Cancer Definition (PDF)

Document Sample
Cancer Definition (PDF) Powered By Docstoc
This page intentionally left blank

Raymond W. Ruddon, M.D., Ph.D.
University of Michigan Medical School
Ann Arbor, Michigan

Oxford University Press, Inc., publishes works that further
Oxford University’s objective of excellence
in research, scholarship, and education.

Oxford New York
Auckland Cape Town Dar es Salaam Hong Kong Karachi
Kuala Lumpur Madrid Melbourne Mexico City Nairobi
New Delhi Shanghai Taipei Toronto
With offices in
Argentina Austria Brazil Chile Czech Republic France Greece
Guatemala Hungary Italy Japan Poland Portugal Singapore
South Korea Switzerland Thailand Turkey Ukraine Vietnam

Copyright # 2007 by Oxford University Press, Inc.

Published by Oxford University Press, Inc.
198 Madison Avenue, New York, New York 10016
Oxford is a registered trademark of Oxford University Press
All rights reserved. No part of this publication may be reproduced,
stored in a retrieval system, or transmitted, in any form or by any means,
electronic, mechanical, photocopying, recording, or otherwise,
without the prior permission of Oxford University Press.

Library of Congress Cataloging-in-Publication Data
Ruddon, Raymond W., 1936–
Cancer biology / Raymond W. Ruddon.— 4th ed.
    p. ; cm.
Includes bibliographical references and index.
ISBN-13: 978-0-19-517543-1 (cloth)
ISBN-13: 978-0-19-517544-8 (pbk.)
1. Cancer. 2. Carcinogenesis. 3. Molecular biology. I. Title.—
[DNLM: 1. Cell Transformation, Neoplastic. 2. Neoplasms
Etiology. QZ202R914C 2007]
RC261.R85 2007
616.99'4071—dc22          2006010326

Printed in the United States of America
on acid-free paper
                  I dedicate this book to my spouse, Lynne Ruddon,
      who has been my best friend and the love of my life for over 45 years.
               Her continual and unflagging patience and support
have made possible whatever success I have experienced in my professional career.
This page intentionally left blank

There have been a significant number of ad-             other multifaceted chronic diseases, an interac-
vances in the field of cancer research since the        tion is required among researchers in many fields,
first edition of Cancer Biology, which was pub-         including molecular biologists, chemists, compu-
lished in 1981. These include advances in defin-        tational scientists, biomedical engineers, epide-
ing the genetic and phenotypic changes in cancer       miologists, and health services researchers, as well
cells, the genetic susceptibility to cancer, mole-     as dedicated physicians, nurses, and other health
cular imaging to detect smaller and smaller tu-        care professionals.
mors, the regulation of gene expression, and the          I would like to thank the many investigators
‘‘-omics’’ techniques of genomics, proteomics, and     who have allowed me to use data from their own
metabolomics, among others. Yet, the goals of the      research to illustrate key points in the text. I would
fourth edition of Cancer Biology remain the same       also like to thank the numerous colleagues who
as those of the earlier editions, namely to provide    have read the earlier editions and used them in
a historical perspective on key developments in        their teaching. Their comments have been help-
cancer research as well as the key advances of sci-    ful in revising the text. I am especially gratified
entific knowledge that will lead to a greatly in-       by the feedback from some individuals who have
creased ability to prevent, diagnose, and treat        said that Cancer Biology was their first exposure
cancer.Unfortunately,manyaspectsoftheexciting          to the field of cancer research and that reading it
breakthroughs in our knowledge of basic cancer         inspired them to seek a career in the field.
biology have yet to be translated into standard           I want to thank Denise Gonzalez for pre-
care for patients. This will require an expanded       paration of some of the early chapters of the
ability of basic scientists and clinical researchers   book. I am greatly indebted to Paulette Thomas
to learn to speak each other’s language and to         for her diligent and patient work on the pre-
collaborate on bringing basic research findings to      paration of the illustrations and on other technical
the bedside. A goal for this book, which may seem      components of the book. I am especially indebted
overly ambitious if not a bit pompous, is to pro-      to Kathy Christopher for her careful preparation
vide part of the lingua franca for these groups        and preliminary editing of the text. Without her,
of experimentalists to better communicate. Now         the book could not have been completed. I also
more than ever it has become clear that to             want to thank the editors and production staff
achieve real breakthroughs in improving much           at Oxford University Press who made the book
needed diagnosis and treatment of cancer and           happen.
This page intentionally left blank

1. CHARACTERISTICS OF                               Interaction of Chemical Carcinogens
   HUMAN CANCER                              3      with Oncogenes and Tumor
                                                    Suppressor Genes                         27
  What Everyone Wants to Know
  about Cancer                               3      Carcinogen-Induced Epigenetic
                                                    Changes                                  27
    Patients                                 3
                                                    Tumor Initiation, Promotion,
    Physicians and Health Care
                                                    and Progression                          27
    Professionals                            3
                                                      Mechanisms of tumor initiation         31
    Cancer Researchers                       3
                                                      Endogenous carcinogenesis              33
  What is Cancer?                            4
                                                      Mechanisms of tumor promotion and
    Definition of Cancer                      4        progression                            34
    Description of Cancer                    4          Central dogma of
  What Significant Events Have                           tumor progression                    35
  Happened in Cancer Research in                        Mechanisms of tumor-promoting
  the Last 20 Years?                          5         agents                               36
  Basic Facts about Cancer                    7
                                                    Experimental Models for the Study
  Hallmarks of Malignant Diseases             9     of Carcinogenesis                        38
  Classification of Human Cancers             12     Validity of Tests for Carcinogenicity    40
  Macroscopic and Microscopic                     Irradiation Carcinogenesis                 43
  Features of Neoplasms                      13     Ionizing Radiation                       44
  Grade and Stage of Neoplasms               14     Ultraviolet Radiation                    45
    Histologic Grade of Malignancy           14   Oxygen Free Radicals, Aging,
    Tumor Staging                            14   and Cancer                                 45
                                                  Genetic Susceptibility and Cancer          47
2. CAUSES OF CANCER                          17   Multiple Mutations in Cancer               47
  The Theory of ‘‘Hits’’                     17   DNA Repair Mechanisms                      48
  Chemical Carcinogenesis                    19   Viral Carcinogenesis                       51
    Historical Perspectives                  19     Historical Perspectives                  51
    Metabolic Activation of Chemical                Role of Viruses in the Causation
    Carcinogens                              21     of Human Cancer                          53
      Donors of simple alkyl groups          21       Association of Epstein-Barr virus
      Cytochrome P-450–mediated activation   21       and human cancers                      54
      2-acetylaminofluorene                   22       Hepatitis virus and hepatocellular
      Other aromatic amines                  23       carcinoma                              54
      Polycyclic aromatic hydrocarbons       24       Papillomaviruses and cervical cancer   55
    DNA Adduct Formation                     26       HTLV-1 and adult T-cell leukemia       55
x                                                                                              CONTENTS

3. THE EPIDEMIOLOGY                                     Cell Phones                                    105
   OF HUMAN CANCER                            62        Electromagnetic Fields                         105
    Trends in Cancer Incidence                          Alcohol                                        106
    and Mortality                             62        Organochlorine Compounds,
      U.S. Data                               62        Polycyclic Aromatic Hydrocarbons,
      Cancer Is a Global Problem              64        and Breast Cancer                              106
    Data for Some Prevalent Human                       Antiperspirants                                107
    Cancers                                   65        Water Chlorination                             107
      Lung Cancer                             65        Abortion or Miscarriage
      Breast Cancer                           67        and Breast Cancer                              108
      Colorectal Cancer                       69        Asbestos                                       108
      Liver Cancer                            70        Saccharin                                      108
      Pancreatic Cancer                       70        Acrylamide in Foods                            109
      Cancers of the Female Reproductive                Alar                                           109
      Tract                                   70        SV40 Virus in Early Polio Vaccines             110
        Cervical cancer                       70
        Ovarian cancer                        71
        Endometrial cancer                    71    4. THE BIOCHEMISTRY
      Prostate Cancer                         71       AND CELL BIOLOGY
      Urinary Bladder Cancer                  72       OF CANCER                                       117
      Lymphoma                                73      Historical Perspectives                          117
      Leukemia                                75      Growth Characteristics
      Skin Cancer                             75      of Malignant Cells                               120
      Cancers of the Central Nervous                    Phenotypic Alterations in Cancer Cells         120
      System                                  77        Immortality of Transformed Cells
    Role of Various Factors in the                      in Culture                                     121
    Development of Cancers                    78        Decreased Requirement for Growth
      Cigarette Smoking                       80        Factors                                        122
      Alcohol                                 83        Loss of Anchorage Dependence                   122
      Diet                                    83        Loss of Cell Cycle Control and
      Sexual Development, Reproductive                  Resistance to Apoptosis                        122
      Patterns, and Sexual Behavior           85        Changes in Cell Membrane Structure
      Industrial Chemicals and Occupational             and Function                                   123
      Cancers                                  85         Alterations in cell surface glycolipids,
                                                          glycoproteins, proteoglycans,
      Herbicides                               86
                                                          and mucins                                   123
      Air and Water Pollutants                 87
                                                          Role of glycosyl transferases and
      Radiation                                89         oligosaccharide processing enzymes           124
        Ultraviolet                            89         Mucins                                       125
        Ionizing radiation                     90         Proteoglycans                                125
        Radon                                  91       Modification of Extracellular
      Drugs                                    92       Matrix Components                              126
      Hormones                                 93       Cell–Extracellular Matrix and
      Infection                                94       Cell–Cell Adhesion                             126
    Aging and Cancer                           94     Cell Proliferation versus Differentiation        128
    Genetic Factors in Cancer                  96       Mechanisms of Cellular Differentiation         129
      Inherited Cancers                        97         Slime molds                                  131
      Gene–Environment Interactions            98         Yeast                                        134
    Avoidability of Cancer                     99         Sea urchin                                   134
      Risk Assessment                         100         Drosophilia melanogaster                     136
    The Great Cancer Myths                    102         Mouse                                        136
      Passive Smoking                         103         Pathways: getting to know all the players    136
      Radon in the Home                       104         Stimulation of cancer cell differentiation   139
CONTENTS                                                                                                   xi

     Stem Cells                                   139         Tumor necrosis factor receptor signaling    205
   Cell Cycle Regulation                          143         Tumor growth factor-b signal transduction   205
     Historical Perspectives                      143         Heat shock protein-mediated events          206
     The Molecular Players                        146     Angiogenesis                                    207
       Cyclin-dependent protein kinases           146       Vascular Endothelial Growth Factor            210
       CDK inhibitors                             146       Platelet-Derived Growth Factor                211
       Cyclins                                    147       Angiopoietins                                 211
       Cell cycle checkpoints                     148       Ephrins                                       212
       Cell cycle regulatory factors as targets             Angiogenesis Inhibitors                       212
       for anticancer agents                      150         Inhibitors of proangiogenic factors         212
   Apoptosis                                      151         Metalloproteinases                          213
     Historical Perspectives                      152         Integrins                                   213
     Biochemical Mechanisms of Apoptosis          153         Endogenous inhibitors                       213
       Caspases                                   154         HIF-1a                                      213
       Bcl-2 family                               156         Miscellaneous anti-angiogenic agents        214
       Role of mitochondria in apoptosis          156         Clinical data                               214
     Anoikis                                      157       Lymphangiogenesis                             215
     Resistance to Apoptosis in Cancer                      Tumor Dormancy                                215
     and Potential Targets for Therapy            157     Biology of Tumor Metastasis                     216
   Growth Factors                                 158       The ‘‘Classic’’ Theory of
     Historical Perspectives                      158       Tumor Metastasis                              216
     Insulin                                      161       Alternate Theory of Tumor Metastasis          219
     Insulin-Like Growth Factors                  161       Invasion and Metastasis: The Hallmarks
     Nerve Growth Factor                          164       of Malignant Neoplasia                        219
     Epidermal Growth Factor                      165       Metastasis Is at Least Partly
     Fibroblast Growth Factor                     171       a Selective Process                           223
     Platelet-Derived Growth Factor               173       Biochemical Characteristics
     Transforming Growth Factors                  176       of Metastatic Tumor Cells                     225
       TGF-a                                      177         Relationship of cancer metastasis
                                                              to normal tissue invasion events            225
       TGF-b                                      178
                                                              Role of lytic enzymes in the
     Hematopoietic Growth Factors                 181         metastasis cascade                          226
     Hepatocyte Growth Factor and                             Role of plasma membrane components
     Scatter Factor                               185         in metastasis                               229
     Miscellaneous Growth Factors                 186         Role of extracellular matrix components
   Signal Transduction Mechanisms                 186         and the basement membranes
     Some Key Signal Transduction                             in tumor metastasis                         230
     Concepts                                     191         Tissue adhesion properties
       Transcriptional regulation by                          of metastatic cells                         232
       signal transduction                        191         Ability of metastatic tumor cells
       Protein–protein interaction domains        191         to escape the host’s immune response        234
       Spatial and temporal regulation            192         Chemotactic factors in cancer
       Signaling networks and crosstalk           193         cell migration                              234
     Overview of Some Signal Transduction                     Role of oncogenes in tumor metastasis       235
     Pathways Important in Cancer                 194       Identification of the ‘‘Metastatic Genes’’
       G protein-linked receptors                 194       and ‘‘Metastasis Suppressor Genes’’           236
       The phosphoinositide 3-kinase pathway      198
       mTOR                                       198   5. MOLECULAR GENETICS
       Tyrosine kinase pathways                   200      OF CANCER                                      257
       Protein phosphatases                       200     Chromatin Structure and Function                258
       JAK-STAT pathway                           201       Components of Chromatin                       258
       Estrogen receptor pathway                  202       Chemical Modifications of
       Hypoxia-inducible factor                   204       Chromatin-Associated Proteins                 259
xii                                                                                           CONTENTS

        Packaging of Chromatin                    262     Aneuploidy                                314
        Structure and Function of Interphase              Disomy                                    316
        Chromosomes                               264     Trinucleotide Expansion                   316
        Nuclear Organization                      266     Microsatellite Instability                317
        Nuclease Sensitivity                      267     Mismatch DNA Repair Defects               317
        Transcriptional Activation and                    Gene Derepression in Cancer Cells         318
        the Cancer Connection                     268       Ectopic hormone production
        Control of Gene Expression during                   by human cancers                        318
        Embryonic Stem Cell Differentiation       269       Possible mechanisms of ectopic
      Split Genes and RNA Processing              270       protein production                      319
      Genetic Recombination                       273     Chromosomal Abnormalities
      Gene Amplification                           277     in Leukemic Patients Exposed
      Cis-Acting Regulatory Elements:                     to Genotoxic Agents                       320
      Promoters and Enchancers                    279     Cancer Genetic Changes Summed Up          321
      Transcription Factors                       282   Oncogenes                                   321
        Structural Motifs of Regulatory                   Historical Perspectives                   321
        DNA-Binding Proteins                      282       The provirus, protovirus,
        Repressors                                284       and oncogene hypothesis                 321
        General (Basal) Transcription Factors     285       The src gene                            323
        Promoter- and Enhancer-Specific                    Oncogene Families                         324
        Transcription Factors                     287     Cell Transforming Ability of onc Genes    326
          AP-1/Fos/Jun                            287     Functional Classes of Oncogenes           328
          ATF/CREB                                287     Characteristics of Individual Oncogenes   330
          SP1                                     290       ras                                     330
          Oct-3                                   290       myc                                     333
          The superfamily of hormone receptors    290       src                                     335
          YY1                                     291       jun and fos                             338
          LEF-1                                   291       ets                                     338
          E2F                                     291       bcr-abl                                 340
          Tissue specific transcription factors    291       myb                                     341
            MyoD                                  292       bcl-2                                   341
            Liver specific transcription factors   293       NF-kB/rel                               342
            Pit-1                                 293       erbA                                    342
            E2A                                   293       sis                                     343
            NF-kB                                 293       erbB                                    344
            POU-domain binding proteins           294       erbB-2 (Her-2/neu)                      344
            Ets1 and Ets2                         294       Other growth factor or growth factor
            Homeobox proteins                     294       receptor oncogenes                      345
      DNA Methylation                             297         fms                                   345
        DNA Methyltransferases                    298         kit                                   345
        Methyl DNA Binding Proteins               299         trk                                   346
        DNA Methylation and Cancer                300         met                                   346
      Genomic Imprinting                          302         Pokemon                               346
      Loss of Heterozygosity                      304   Cellular onc Gene Expression during
      Telomeres and Telomerase                    304   Normal Embryonic Development                346
      Post-transcriptional Regulation             305   DNA Tumor Viruses                           347
      Molecular Genetic Alterations                       SV40 and Polyoma                          347
      in Cancer Cells                             307     Papilloma Viruses E6 and E7               349
        Translocations and Inversions             308     Adenoviruses E1A and E1B                  350
        Chromosomal Deletions                     312     Hepatitis B Virus                         351
        Gene Amplification                         314     Herpes Viruses                            351
        Point Mutations                           314   Tumor Suppressor Genes                      352
CONTENTS                                                                                               xiii

     Historical Perspectives                           352       T Lymphocytes and T Cell Activation   406
     Properties of Individual Tumor                              The Immunological Synapse             408
     Suppressor Genes                                  354       B Lymphocytes and B Cell Activation   409
       rb                                              354       Natural Killer Cells                  410
            Characterization of the rb protein         354       Cell-Mediated Cytotoxicity            411
            Interactions of Rb proteins                355       Danger Theory                         412
            Role of rb in reversing the                        Role of Gene Rearrangement
            malignant phenotype                        356     in the Tumor Response                   413
            Requirement of a functional rb-1                   Heat Shock Proteins as Regulators
            gene in development                        356     of the Immune Response                  414
            Cell cycle regulation by Rb                356     Inflammation and Cancer                  414
            Interactions of Rb protein with                    Immunotherapy                           415
            transcription factors and DNA
                                                                 Rationale for Immunotherapy           415
            regulatory elements                        357
                                                                 Identification and Characterization
       p53                                             357       of Tumor-Derived Antigenic Peptides   417
            Characteristics of p53 and its mutations   357       Cytokines                             417
            Mutagenesis of p53                         359         Interferons                         418
            Ability of p53 to reverse cellular                     Interleukins                        420
            transformation and tumorigenesis           359
                                                                   Tumor necrosis factor               421
            Role of p53 in cell cycle progression
            and in inducing apoptosis                  360       Adoptive Immunotherapy                422
            Mechanism of p53’s actions                 360       Vaccines                              424
       Wilms’ tumor suppressor gene wt-1               362       Monoclonal Antibodies                 424
       Adenomatous polyposis coli (apc) gene           364     How Tumor Cells Avoid the
                                                               Immune Response                         424
       Deleted in colorectal cancer (dcc) gene         364
       Hereditary nonpolyposis colorectal
       cancer (hnpcc) gene                             364   7. CANCER DIAGNOSIS                       429
       Neurofibromatosis genes nf-1 and nf-2            365     Medical and Scientific Drivers for
       Von Hippel-Lindau syndrome and                          Expanded Cancer Diagnostic
       renal cell carcinoma gene                       365     Techniques                              429
       BRCA1 and BRCA2                                 366     Categories of Tumor Markers             433
     Identification of Tumor Suppressor                           Nucleic Acid-Based Markers            433
     Genes                                             366         Cancer-associated mutations         434
   Mechanisms of Gene Silencing                        367         Loss of heterozygosity and
     Antisense                                         367         microsatellite instability          434
     Ribozymes                                         368         DNA methylation patterns            435
     DNAzymes                                          370         Mitochondrial DNA mutations         435
     RNAi                                              370         Viral DNA                           435
       Transitive RNAi                                 372     Gene Expression Microarrays             436
       Micro-RNA                                       373       Laser-Capture Microdissection         437
       Small temporal RNA                              374       Comparative Genome Hybridization      437
       Short hairpin RNA                               374       Tissue Arrays                         439
   Gene Therapy                                        374       Gene Expression Microarrays
     Gene Therapy for Cancer                           375       in Individual Cancer Types            439
   Personalized Medicine and                                       Lymphoma                            439
   Systems Biology                                     376         Leukemia                            440
                                                                   Breast cancer                       440
6. TUMOR IMMUNOLOGY                                    400         Ovarian cancer                      442
   Historical Perspectives                             400         Prostate cancer                     442
   Mechanisms of the Immune Response                               Colorectal cancer                   443
   to Cancer                                           404         Lung cancer                         444
     Antigen Presenting Cells                          404         Renal cancer                        444
     How Antigens Are Processed                        406         Hepatic cancer                      445
xiv                                                                                             CONTENTS

          Other cancers and cancer-related                 Platelets                                 478
          phenotypes                           445         Thrombosis                                478
      Proteomics                               446       Fever and Infection                         479
        Proteomics Methods                     447       Hormonal Effects                            481
          Two-dimensional electrophoresis      447       Hypercalcemia                               481
          Isotope-coded affinity tags (ICAT)    447       Neurologic Effects                          482
          Mass spectrometry-based proteomics   447       Dermatologic Effects                        483
          Protein chips                        449       Fatigue                                     483
          Surface-enhanced laser desorption/           Sequelae of Cancer Treatment                  484
          ionization (SELDI)                   449
          Yeast two-hybrid system              450
          Phage display                        450   9. CANCER PREVENTION                            487
          Organelle proteomics                 451     Molecular Mechanisms of Aging
                                                       and Its Prevention                            487
          Plasma proteome                      451
                                                         Somatic Mutation                            487
          Tissue proteomics: imaging
          mass spectrometry                    451       Telomere Loss                               487
          Pattern recognition                  452       Mitochondrial Damage                        488
          The unfolded protein response        452       Formation of Oxygen-Free Radicals           488
        Proteomics in Cancer Diagnosis         453       Cell Senescence                             488
          Lung cancer                          454       DNA Repair and Genome Stability             488
          Ovarian cancer                       454       Caloric Restriction                         490
          Breast cancer                        454     Diet and Cancer Prevention                    491
          Prostate cancer                      454     Chemoprevention                               493
          Pancreatic cancer                    455       Molecular Targets for
                                                         Chemoprevention                             494
      Circulating Epithelial Cells             455
                                                         Antimutagens and Carcinogen-
      Circulating Endothelial Cells and                  Blocking Agents                             494
      Endothelial Progenitor Cells             456
                                                           Isothiocyanates                           494
      Molecular Imaging                        458
                                                           Oltipraz                                  495
        Protein–Protein Interactions           459
                                                           Other organosulfur compounds              495
        Protein Degradation                    459
                                                           Ellagic acid                              496
        Imaging Gene Expression In Vivo        459
                                                           Dehydroepiandrosterone (DHEA)             496
          Bioluminescent detection             460
                                                         Antiproliferative Agents                    496
        Magnetic Resonance
        Spectroscopy                           461         Retinoids and b-carotene                  496
          Ultrasound Imaging                   461         Hormonal chemoprevention                  498
      Nanotechnology                           461           Oral contraceptives                     498
                                                             Gonadotropin-releasing hormone
        Gray Goo                               464
                                                             analogs (GNRHAs)                        498
      Pharmacogenomics and
      Pharmacogenetics                         464           Hormone replacement therapy             498
                                                             Tamoxifen, Raloxifene, and
        Importance of Pharmacogenomics
        in Cancer                              465           aromatase inhibitors                    499
      Haplotype Mapping                        466           Antiandrogens                           499
                                                           Anti-inflammatory agents                   499
                                                           Cyclooxygenase-2 inhibitors               500
                                                           Ornithine decarboxylase inhibitors        500
   AND ITS TREATMENT                           472
                                                         Antioxidants                                500
      Patient–Tumor Interactions               472
                                                         Protease Inhibitors                         501
        Pain                                   472
                                                         Histone Deacetylase Inhibitors              501
        Nutritional Effects                    474
                                                         Statins                                     501
        Hematologic Effects                    477
                                                         Multiagent chemoprevention                  502
          Erythropoiesis                       477
          Leukopoiesis                         478     INDEX                                         507
This page intentionally left blank

Characteristics of Human Cancer

                                                          Physicians and Health Care
                                                          The members of the health care team who take
                                                          care of cancer patients have a different set of
During my career as a cancer scientist, I have            questions. These may include the following: What
frequently received calls from individuals who            are the most appropriate diagnostic tests with low
recently heard a physician tell them the ominous          false negatives and false positives? What are the
words ‘‘You have cancer,’’ or from people who             differential diagnoses that need to be ruled out?
have heard that statement about a family mem-             And once the diagnosis is made, what is the stage
ber or close friend. The first question usually is         and histological grade? Is the disease local, re-
‘‘What can you tell me about this kind of can-            gional, or metastatic? What is the likely prognosis
cer?’’ They may have already visited several              and the best therapeutic approach? How often is
Internet sites and have some information, not al-         follow-up of the patient required and for how
ways accurate or scientifically based. If the pa-          long? If the disease progresses, how may the
tient is a child and the inquiry comes from               treatment approaches change? Some of the data
parents, they frequently have a great feeling of          that relate to answering these questions will also
guilt and want to know what they did wrong, or            be discussed in the book.
they may lash out at some perceived environ-
mental agent that they think is the cause, such as
                                                          Cancer Researchers
water pollutants or electromagnetic fields from
high-power lines in their neighborhood. Indi-             Basic scientists and clinicians working in the field
viduals or their family members then want to              of cancer research, by contrast, have yet another
know what caused the cancer, what the meaning             set of fundamental questions: What are the basic
of the test results is, what the treatment options        mechanisms of malignant transformation of cells?
are, and, if the tumor has spread, if there are any       What causes of cancer can be identified? Know-
preventive measures that can be taken to stop             ing that, what preventive measures can be taken?
further spread of the cancer. If cancer is in the         Are there genetic profiles, hereditary or induced
family, they may ask what their chances are of            by spontaneous mutations, that correlate with
getting cancer. These are questions that are al-          susceptibility or progression of cancer? Can the
ways difficult to answer. One of the goals of this         gene expression patterns of cancer cells be used
book is to try to provide the scientific basis for         to identify targets for cancer diagnosis or ther-
approaching these questions.                              apy? What proof-of-principle studies are needed

4                                                                                      CANCER BIOLOGY

to verify these targets? What type of clinical trials   in the expression of multiple genes, leading to
is needed to determine the toxicity and efficacy of      dysregulation of the normal cellular program for
a new therapeutic modality? These questions will        cell division and cell differentiation. This results
also be addressed.                                      in an imbalance of cell replication and cell death
                                                        that favors growth of a tumor cell population.
                                                        The characteristics that delineate a malignant
WHAT IS CANCER?                                         cancer from a benign tumor are the abilities to
                                                        invade locally, to spread to regional lymph nodes,
A few years ago I was at a small meeting with a         and to metastasize to distant organs in the body.
group of distinguished cancer biologists and cli-       Clinically, cancer appears to be many different
nicians. It was an interesting meeting because          diseases with different phenotypic characteris-
there were also distinguished scientists from           tics. As a cancerous growth progresses, genetic
other fields. The idea of the meeting was to             drift in the cell population produces cell het-
stimulate cross-fertilization of ideas from dif-        erogeneity in such characteristics as cell anti-
ferent scientific disciplines, with the hope that        genicity, invasiveness, metastatic potential, rate
new paradigms for approaching the causes of             of cell proliferation, differentiation state, and
cancer and its course would be conceived.               response to chemotherapeutic agents. At the
   One of the first questions that one of the non-       molecular level, all cancers have several things
cancer researchers asked was, what is the defi-          in common, which suggests that the ultimate
nition of cancer? It was somewhat startling to          biochemical lesions leading to malignant trans-
hear the vigorous discussion and even squabbling        formation and progression can be produced by a
among the distinguished cancer scientists in their      common but not identical pattern of alterations
attempt to define cancer. Although most could            of gene readout. In general, malignant cancers
agree on a few key characteristics, everyone had        cause significant morbidity and will be lethal to
their own caveats or additional variations to add.      the host if not treated. Exceptions to this appear
So, like all good academic groups, they appointed       to be latent, indolent cancers that may remain
a committee to come up with a consensus defi-            clinically undetectable (or in situ), allowing the
nition. As the most gullible person there, I agreed     host to have a standard life expectancy.
to chair the committee. After many phone calls             Some points in the description may not seem
and E-mails going back and forth, we came up            intuitively obvious. For example, cancer doesn’t
with the definition and more detailed description        just occur in humans, or just mammals for that
below. I should note that the definition is the sort     matter. Cancer (or at least tumorous growths—
of thing that would appear in a dictionary and the      these may or may not have been observed to
description contains some of the points and ca-         metastasize) has been observed in phyla as old as
veats thought crucial for taking into account the       Cnidaria, which appeared almost 600 million
characteristics of this multifaceted disease.           years before the present, and in other ancient
                                                        phylasuch asEchinodermata (> 500 million years
                                                        old), Cephalopoda (500 million years old), Am-
Definition of Cancer
                                                        phibia (300 million years old), and Aves (150
Cancer is an abnormal growth of cells caused by         million years old). Curiously, cancer has never
multiple changes in gene expression leading to          been seen (or at least reported) in a number of
dysregulated balance of cell proliferation and          phyla such as Nematoda, Tradigrada, and Roti-
cell death and ultimately evolving into a popu-         fera. It is intriguing to consider that these or-
lation of cells that can invade tissues and me-         ganisms may have some protective mechanisms
tastasize to distant sites, causing significant          that prevent them from getting tumors. If so, it
morbidity and, if untreated, death of the host.         would be important to find out what these
                                                        mechanisms are.
                                                           One thing is clear, though, which is that
Description of Cancer
                                                        cancer is a disease of multicellular organisms.
Cancer is a group of diseases of higher multicel-       This trait implies that there is something in-
lular organisms. It is characterized by alterations     herent in the ability of cells to proliferate in
CHARACTERISTICS OF HUMAN CANCER                                                                          5

clumps or to differentiate into different cell        to becoming cancerous, at least two genetic hits
types and move around in the body to sites of         are required. One may be inherited and another
organogenesis that is key to the process of tu-       accrued after birth or both may be accrued after
morigenesis. Problems occur when these pro-           birth (so-called somatic, or spontaneous, hits).
cesses become dysregulated.                           The kinds of genes involved are oncogenes,
   One might also argue that evolution itself has     which when activated lead to dysregulated cell
played some tricks on us because some of the          proliferation, and tumor suppressor genes, which
properties selected for may themselves be pro-        become inactivated or deleted, producing a loss
cesses that cancer cells use to become invasive       of the cell’s checks and balances controlling cell
and metastatic. Or to phrase it differently: Is       proliferation and differentiation.
cancer an inevitable result of a complex evolu-          The single most common, if not universal,
tionary process that has advantages and disad-        trait that occurs in all cancers is genetic drift. or
vantages? Some of these processes might be the        the ability of cells to lose the stringent require-
following:                                            ment for precise DNA replication and to acquire
                                                      the ability to undergo sequential progressive
   1. The mechanism of cell invasiveness that
                                                      changes in their genome, through mutations,
      allows the implantation of the early em-
                                                      gene rearrangement, or gene deletion. This
      bryo into the uterine wall and the devel-
                                                      has sometimes been called the acquisition of a
      opment of a placenta.
                                                      ‘‘mutator phenotype.’’
   2. Cell motility that allows neural cells, for
      example, to migrate from the original neu-
      ral crest to form the nervous system.
   3. The development of a large, complex ge-         WHAT SIGNIFICANT EVENTS HAVE
      nome of up to 40,000 genes that must be         HAPPENED IN CANCER RESEARCH
      replicated perfectly every time a cell di-      IN THE LAST 25 YEARS?
   4. The large number of cells in a human or         As I was beginning to gather my thoughts for the
      higher mammal that must replicate and           fourth edition of Cancer Biology, one of my
      differentiate nearly perfectly every time       colleagues mentioned that he thought it would
      (some can be destroyed if they become           be of interest to describe the significant things
      abnormal).                                      that have happened in cancer biology in the
   5. The long life span of humans and higher         25 years since the first edition was published
      mammals, increasing the chance for a            (1981). Many things have happened since then,
      genetic ‘‘hit’’ to occur and lead a cell down   of course, and everyone has their favorite list.
      a malignant path.                               But looking back at the table of contents for the
                                                      first edition and at the outline for this edition,
  As we shall see in later chapters of this book,
                                                      several things struck me, as listed below.
cancer cells take advantage of a number of these
events and processes.                                     1. Cancer susceptibility genes. In 1981 we
  Other questions that arose at the gathering                knew that familial clustering of some can-
above from scientists not in the field of cancer              cers occurred, for example, with colon can-
were the following:                                          cer, but the genes involved in this hadn’t
                                                             been determined. The APC, BRCA-1,
   1. Is there a single trait or traits that all
                                                             BRCA-2, and p53 inherited mutations, for
      cancer cells have?
                                                             example, were not known at that time. Re-
   2. How many genetic ‘‘hits’’ does it take to
                                                             search in this area has identified a number
      make a cancer cell?
                                                             of genes involved in cancer susceptibility,
   3. What kinds of genes are involved in these
                                                             and with modern cloning techniques, more
                                                             are identified every few months.
  These questions are all dealt with in later             2. The techniques of modern molecular
chapters. Suffice it to say here that for a cell to           biology were in their infancy at that time.
become cancerous or at least take the first steps             Polymerase chain reaction (PCR), DNA
6                                                                                    CANCER BIOLOGY

         microarrays, protein chips, and bioin-               of the immune response and the ability
         formatics were not terms in anybody’s                to manipulate it with cytokines, activated
         dictionary.                                          dendritic cells, and vaccines. Such ma-
    3.   Genes involved in cancer initiation and              nipulation was not in the treatment ar-
         promotion were very poorly defined. Al-               mamentarium.
         though we knew that chemicals and irra-         8.   The first treatment of a patient with
         diation could damage DNA and initiate                gene therapy occurred in 1990. Several
         cancer in animals and humans, the spe-               gene therapy clinical trials for cancer are
         cific genes altered were almost completely            under way and some gene therapy modal-
         unknown. We now know a lot about the                 ities will likely be approved in the next
         genes involved at various stages of a num-           few years.
         ber of cancers. For example, the work of        9.   The viral etiology of cancer was still be-
         Bert Vogelstein and colleagues has de-               ing widely debated in 1981. The involve-
         fined a pathway sometimes called the                  ment of Epstein-Barr virus in Burkitt’s
         ‘‘Vogelgram’’ for the progression of colon           lymphoma and of hepatitis B virus in
         cancer (see Chapter 5). We knew that                 liver cancer was becoming accepted, but
         DNA repair was important and that herita-            the role of viruses in these diseases and
         ble conditions of defective DNA repair               in cervical cancer, Kaposis’ sarcoma, and
         (e.g., xeroderma pigmentosum) could lead             in certain T-cell lymphomas became
         to cancer, but the ideas about the mech-             clearer much later.
         anisms of DNA repair were primitive.           10.   Although some growth factors that affect
    4.   The identification of oncogenes didn’t                cancer cell replication, such as IGF-1
         really start until the early 1980s. The src          and IGF-2, FGF, NGF, PDGF, and
         gene was identified in 1976 by Stehelin               EGF, were known in 1981, knowledge
         et al., and erb, myc, and myb oncogenes              about their receptors and signal trans-
         were identified in the late 1970s, but this           duction mechanisms was primitive in-
         was about the limit of our knowledge                 deed. Tumor growth factor a was known
         (see Chapter 5).                                     as sarcoma growth factor (SGF), and the
    5.   The term tumor suppressor gene wasn’t                existence of its partner, TGF-b, was only
         even coined until the early 1980s, al-               implied from what was thought to be
         though their existence had been implied              a contaminating HPLC peak from the
         from the cell fusion experiments of                  purification procedure. The explosion
         Henry Harris, (Chapter 5) who showed                 of knowledge about signal transduction
         that if a normal cell was fused with a               mechanisms and how these pathways in-
         malignant cell, the phenotype was usu-               teract has been a tremendous boon to our
         ally nonmalignant. The RB gene was the               understanding of how cells respond to
         first one cloned, in 1983 by Cavenee et al.           signals in their environment and commu-
         (Chapter 5) p53 was originally thought               nicate with each other.
         of as an oncogene. It wasn’t realized until    11.   Knowledge about the regulation of gene
         1989 that wild-type p53 could actually               expression has greatly increased in the
         suppress malignant transformation. A                 past 25 years, on the basis of our current
         number of tumor suppressor genes have,               information on the packaging of chro-
         of course, been identified since then.                matin, transcription factors, coinducers
    6.   Starting in the 1970s, cell cycle check-             and corepressors, and inhibitory RNA
         points were identified in yeast by Lee                (siRNA).
         Hartwell and colleagues, but the identi-       12.   While not topics discussed in detail in
         fication of human homologs of these genes             the earlier editions of Cancer Biology, ad-
         didn’t occur until the late 1980s (see Chap-         vances in diagnostic imaging such as mag-
         ter 4).                                              netic resonance imaging (MRI), computed
    7.   Tumor immunology was still poorly un-                tomography (CT), and positron emission
         derstood in 1981—both the mechanism                  tomography (PET) have significantly im-
CHARACTERISTICS OF HUMAN CANCER                                                                          7

       proved cancer diagnosis. Improved radia-       with the others. The highest mortality rates are
       tion therapy, combined modality therapy,       seen with lung, colorectal, breast, and prostate
       bone marrow transplant, and supportive         cancers (Fig. 1–1). Over 570,000 people die
       care have also improved significantly.          each year in the United States from these and
                                                      other cancers. More people die of cancer in 1
                                                      year in the United States than the number of
BASIC FACTS ABOUT CANCER                              people killed in all the wars in which the United
                                                      States was involved in the twentieth century
Cancer is a complex family of diseases, and car-      (Fig. 1–2).
cinogenesis, the events that turn a normal cell in       In many cases the causes of cancer aren’t
the body into a cancer cell, is a complex multi-      clearly defined, but both external (e.g., environ-
step process. From a clinical point of view, can-     mental chemicals and radiation) and internal
cer is a large group of diseases, perhaps up to a     (e.g., immune system defects, genetic predispo-
hundred or more, that vary in their age of onset,     sition) factors play a role (see Chapter 2). Clearly,
rate of growth, state of cellular differentiation,    cigarette smoking is a major causative factor.
diagnostic detectability, invasiveness, metastatic    These causal factors may act together to initiate
potential, response to treatment, and prognosis.      (the initial genetic insult) and promote (stimu-
From a molecular and cell biological point of         lation of growth of initiated cells) carcinogene-
view, however, cancer may be a relatively small       sis. Often 10 to 20 years may pass before an
number of diseases caused by similar molecular        initiated neoplastic cell grows into a clinically
defects in cell function resulting from common        detectable tumor.
types of alterations to a cell’s genes. Ultimately,      Although cancer can occur at any age, it is
cancer is a disease of abnormal gene expression.      usually considered a disease of aging. The av-
There are a number of mechanisms by which             erage age at the time of diagnosis for cancer of
this altered gene expression occurs. These mech-      all sites is 67 years, and about 76% of all cancers
anisms may occur via a direct insult to DNA,          are diagnosed at age 55 or older. Although can-
such as a gene mutation, translocation, amplifi-       cer is relatively rare in children, it is the second-
cation, deletion, loss of heterozygosity, or via a    leading cause of death in children ages 1–14. In
mechanism resulting from abnormal gene tran-          this age group leukemia is the most common
scription or translation. The overall result is an    cause of death, but other cancers such as osteo-
imbalance of cell replication and cell death in       sarcoma, neuroblastoma, Wilms’ tumor (a kidney
a tumor cell population that leads to an expan-       cancer), and lymphoma also occur.
sion of tumor tissue. In normal tissues, cell pro-       Over eight million Americans alive today have
liferation and cell loss are in a state of equilib-   had some type of cancer. Of these, about half
rium.                                                 are considered cured. It is estimated that about
   Cancer is a leading cause of death in the          one in three people now living will develop some
Western world. In the United States and a num-        type of cancer.
ber of European countries, cancer is the second-         There has been a steady rise in cancer death
leading killer after cardiovascular disease, al-      rates in the United States during the past 75
though in the United States since 1999 cancer         years. However, the major reason why cancer
has surpassed heart disease as the number one         accounts for a higher proportion of deaths now
cause of death in people younger than 85.1 Over       than it did in the past is that today more people
1.3 million new cases of cancer occur in the          live long enough to get cancer, whereas earlier
United States each year, not including basal cell     in the twentieth century more people died of
and squamous cell skin cancers, which add an-         infectious disease and other causes. For exam-
other 1 million cases annually. These skin cancers    ple, in 1900 life expectancy was 46 years for men
are seldom fatal, do not usually metastasize, and     and 48 years for women. By 2000, the expec-
are curable with appropriate treatment, so they       tancy had risen to age 74 for men and age 80 for
are usually considered separately. Melanoma,          women. Thus, even though the overall death
by contrast, is a type of skin cancer that is more    rates due to cancer have almost tripled since
dangerous and can be fatal, so it is considered       1930 for men and gone up over 50% for women,
8                                                                                                CANCER BIOLOGY

    Estimated New Cases*

                                                        Males   Females
                           Prostate    232,090    33%                 Breast                           211,240   32%
                Lung and Bronchus       93,010    13%                 Lung and Bronchus                 79,560   12%
                 Colon and Rectum       71,820    10%                 Colon and Rectum                  73,470   11%
                    Urinary Bladder     47,010     7%                 Uterine Corpus                    40,880    6%
             Melanoma of the Skin       33,580     5%                 Non-Hodgkin Lymphoma              27,320    4%
          Non-Hodgkin Lymphoma          29,070     4%                 Melanoma of the Skin              26,000    4%
           Kidney and Renal Pelvis      22,490     3%                 Ovary                             22,220    3%
                          Leukemia      19,640     3%                 Thyroid                           19,190    3%
           Oral Cavity and Pharynx      19,100     3%                 Urinary Bladder                   16,200    2%
                          Pancreas      16,100     2%                 Pancreas                          16,080    2%
                           All Sites   710,040   100%                 All Sites                        662,870   100%

    Estimated Deaths
                                                        Males   Females
                Lung and Bronchus       90,490   31%                  Lung and Bronchus                 73,020   27%
                           Prostate     30,350   10%                  Breast                            40,410   15%
                Colon and Rectum        28,540   10%                  Colon and Rectum                  25,750   10%
                          Pancreas      15,820    5%                  Ovary                             16,210    6%
                          Leukemia      12,540    4%                  Pancreas                          15,980    6%
                        Esophagus       10,530    4%                  Leukemia                          10,030    4%
    Liver and Intrahepatic Bile Duct    10,330    3%                  Non-Hodgkin Lymphoma               9050     3%
          Non-Hodgkin Lymphoma          10,150    3%                  Uterine Corpus                     7310     3%
                    Urinary Bladder      8970     3%                  Multiple Myeloma                   5640     2%
           Kidney and Renal Pelvis       8020     3%                  Brain and Other Nervous System     5480     2%
                          All Sites    295,280   100%                 All Sites                        275,000   100%

                     Figure 1–1. Ten leading cancer types for estimated new cancer cases and
                     deaths, by sex, United States, 2005. *Excludes basal and squamous cell skin
                     cancers and in situ carcinoma except urinary bladder. Estimates are rounded
                     to the nearest 10. Percentage may not total 100% due to rounding. (From
                     American Cancer Society, Surveillance Research, 2005. CA Cancer J Clin
                     2005; 55:10–30, with permission.)

the age-adjusted cancer death rates in men have                    It is instructive to examine the trends in can-
only increased 54% in men and not at all for                    cer mortality over time to get some clues about
women.2                                                         the causes of cancer. For males, lung cancer
   The major increase has been in deaths due to                 remains the number one cancer killer (Fig. 1–3).
lung cancer. Thus, cigarette smoking is a highly                With a lag of about 20 years, its rise in mortality
suspect culprit in the observed increases. In ad-               parallels the increase in cigarette smoking
dition, pollution, diet, and other lifestyle changes            among men, which has an almost identical curve
may have contributed to this increase in cancer                 starting in the early 1900s. Lung cancer mor-
mortality rates (Chapter 3). The mortality rates                tality rates for men have decreased somewhat
for some cancers has decreased in the past 50                   since 1990, and death rates for colorectal cancer
years (e.g., stomach, uterine cervix); however, the             have dropped slightly in recent years, whereas
mortality rates have been essentially flat for many              prostate cancer mortality has increased some-
of the major cancers such as breast, colon, and                 what. Stomach cancer mortality has dropped
prostate, although 5-year survival rates have im-               significantly since the early 1900s, presumably
proved for these cancers (see Chapter 3).                       because of better methods of food preservation
CHARACTERISTICS OF HUMAN CANCER                                                                                           9


  Death in Thousands




                                         48          54                                                41
                              0.01                                                                                25

                             Gulf    Vietnam      WWI            Korea        WWII      Cancer       AIDS      Murder

                                           Total Battle Deaths                                    Each Year

                             Figure 1–2. Total battle deaths from all wars with U.S. involvement in the
                             twentieth century, compared to number of deaths each year from cancer,
                             AIDS, and murder in the United States. (Personal communication from Don
                             Coffey, Johns Hopkins University, with permission.)

(e.g., better refrigeration, less addition of nitrate                       The good news is that more and more people
and nitrate preservatives). Cancer of the gastro-                        are being cured of their cancers today. In the
esophageal junction, however, has risen signifi-                          1940s, for example, only one in four persons
cantly in recent years, perhaps due to obesity                           diagnosed with cancer lived at least 5 years after
and increased incidence of gastric reflux into the                        treatment; in the 1990s that figure rose to 40%.
esophagus in the U.S. population.                                        When normal life expectancy is factored into
   Somewhat surprising, perhaps, is the fact that                        this calculation, the relative 5-year survival rate
lung cancer has overtaken breast cancer as the                           is about 64% for all cancers taken together.1
number one cancer killer in women (Fig. 1–4).                            Thus, the gain from 1 in 3 to 4 in 10 survivors
This increase occurred in the late 1980s and, as                         means that almost 100,000 people are alive now
was the case for males, parallels the rise in the                        who would have died from their disease in less
percentage of women who smoke. Smoking                                   than 5 years if they had been living in the 1940s.
started to increase dramatically during World                            This progress is due to better diagnostic and
War II. Rosie the Rivetter picked up some bad                            treatment techniques, many of which have come
male habits along with increased access to tra-                          about from our increasing knowledge of the
ditionally male jobs.                                                    biology of the cancer cell.
   Breast cancer mortality rates have remained
stubbornly stable, although a small decrease
(5%) has occurred since 1990. Uterine cancer                             HALLMARKS OF MALIGNANT
death rates have been going down, primarily                              DISEASES
through earlier detection and treatment of cer-
vical cancer. Female colon cancer mortality has                          Malignant neoplasms or cancers have several
been decreasing, but the reasons for this aren’t                         distinguishing features that enable the patholo-
clear. As in males, stomach cancer mortality in                          gist or experimental cancer biologist to charac-
women has been going down for many years.                                terize them as abnormal. The most common
10                                                                                                         CANCER BIOLOGY

                                                                                                  Lung and Bronchus


     Rate per 100,000 Males


                               40                                                                          Prostate
                                                                      Colon and Rectum


                                              Leukemia                                   Liver
                                                                         Year of Death

                                    Figure 1–3. Annual age-adjusted cancer death rates* among males for se-
                                    lected cancer types, United States, 1930 to 2001. *Rates are age adjusted to
                                    the 2000 U.S. standard population. Because of changes in ICD coding, nu-
                                    merator information has changed over time rates for cancers of the lung and
                                    bronchus, colon and rectum, and liver are affected by these changes. (From
                                    U.S. Mortality Public Use Data Tapes, 1960 to 2001, U.S. Mortality Volumes,
                                    1930 to 1959, National Center for Health Statistics, Centers for Disease
                                    Control and Prevention, with permission.)

types of human neoplasms derive from epitheli-                                The term neoplasm, meaning new growth, is
um, that is, the cells covering internal or external                       often used interchangeably with the term tumor
surfaces of the body. These cells have a sup-                              to signify a cancerous growth. It is important to
portive stroma of blood vessels and connective                             keep in mind, however, that tumors are of two
tissue. Malignant neoplasms may resemble nor-                              basic types: benign and malignant. The ability to
mal tissues, at least in the early phases of their                         distinguish between benign and malignant tu-
growth and development. Neoplastic cells can                               mors is crucial in determining the appropriate
develop in any tissue of the body that contains                            treatment and prognosis of a patient who has
cells capable of cell division. Though they may                            a tumor. The following are features that differ-
grow fast or slowly, their growth rate frequently                          entiate a malignant tumor from a benign tumor:
exceeds that of the surrounding normal tissue.
This is not an invariant property, however, be-                               1. Malignant tumors invade and destroy ad-
cause the rate of cell renewal in a number of                                    jacent normal tissue; benign tumors grow
normal tissues (e.g., gastrointestinal tract epi-                                by expansion, are usually encapsulated,
thelium, bone marrow, and hair follicles) is as                                  and do not invade surrounding tissue.
rapid as that of a rapidly growing tumor.                                        Benign tumors may, however, push aside
CHARACTERISTICS OF HUMAN CANCER                                                                                                11

  Rate per 100,000 Females

                              45                                                                       Lung and Bronchus
                              35                                                 Breast
                                                                                                            Colon and Rectum
                              20                                      Uterus
                              10             Ovary
                                                                       Year of Death

                                   Figure 1–4. Annual age-adjusted cancer death rates* among females for
                                   selected cancer types, United States, 1930 to 2001. *Rates are age adjusted to
                                   the 2000 U.S. standard population. Because of ICD coding, numerator
                                   information has changed over time, rates for cancers of the uterus, ovary,
                                   lung and bronchus, and colon and rectum are affected by these changes.
                                   Uterus cancers are for uterine cervix and uterine corpus combined. (From
                                   U.S. Mortality Public Use Data Tapes, 1960 to 2001, U.S. Mortality Volumes,
                                   1930 to 1959, National Center for Health Statistics, Centers for Disease
                                   Control and Prevention, with permission.)

     normal tissue and may become life threat-                                    Some malignant neoplastic cells at first
     ening if they press on nerves or blood                                       structurally and functionally resemble the
     vessels or if they secrete biologically active                               normal tissue in which they arise. Later, as
     substances, such as hormones, that alter                                     the malignant neoplasm progresses, invades
     normal homeostatic mechanisms.                                               surrounding tissues, and metastasizes, the
  2. Malignant tumors metastasize through lym-                                    malignant cells may bear less resemblance
     phatic channels or blood vessels to lymph                                    to the normal cell of origin. The develop-
     nodes and other tissues in the body. Be-                                     ment of a less well-differentiated malignant
     nign tumors remain localized and do not                                      cell in a population of differentiated normal
     metastasize.                                                                 cells is sometimes called dedifferentiation.
  3. Malignant tumor cells tend to be ‘‘anaplas-                                  This term is probably a misnomer for the
     tic,’’ or less well differentiated than normal                               process, because it implies that a differen-
     cells of the tissue in which they arise. Be-                                 tiated cell goes backwards in its develop-
     nign tumors usually resemble normal tissue                                   mental process after carcinogenic insult. It
     more closely than malignant tumors do.                                       is more likely that the anaplastic malignant
12                                                                                         CANCER BIOLOGY

        cell type arises from the progeny of a tissue        Differential diagnosis of cancer from a benign
        ‘‘stem cell’’ (one that still has a capacity for   tumor or a nonneoplastic disease usually involves
        renewal and is not yet fully differentiated),      obtaining a tissue specimen by biopsy, surgical
        which has been blocked or diverted in its          excision, or exfoliative cytology. The latter is an
        pathway to form a fully differentiated cell.       examination of cells obtained from swabbings,
            Examples of neoplasms that maintain a          washings, or secretions of a tissue suspected to
        modicum of differentiation include islet cell      harbor cancer: the ‘‘Pap test’’ involves such an
        tumors of the pancreas that still make insu-       examination.
        lin, colonic adenocarcinoma cells that form
        glandlike epithelial structures and secrete
        mucin, and breast carcinomas that make             CLASSIFICATION OF HUMAN
        abortive attempts to form structures resem-        CANCERS
        bling mammary gland ducts. Hormone-
        producing tumors, however, do not respond          Although the terminology applied to neoplasms
        to feedback controls regulating normal tis-        can be confusing for a number of reasons, certain
        sue growth or to negative physiologic feed-        generalizations can be made. The suffix oma,
        back regulating hormonal secretion. For            applied by itself to a tissue type, usually indicates
        example, an islet cell tumor may continue          a benign tumor. Some malignant neoplasms,
        to secrete insulin in the face of extreme          however, may be designated by the oma suffix
        hypoglycemia, and an ectopic adrenocortio-         alone; these include lymphoma, melanoma, and
        cotropic hormone (ACTH)-producing lung             thymoma. Rarely, the oma suffix is used to de-
        carcinoma may continue to produce ACTH             scribe a nonneoplastic condition such as granu-
        even though circulating levels of adreno-          loma, which is often not a true tumor, but a mass
        cortical steroids are sufficient to cause           of granulation tissue resulting from chronic in-
        Cushing’s syndrome (see Chapter 6). Many           flammation or abscess. Malignant tumors are
        malignant neoplasms, particularly the more         indicted by the terms carcinoma (epithelial in
        rapidly growing and invasive ones, only            origin) or sarcoma (mesenchymal in origin) pre-
        vaguely resemble their normal counterpart          ceded by the histologic type and followed by the
        tissue structurally and functionally. They         tissue of origin. Examples of these include ade-
        are thus said to be ‘‘undifferentiated’’ or        nocarcinoma of the breast, squamous cell carci-
        ‘‘poorly differentiated.’’                         noma of the lung, basal cell carcinoma of skin,
     4. Malignant tumors usually, though not in-           and leiomyosarcoma of the uterus. Most human
        variably, grow more rapidly than benign            malignancies arise from epithelial tissue. Those
        tumors. Once they reach a clinically detect-       arising from stratified squamous epithelium are
        able stage, malignant tumors generally show        designated squamous cell carcinomas, whereas
        evidence of significant growth, with involve-       those emanating from glandular epithelium are
        ment of surrounding tissue, over weeks or          termed adenocarcinomas. When a malignant
        months, whereas benign tumors often grow           tumor no longer resembles the tissue of origin, it
        slowly over several years.                         may be called anaplastic or undifferentiated. If a
                                                           tumor is metastatic from another tissue, it is
   Malignant neoplasms continue to grow even               designated, for example, an adenocarcinoma of
in the face of starvation of the host. They press          the colon metastatic to liver. Some tumors arise
on and invade surrounding tissues, often inter-            from pluripotential primitive cell types and may
rupting vital functions; they metastasize to vital         contain several tissue elements. These include
organs, for example, brain, spine, and bone mar-           mixed mesenchymal tumors of the uterus, which
row, compromising their functions; and they in-            containcarcinomatous and sarcomatous elements,
vade blood vessels, causing bleeding. The most             and teratocarcinomas of the ovary, which may
common effects on the patient are cachexia                 contain bone, cartilage, muscle, and glandular
(extreme body wasting), hemorrhage, and in-                epithelium.
fection. About 50% of terminal patients die from              Neoplasms of the hematopoietic system usu-
infection (see Chapter 8).                                 ally have no benign counterparts. Hence the
CHARACTERISTICS OF HUMAN CANCER                                                                        13

terms leukemia and lymphoma always refer to a          primary tumor, or if other diagnostic techniques
malignant disease and have cell-type designa-          fail to reveal other tumor masses, the clinician
tions such as acute or chronic myelogenous             has to treat blindly, and thus might not choose
leukemia, Hodgkin’s or non-Hodgkin’s lym-              the best mode of therapy.
phoma, and so on. Similarly, the term melanoma            Another consideration is the accessibility of a
always refers to a malignant neoplasm derived          tumor. If a tumor is surgically inaccessible or too
from melanocytes.                                      close to vital organs to allow complete resection,
                                                       surgical removal is impossible. For example, a
                                                       cancer of the common bile duct or head of the
MACROSCOPIC AND MICROSCOPIC                            pancreas is often inoperable by the time it is
FEATURES OF NEOPLASMS                                  diagnosed because these tumors invade and at-
                                                       tach themselves to vital structures early, thus
The pathologist can gain valuable insights about       preventing curative resection. Similarly, if ad-
the nature of a neoplasm by careful examination        ministered anticancer drugs cannot easily reach
of the overall appearance of a surgical specimen.      the tumor site, as is the case with tumors growing
Often, by integrating the clinical findings with        in the pleural cavity or in the brain, these agents
macroscopic characteristics of a tumor, a ten-         might not be able to penetrate in sufficient
tative differential diagnosis can be reached.          quantities to kill the tumor cells.
Also, notation of whether the tumor is encap-             The site of the primary tumor also frequently
sulated, has extended through tissue borders, or       determines the mode of, and target organs for,
reached to the margins of the excision provides        metastatic spread. In addition to local spread,
important diagnostic information.                      cancers metastasize via lymphatic channels or
   The location of the anatomic site of the neo-       blood vessels. For example, carcinomas of the
plasm is important for several reasons. The site of    lung most frequently metastasize to regional
the tumor dictates several things about the clin-      lymph nodes, pleura, diaphragm, liver, bone,
ical course of the tumor, including (1) the likeli-    kidneys, adrenals, brain, thyroid, and spleen.
hood and route of metastatic spread, (2) the           Carcinomas of the colon metastasize to regional
effects of the tumor on body functions, and (3)        lymph nodes, and by local extension, they ul-
the type of treatment that can be employed. It is      cerate and obstruct the gastrointestinal tract.
also important to determine whether the ob-            The most common site of distant metastasis of
served tumor mass is the primary site (i.e., tissue    colon carcinomas is the liver, via the portal vein,
of origin) of the tumor or a metastasis. A primary     which receives much of the venous return from
epidermoid carcinoma of the lung, for example,         the colon and flows to the liver. Breast carci-
would be treated differently and have a different      nomas most frequently spread to axillary lymph
prognosis than an embryonal carcinoma of the           nodes, the opposite breast through lymphatic
testis metastatic to the lung. It is not always easy   channels,lungs,pleura,liver,bone,adrenals,brain,
to determine the primary site of a neoplasm,           and spleen.
particularly if the tumor cells are undifferenti-         Some tissues are more common sites of me-
ated. The first signs of a metastatic tumor may         tastasis than others. Because of their abundant
be a mass in the lung noted on CT scan or a            blood and lymphatic supply, as well as their
spontaneous fracture of a vertebra that had been       function as ‘‘filters’’ in the circulatory system,
invaded by cancer cells. Because the lungs and         the lungs and the liver are the most common
bones are frequent sites of metastases for a vari-     sites of metastasis from tumors occurring in
ety of tumors, the origin of the primary tumor         visceral organs. Metastasis is usually the single
may not be readily evident. This is a very diffi-       most important criterion determining the pa-
cult clinical situation, because to cure the pa-       tient’s prognosis. In breast carcinoma, for ex-
tient or to produce long-term remission, the           ample, the 5-year survival rate for patients with
oncologist must be able to find and remove or           localized disease and no evidence of axillary
destroy the primary tumor to prevent its con-          lymph node involvement is about 85%; but when
tinued growth and metastasis. If histologic ex-        more than four axillary nodes are involved, the
amination does not reveal the source of the            5-year survival is about 30%, on average.3
14                                                                                  CANCER BIOLOGY

   The anatomic site of a tumor will also deter-            common in malignant tissue than in nor-
mine its effect on vital functions. A lymphoma              mal tissue.
growing in the mediastinum may press on ma-              5. Obvious evidence of invasion of normal
jor blood vessels to produce the superior vena              tissue by a neoplasm may be seen, indi-
caval syndrome, manifested by edema of the                  cating that the tumor has already become
neck and face, distention of veins of the neck,             invasive and may have metastasized.
chest, and upper extremities, headache, dizzi-
ness, and fainting spells. Even a small tumor
growing in the brain can produce such dramatic        GRADE AND STAGE OF NEOPLASMS
central nervous system effects as localized weak-
ness, sensory loss, aphasia, or epileptic-like sei-
                                                      Histologic Grade of Malignancy
zures. A lung tumor growing close to a major
bronchus will produce airway obstruction ear-         The histologic grading of malignancy is based on
lier than one growing in the periphery of the         the degree of differentiation of a cancer and on
lung. A colon carcinoma may invade surround-          an estimate of the growth rate as indicated by
ing muscle layers of the colon and constrict the      the mitotic index. It was generally believed that
lumen, causing intestinal obstruction. One of the     less differentiated tumors were more aggressive
frequent symptoms of prostatic cancer is inabil-      and more metastatic than more differentiated
ity to urinate normally.                              tumors. It is now appreciated that this is an
   The cytologic criteria that enable the pathol-     oversimplification and, in fact, not a very accu-
ogist to confirm the diagnosis, or at least to         rate way to assess the degree of malignancy for
suspect that cancer is present (thus indicating       certain kinds of tumors. However, for certain
the need for further diagnostic tests), are as        epithelial tumors, such as carcinomas of the
follows:                                              cervix, uterine endometrium, colon, and thy-
                                                      roid, histologic grading is a fairly accurate index
     1. The morphology of cancer cells is usually     of malignancy and prognosis. In the case of
        different from and more variable than that    epidermoid carcinomas, for example, in which
        of their counterpart normal cells from the    keratinization occurs, keratin production pro-
        same tissue. Cancer cells are more vari-      vides a relatively facile way to determine the
        able in size and shape.                       degree of differentiation. On the basis of this
     2. The nucleus of cancer cells is often larger   criterion, and others like it, tumors have been
        and the chromatin more apparent (‘‘hy-        classified as grade I (75% to 100% differentia-
        perchromatic’’) than the nucleus in nor-      tion), grade II (50% to 75%), grade III (25% to
        mal cells; the nuclear-to-cytoplasmic ratio   50%), and grade IV (0% to 25%).4 More recent
        is often higher; and the cancer cell nuclei   methods of malignancy grading also take into
        contain prominent, large nucleoli.            consideration mitotic activity, amount of infil-
     3. The number of cells undergoing mitosis is     tration into surrounding tissue, and amount of
        usually greater in a population of cancer     stromal tissue in or around the tumor. The chief
        cells than in a normal tissue population.     value of grading is that it provides, for certain
        Twenty or more mitotic figures per 1000        cancers, a general guide to prognosis and an
        cells would not be an uncommon finding         indicator of the effectiveness of various thera-
        in cancerous tissue, whereas less than 1      peutic approaches.
        per 1000 is usual for benign tumors or
        normal tissue.4 This number, of course,
                                                      Tumor Staging
        would be higher in normal tissues that
        have a high growth rate, such as bone         Although the classification of tumors based on
        marrow and crypt cells of the gastroin-       the preceding descriptive criteria helps the on-
        testinal mucosa.                              cologist determine the malignant potential of a
     4. Abnormal mitosis and ‘‘giant cells,’’ with    tumor, judge its probable course, and determine
        large, pleomorphic (variable size and         the patient’s prognosis, a method of discovering
        shape) or multiple nuclei, are much more      the extent of disease on a clinical basis and a
CHARACTERISTICS OF HUMAN CANCER                                                                        15

universal language to provide standardized cri-       patient; however, as more information becomes
teria among physicians are needed. Attempts to        available following a more extensive workup,
develop an international language for describing      such as a biopsy or surgical exploration, this in-
the extent of disease have been carried out by        formation is, of course, taken into consideration
two major agencies—the Union Internationale           in determining treatment and estimating prog-
Contre le Cancer (UICC) and the American Joint        nosis. Staging provides a useful way to estimate at
Committee for Cancer Staging and End Results          the outset what a patient’s clinical course and
Reporting (AJCCS). Some of the objectives of the      initial treatment should be. The actual course of
classification system developed by these groups        the disease indicates its true extent. As more is
are (1) to aid oncologists in planning treatment;     learned about the natural history of cancers, and
(2) to provide categories for estimating prognosis    as more sophisticated diagnostic techniques be-
and evaluating results of treatment; and (3) to       come available, the criteria for staging will likely
facilitate exchange of information.5 Both the         change and staging should become more accu-
UICC and AJCCS schemes use the T, N, M                rate (see Chapter 7).
classification system, in which T categories define        It is important to remember that staging does
the primary tumor; N, the involvement of re-          not mean that any given cancer has a predict-
gional lymph nodes; and M, the presence or ab-        able, ineluctable progression. Although some
sence of metastases. The definition of extent of       tumors may progress in a stepwise fashion from
malignant disease by these categories is termed       a small primary tumor to a larger primary tumor,
staging. Staging defines the extent of tumor           and then spread to regional nodes and distant
growth and progression at one point in time; four     sites (i.e., progressing from stage I to stage IV),
different methods are involved:                       others may spread to regional nodes or have
                                                      distant metastases while the primary tumor is
   1. Clinical staging: estimation of disease pro-
                                                      microscopic and clinically undetectable. Thus,
      gression based on physical examination,
                                                      staging is somewhat arbitrary, and its effective-
      clinical laboratory tests, X-ray films, and
                                                      ness is really based on whether it can be used as
      endoscopic examination.
                                                      a standard to select treatment and to predict the
   2. Tumor imaging: evaluation of progression
                                                      course of disease.
      based on sophisticated radiography—for
                                                         Although the exact criteria used vary with
      example, CT scans, arteriography, lymph-
                                                      each organ site, the staging categories listed be-
      angiography, and radioisotope scanning;
                                                      low represent a useful generalization.6
      MRI; and PET.
   3. Surgical staging: direct exploration of the
                                                      Stage I (T1 N0 M0): Primary tumor is limited to
      extent of the disease by surgical proce-
                                                        the organ of origin. There is no evidence of
                                                        nodal or vascular spread. The tumor can
   4. Pathologic staging: use of biopsy proce-
                                                        usually be removed by surgical resection.
      dures to determine the degree of spread,
                                                        Long-term survival is from 70% to 90%.
      depth of invasion, and involvement of
                                                      Stage II (T2 N1 M0): Primary tumor has spread
      lymph nodes.
                                                        into surrounding tissue and lymph nodes
   These methods of staging are not used inter-         immediately draining the area of the tumor
changeably, and their use depends on agreed-            (‘‘first-station’’ lymph nodes). The tumor is
upon procedures for each type of cancer. For            operable, but because of local spread, it may
example, operative findings are used to stage cer-       not be completely resectable. Survival is 45%
tain types of cancer (e.g., ovarian carcinomas) and     to 55%.
lymphangiography is required to stage Hodgkin’s       Stage III (T3 N2 M0): Primary tumor is large,
disease. Although this means that different stag-       with fixation to deeper structures. First-station
ing methods are used to stage different tumors,         lymph nodes are involved; they may be more
each method is generally agreed on by oncolo-           than 3 cm in diameter and fixed to underlying
gists, thus allowing a comparison of data from          tissues. The tumor is not usually resectable,
different clinical centers. Once a tumor is clini-      and part of the tumor mass is left behind.
cally staged, it is not usually changed for that        Survival is 15% to 25%.
16                                                                                      CANCER BIOLOGY

Stage IV (T4 N3 Mþ): Extensive primary tumor            foreseeable future. The ultimate diagnosis, prog-
  (may be more than 10 cm in diameter) is pres-         nosis, and selection of a treatment course will
  ent. It has invaded underlying or surrounding         depend on this.
  tissues. Extensive lymph node involvement has            Although the TNM system is useful for staging
  occurred, and there is evidence of distant me-        malignant tumors, it is primarily based on a tem-
  tastases beyond the tissue of origin of the pri-      poral model that assumes a delineated progres-
  mary tumor. Survival is under 5%.                     sion over time from a small solitary lesion to
                                                        one that is locally invasive, then involves lymph
   The criteria for establishing lymph node in-         nodes, and finally spreads through the body.
volvement (N categories) are based on size, firm-        While this is true for some cancers, the linearity
ness, amount of invasion, mobility, number of           of this progression model is an oversimplifica-
nodes involved, and distribution of nodes in-           tion. For example, some patients have aggressive
volved (i.e., ipsilateral, contralateral, distant in-   tumors almost from the outset and may die be-
volvement): N0 indicates that there is no evidence      fore lymph node involvement becomes evident,
of lymph node involvement; N1 indicates that            whereas others may have indolent tumors that
there are palpable lymph nodes with tumor in-           grow slowly and remain localized for a long time,
volvement, but they are usually small (2 to 3 cm        even though they may become large.
in diameter) and mobile; N2 indicates that there           In addition, the TNM staging system does not
are firm, hard, partially movable nodes (3 to 5 cm       take into account the molecular markers that we
in diameter), partially invasive, and they may feel     now know can more clearly define the status of a
as if they were matted together; N3 indicates that      cancer, e.g., its gene array and proteomic pro-
there are large lymph nodes (over 5 cm in diam-         files (see Chapter 7). Nor does the TNM system,
eter) with complete fixation and invasion into           as a prognostic indicator, take into account the
adjacent tissues; N4 indicates extensive nodal in-      varied responsiveness of tumors to various
volvement of contralateral and distant nodes.           therapeutic modalities. Thus, treatment choices
   The criteria applied to metastases (M cate-          and prognostic estimates should be based more
gories) are as follows: M0, no evidence of metas-       on the molecular biology of the tumor than the
tasis; M1, isolated metastasis in one other organ;      tumor’s size, location, or nodal status at the time
M2, multiple metastases confined to one organ,           of diagnosis.7
with minimal functional impairment; M3, mul-
tiple organs involved with no to moderate
functional impairment; M4, multiple organ in-           References
volvement with moderate to severe functional
                                                        1. A. Jemal, T. Murray, E. Ward, A. Samuels, R. C.
impairment. Occasionally a subscript is used to            Tivari, A. Ghafoor, E. J. Feuer, and M. J. Thun:
indicate the site of metastasis, such as Mp, Mh,           Cancer statistics, 2005. CA Cancer J Clin 55:10,
Mo for pulmonary, hepatic, and osseous metas-              2005.
tases, respectively.                                    2. P. A. Wingo, C. J. Cardinez, S. H. Landis, R. T.
   Diagnostic procedures are getting more so-              Greenlee, A. G. Ries, R. N. Anderson, and M. J.
                                                           Thun: Long-term trends in cancer mortality in the
phisticated all the time. Improved CT, MRI, and            United States, 1930–1998. Cancer 97:3133, 2003.
PET scanners, as well as ultrasound techniques,         3. I. C. Henderson and G. P. Canellos: Cancer of the
are being developed to better localize tumors              breast—The past decade. N Engl J Med 302:17,
and determine their metabolic rate. One can                1980.
visualize the day when ‘‘noninvasive biopsies,’’        4. S. Warren: Neoplasms. In W. A. D. Anderson, ed.:
                                                           Pathology. St. Louis: C. V. Mosby, 1961, pp. 441–
based on the ability to carry out molecular and            480.
cellular imaging by means of external detection         5. P. Rubin: A unified classification of cancers: An
of internal signals, may at least partially replace        oncotaxonomy with symbols. Cancer 31:963, 1973.
the need for biopsy or surgical specimens to get        6. P. Rubin: Statement of the clinical oncologic prob-
diagnostic information (see Chapter 7). There              lem. In P. Rubin, ed.: Clinical Oncology. Roche-
                                                           ster: American Cancer Society, 1974, pp. 1–25.
will always be the need, however, for clinical          7. H. B. Burke: Outcome prediction and the future
pathologists to examine tissue specimens to con-           of the TNM staging system. J Natl Cancer Inst
firm noninvasive procedures, at least for the               96:1408, 2004.

Causes of Cancer

Perhaps the most important question in cancer              proliferation must occur, but it may originally be
biology is what causes the cellular alterations            limited by host defenses or lack of access to the
that produce a cancer. The answer to this ques-            host’s blood supply. During the process of tu-
tion has been elusive. If the actual cause of these        mor progression, however, escape from the host’s
alterations were known, the elimination of fac-            defense mechanisms and vascularization of the
tors that produce cancer and the development               growing tumor ultimately occur.
of better treatment modalities would likely fol-              The genetic instability of cancer cells leads to
low. Cancer prevention might become a reality.             the emergence of a more aggressively growing
   A cancerous growth has a number of predict-             tumor frequently characterized by the appear-
able properties. The incidence rates of various            ance of poorly differentiated cells with certain
cancers are strongly related to environmental fac-         properties of a more embryonic phenotype. Dur-
tors and lifestyle, and cancers have certain growth        ing tumor progression, considerable biochemical
characteristics, among which are the abilities             heterogeneity becomes manifest in the growing
to grow in an uncontrolled manner, invade sur-             tumor and its metastases, even though all the
rounding tissues, and metastasize. Also, when              neoplastic cells may have arisen originally from a
viewed microscopically, cancer cells appear to be          single deranged cell. Any theory that seeks to
less well differentiated than their normal coun-           explain the initiation of cancer and its progression
terparts and to have certain distinguishing fea-           must take these observations into consideration.
tures, such as large nuclei and nucleoli. Most                In this chapter, we will examine what is known
cancers arise from a single clone of cells, whose          about various chemical, physical, and viral carci-
precursor may have been altered by insult with a           nogenic agents and discuss the putative mech-
carcinogen. In most cases cancer is a disease of           anisms by which they cause cancer.
aging. The average age at diagnosis is over 65 and
malignant cancers arise from a lifetime accumu-
lation of ‘‘hits’’ on a person’s DNA. These hits           THE THEORY OF ‘‘HITS’’
may result from genetic susceptibility to envi-
ronmental agents such as chemicals; radiation; or          As noted above, with the exception of childhood
viral, bacterial, or parasitic infections; or from         malignancies such as leukemias and sarcomas
endogenously generated agents such as oxygen               that occur in children, cancer incidence in-
radicals. It is often said that we would all get           creases with age. Most of the common adult solid
cancer if we lived long enough.                            tumors begin to increase after age 45 and go up
   There is frequently a long latent period, in            logarithmically with age after that, as shown for
some cases 20 years or more, between the ini-              colorectal cancer (Fig. 2–1).1 This has led to the
tiating insult and the appearance of a clinically          idea that it takes multiple cellular hits to explain
detectable tumor. During this time, cellular               the age-related incidence of malignancy. Most

18                                                                                                   CANCER BIOLOGY

                              white males (1984)                         black males (1984)

                   2 10 5

                              white females (1984)                       black females (1984)


                                  20      40         60    80      100       20      40         60    80   100
                                            age (years)                                age (years)

                            Figure 2–1. Observed (squares) and predicted (lines) incidence of colorectal
                            cancer by race and gender in the Surveillance, Epidemiology, and End Results
                            (SEER) registry (1984). (From Luebeck and Moolgavkar,1 with permission.)

of these hits are thought to be mutational in                      sine kinase promoting cell proliferation (see
origin and to result from chromosomal damage                       Chapter 4). Thus, CML appears to be triggered
or base changes in DNA. The number of hits                         by this one-hit event and is probably the reason
needed to produce the initiation of a malignant                    why the drug Gleevec, which targets this kinase,
event may vary from one to six or more. How-                       is effective as a single agent in CML.
ever, progression to a full-blown invasive met-                       A second example is retinoblastoma. There are
astatic cancer almost always requires multiple                     two forms of this disease, hereditary and spon-
hits. A few examples will make the point.                          taneous. Both forms appear to require two ini-
   In chronic myleogenous leukemia (CML),                          tiating genetic events, leading Knudson, who
there is an inciting chromosomal transloca-                        studied this disease in detail, to postulate the
tion that involves a piece of chromosome 22 be-                    two-hit hypothesis.4 In the hereditary form, one
ing lost. This was first observed by Nowell and                     genetic mutation is inherited at birth and a
Hungerford,2 who named this small chromo-                          second one occurs later (Fig. 2–3). This must be
some the Philadelphia chromosome. It was later                     the case, since every cell in the eye contains
shown by Rowley3 that this was a reciprocal                        the hereditary mutation, but only three to four
translocation between chromosomes 9 and 22                         tumors on average develop in a retinal cell pop-
(Fig. 2–2), which produces a chimeric protein                      ulation of several million cells in affected indi-
called Bcr/Abl that is a constitutively active tyro-               viduals.
CAUSES OF CANCER                                                                                       19

                                                       unlikely for most solid tumors. Most likely, mul-
                                                       tiple aberrant cell signaling pathways will need to
                                                       be inhibited for effective chemotherapeutic reg-
                                                       imens to be achieved. However, if there are
                                                       identifiable time intervals between the multiple
                                                       hits that lead to cancer, perhaps detectable by
                                                       early screening for surrogate markers of progres-
                                                       sion, there may be a window of opportunity for
                                                       preventive agents (see Chapter 9).

                                                       CHEMICAL CARCINOGENESIS

                                                       Historical Perspectives
                                                       Carcinogenic chemicals and irradiation (ioniz-
                                                       ing and ultraviolet) are known to affect DNA
                                                       and to be mutagenic under certain conditions.
                                                       Thus, one of the long-standing theories of car-
                                                       cinogenesis is that cancer is caused by a genetic
                                                       mutation; however, it is now known that epige-
                                                       netic mechanisms are also involved.
                                                          Evidence that chemicals can induce cancer in
                                                       humans has been accumulating since the six-
                                                       teenth century (reviewed in Reference 7). In
                                                       1567, Paracelsus described a ‘‘wasting disease of
                                                       miners’’ and proposed that exposure to some-
                                                       thing in the mined ores caused the condition. A
                                                       similar condition was described in 1926 in Sax-
Figure 2–2. A comparison of karyotypes. a. Chronic
myelogenous leukemia, showing the typical 9;22
                                                       ony and was later identified as the ‘‘lung cancer
translocation and an otherwise normal karyotype. b.    of the Schneeberg mines.’’ It was realized much
Non–small cell carcinoma of the lung, showing ab-      later that the cause of this was probably expo-
normalities of both number and structure. The arrows   sure to radon. Nevertheless, Paraclesus could
indicate aberrant chromosomes. (From Knudson,4         probably be called ‘‘the father of occupational
reprinted by permission from Macmillan Publishers
                                                       carcinogenesis.’’ It is Bernadini Ramazzini, how-
                                                       ever, who published a systematic account of
                                                       work-related diseases in 1700, who is more
                                                       logically considered the founder of occupational
   Most adult solid cancers (e.g., colon, lung,        medicine.7
breast, prostate) likely require several hits to          Later in the eighteenth century, the first direct
achieve a full malignant state. The best example       observation associating chemicals was made by
of this is colon cancer, for which at least five        John Hill, who in 1761 noted that nasal cancer
hits appear to be required to produce an inva-         occurred in people who used snuff excessively.
sive carcinoma (Fig. 2–4). Because of genetic          In 1775, Percival Pott reported a high incidence
instability, a characteristic of most solid cancers,   of scrotal skin cancer among men who had spent
many more genetic alterations are frequently           their childhood as chimney sweeps. One hun-
seen in later stages of cancer progression.5 This      dred years later, von Volkman, in Germany, and
has been ascribed to a ‘‘mutator phenotype’’ ob-       Bell, in Scotland, observed skin cancer in work-
served in many cancers.6 In contrast to single         ers whose skin was in continuous contact with
genetic defect cancers such as CML, the prospect       tar and paraffin oils, which we now know contain
of finding effective single therapeutic agents is       polycyclic aromatic hydrocarbons. In 1895, Rehn
20                                                                                      CANCER BIOLOGY

     Hereditary                                        Nonhereditary

                                               Tumor                                            Tumor

               Figure 2–3. Two-hit tumor formation in both hereditary and nonhereditary
               retinoblastoma. A ‘‘one-hit’’ clone is a precursor to the tumor in nonheredi-
               tary retinoblastomas, whereas all retinoblasts (indeed, all cells) are one-hit
               clones in hereditary retinoblastoma. (From Knudson,4 reprinted by permis-
               sion from Macmillan Publishers Ltd.)

reported the development of urinary bladder            is a carcinogen, and from the identification of
cancer in aniline dye workers in Germany. Sim-         the carcinogen 3,4-benzpyrene in coal tar by
ilar observations were later made in a number of       Cook, Hewitt, and Hieger. Induction of tumors
countries and established a relationship between       by other chemical and hormonal carcinogens
heavy exposure to 2-naphthylamine, benzidine,          was described in the 1930s, including the induc-
or 4-aminobiphenyl and bladder cancer. Thus,           tion of liver tumors in rats and mice with 20 ,
the first observations of chemically induced can-       3-dimethyl-4-aminoazobenzene by Yoshida, of
cer were made in humans. These observations            urinary bladder cancer in dogs with 2-naphthyl-
led to attempts to induce cancer in animals with       amine by Hueper, Wiley, and Wolfe, and of
chemicals. One of the first successful attempts         mammary cancer in male mice with estrone by
was made in 1915, when Yamagiwa and Ichikawa           Lacassagne. The list of known carcinogenic
induced skin carcinomas by the repeated appli-         chemicals expanded in the 1940s with the dis-
cation of coal tar to the ears of rabbits. This and    covery of the carcinogenicity of 2-acetylamino-
similar observations by other investigators led to     fluorene, halogenated hydrocarbons, urethane,
a search for the active carcinogen in coal tar and     beryllium salts, and certain anticancer alkylating
to the conclusion that the carcinogenic agents in      agents. Since the 1940s, various nitrosamines,
tars are the polycyclic aromatic hydrocarbons.         intercalating agents, nickel and chromium com-
Direct evidence for that came in the 1930s from        pounds, asbestos, vinyl chloride, diethylstilbes-
the work of Kennaway and Heiger, who demon-            trol, and certain naturally occurring substances,
strated that synthetic 1,2,5,6-dibenzanthracene        such as aflatoxins, have been added to the list of
CAUSES OF CANCER                                                                                       21

                              Normal colon cells:      tain carcinogenic agents is necessary to produce
                              two APC mutations
                                                       the ‘‘ultimate carcinogen’’ that actually reacts
                                                       with crucial molecules in target cells. With the
                                                       exception of the very chemically reactive alky-
                                                       lating agents, which are activated in aqueous
                              Adenomatous polyp:       solution at physiologic pH (e.g., N-methyl-N-
                              one RAS mutation         nitrosourea), and the agents that intercalate into
                                                       the DNA double helix by forming tight non-
                                                       covalent bonds (e.g., daunorubicin), most of the
                                                       known chemical carcinogens undergo some
                              Dysplastic polyp:
                              two TP 53 mutations      metabolic conversions that appear to be re-
                                                       quired for their carcinogenic action. Some ex-
                                                       amples of these metabolic conversions are given
                              Colon carcinoma:         next.
                              Other events;
                                                            Donors of Simple Alkyl Groups
                                                       Included in this group are the dialkylnitro-
                                                       samines, dialkylhydrazines, aryldialkyltriasenes,
                                                       alkylnitrosamides, and alkylnitrosimides. The al-
                              Metastatic carcinoma     kylnitrosamides and alkylnitrosimides do not
                                                       require enzymatic activation because they can
                                                       react directly with water or cellular nucleophilic
                                                       groups. The alkylnitrosamines, alkylhydrazines,
Figure 2–4. A possible five-hit scenario for colorec-
tal cancer, showing the mutational events that cor-    and alkyltriazenes, however, undergo an enzyme-
relate with each step in the adenoma–carcinoma         mediated activation step to form the reactive
sequence. (From Knudson,4 reprinted by permission      electrophile (Fig. 2–6). These agents are meta-
from Macmillan Publishers Ltd.)                        bolically dealkylated by the mixed-function oxi-
                                                       dase system in the microsomal fraction (endo-
known carcinogens. A list of some known human          plasmic reticulum) of cells, primarily liver cells.
carcinogens is found in Table 2–1, and the struc-      The monoalkyl derivatives then undergo a non-
tures of some known carcinogens are shown in           enzymatic, spontaneous conversion to mono-
Figure 2–5.                                            alkyldiazonium ions that donate an alkyl to cel-
                                                       lular nucleophilic groups in DNA, RNA, and
Metabolic Activation of
Chemical Carcinogens
                                                            Cytochrome P-450–Mediated
As studies on the reactions of carcinogens with
cellular macromolecules progressed, it became
apparent that most of these interactions resulted      A number of carcinogenic chemicals are chemi-
from covalent bond formation between an elec-          cally inert nucleophilic agents until they are con-
trophilic form of the carcinogen and the nucle-        verted to active nucleophiles by the cytochrome
ophilic sites in proteins (e.g., sulfur, oxygen, and   p-450–dependent mixed function oxidases, or
nitrogen atoms in cysteine, tyrosine, and histi-       CYPs So far, 57 genes encoding these enzymes
dine, respectively) and nucleic acids (e.g., pu-       have been identified in the human genome. The
rine or pyrimidine ring nitrogens and oxygens).        CYPs most involved in carcinogen activation are
Frequently, the parent compound itself did not         CYP1A1, 1A2, 1B1, 2A6, and 3A4. A wide variety
interact in vitro with macromolecules until it         of chemical carcinogens such as aromatic and
had been incubated with liver homogenates or           heterocyclic amines, aminoazo dyes, polycyclic
liver microsomal fractions. These studies led to       aromatic hydrocarbons, N-nitrosamines, and hal-
the realization that metabolic activation of cer-      ogenated olefins are activated by one or more of
22                                                                                      CANCER BIOLOGY

Table 2–1. Selected Human Chemical Carcinogens
                                                                                       Affected Organs
Compounds                             Main Sources and Uses                            and Cancer Type
o-Aminoazotoluene                     Pigments, coloring oils; immunosuppressant       Liver, lung, bladder
N,N-dimethyl-4-aminoazobenzene        Color polishes, waxes (no longer in use)         Lung, liver
Melphalan                             Chemotherapy                                     Leukemia
Thiotepa                              Chemotherapy (no longer in use)                  Leukemia
2-Naphthylamine                       Dyes; antioxidant (no longer in use)             Bladder
4-Aminobiphenyl                       Dyes; antioxidant (no longer in use), research   Bladder
2-Acetylaminofluorene                  Model compound; tested as a pesticide            Liver, bladder
Benzo[a]pyrene                        Coal tar, roofing, cigarette smoke                Skin, lung, stomach
2,3,7,8-Tetrachlorodibenzo-p-dioxin   No commercial use; tested as a pesticide         Lung, lymphoma, liver
Polychlorinated biphenyls             Flame retardants, hydraulic fluids                Liver, skin
Arsenic                               Natural ores, alloys; pharmaceutical agent       Skin, lung, liver
Cadmium                               Natural ores; pigments, batteries, ceramics      Lung, prostate, kidney
Nickel                                Natural ores; alloys, electrodes, catalysts      Lung, nasal cavity
Aflatoxin B1                           A mycotoxin (found in contaminated food)         Liver
Asbestos (fibrous silicates)           Thermal insulation, gaskets (declining usage)    Lung, mesothelioma
N-Nitrosodimethylamine                Polymers, batteries, nematocide (no longer       Liver, lung, kidney
                                        in use)
4-(Methylnitrosamino)-1-              Cigarette smoke; research tool                   Lung, liver
Ethylene oxide                        Glycol and polyester production; sterilization   Leukemia, lymphoma
Vinyl chloride (VC)                   Plastics (PVC), copolymers                       Liver (angiosarcoma)
Trichloroethylene                     Degreasing operations, adhesives, lubricants     Liver, kidney
1,2-Dichloroethane                    VC production, solvent, degreaser                Liver, lung, breast
                                        (no longer in use)
Bis(chloromethyl)ether                Technical applications (rarely used)             Lung
Mustard gas (sulphur mustard)         Chemical warfare in World War I; research        Lung
Nitrogen mustard                      Limited application as antineoplastic agent      Lung, skin, lymphoma

these CYPs (Fig. 2–7). Some of these compounds
are further activated by subsequent steps; for
example, 2-acetylaminofluorene (AAF) is further          The metabolic interconversions of this compound
modified by a sulfotransferase to form the ulti-         were studied in detail by the Millers and col-
mate DNA-binding moiety.                                leagues.9,10 In 1960, it was shown that AAF is con-
   Somewhat surprisingly, glutathione-S-trans           verted to a more potent carcinogen, N-hydroxy-
ferase (GST), which had been thought to be              AAF, after the parent compound was fed to rats.
involved only in detoxifying carcinogens, has           Although both AAF and N-hydroxyl-AAF are
been shown to activate some industrial chemi-           carcinogenic in vivo, neither compound reacted
cals,7 so GST appears to have a dual role, de-          in vitro with nucleic acids or proteins, suggesting
pending on the chemical.                                that the ultimate carcinogen was another, as-yet
CAUSES OF CANCER                                                                                      23

               Figure 2–5. Structures of some known carcinogens. (Used with permission.)

unidentified metabolite. Subsequent studies            and may be the ultimate carcinogen in nonhepatic
showed that N-hydroxy-AAF is converted in             tissues.
rat liver to a sulfate, N-sulfonoxy-AAF, by means
of a cytosol sulfotransferase activity (Fig. 2–7).
                                                           Other Aromatic Amines
This compound reacts with nucleic acids and
proteins and appears to be the ultimate carcino-      Electrophilic forms of the aromatic amines result
gen in vivo. It is also highly mutagenic, as deter-   from their metabolic activation, and the positively
mined by assays of DNA-transforming activity          charged nitrenium ion formed from naphthyl-
(see below).                                          amine and aminobiphenyl compounds has been
   Other enzymatic conversions of AAF occur in        implicated as the ultimate urinary bladder car-
rat liver, for example, N-hydroxy-AAF is converted    cinogen in dogs and humans. Hydroxylamine de-
to N-acetoxy-AAF, N-acetoxy-2-aminofluorene            rivatives of these compounds are formed in
and the O-glucuronide (conjugate with glucoronic      the liver and then converted to a glucuronide. The
acid). These enzymatic reactions may also be in-      glucuronide conjugate is excreted in the urine,
volved in the conversion of AAF to carcinogenic       where the acid pH can convert it back to hydrox-
metabolites,especiallyinnonhepatictissues,which       ylamine and subsequently to a protonated hy-
often have low sulfotransferase activity for N-       droxylamine, which rearranges to form a nitre-
hydroxy-AAF. The acetyltransferase-mediated           nium ion by a loss of water. The electrophilic
activity converts N-hydroxy-AAF to N-acetoxy-2-       nitrenium ion can then react with nucleophilic
aminofluorene, which is also a strong electrophile     targets in the urinary bladder epithelium.
24                                                                                      CANCER BIOLOGY

                                                         cial one. Moreover, because carcinogenesis is a
                                                         multistage process involving initiation, a lag time,
                                                         promotion, and tumor progression, multiple
                                                         actions of a carcinogen—or alternatively, the
                                                         actions of multiple carcinogens—appear to be
                                                         necessary to produce a clinically detectable
                                                         malignant neoplasm. An important point to
                                                         note, however, is that although the PAH diol-
                                                         epoxides vary considerably in their biological
                                                         reactivity, the level of mutations in cells is quan-
                                                         titatively related to the level of diol-epoxide-
                                                         DNA adducts, and the carcinogenicity of dif-
                                                         ferent PAHs correlates with the DNA adducts in
                                                         lung tissue.7
                                                             Of particular interest is the association of
Figure 2–6. The enzymatic and nonenzymatic acti-         CYP1A1 levels with cigarette smoking. CYP1A1
vations of dimethylnitrosamine and N-methyl-N-           is inducible in various extrahepatic tissues by the
nitroso reactive nucleophiles. (Used with permission.)   PAH contained in cigarette smoke. This has led
                                                         a number of investigators to examine the rela-
                                                         tionship between inducibility of CYP1A1 and
     Polycyclic Aromatic Hydrocarbons
                                                         susceptibility to lung cancer. Early studies of
In 1950, Boyland11 suggested that the carcinoge-         Kellerman et al.13 showed a correlation between
nicity of polycyclic aromatic hydrocarbons (PAH)         aryl hydrocarbon hydroxylase induction in pe-
was mediated through metabolically formed ep-            ripheral blood lymphocytes and the incidence of
oxides. It was originally thought that the key ex-       lung cancer. More recent studies have shown
poxide formation involved the K region of the hy-        the formation of DNA-benzo(a)pyrene adducts
drocarbon ring structure.12 However, subsequent          in pulmonary tissue from cigarette smokers and
studies demonstrated that K-region epoxides              a higher level of CYP1A1 expression in lung tis-
had little carcinogenecity in vivo. An extensive         sue from cigarette smokers than in nonsmokers
amount of work has gone on since the 1950s to            (89% vs. 0% respectively).14 In addition, CYP1A1
characterize the metabolism and carcinogenic             was elevated in about half the lung cancers from
potential of the PAH (reviewed in Reference 7).          smokers, compared to only 25% of lung cancers
It is now generally accepted that conversion of          from nonsmokers. A genetic polymorphism of
PAH to dihydrodiol epoxides is a crucial path-           CYP1A1 combined with a genetic dificiency in
way in the formation of the ultimate carcinogen.         GST (which detoxifies the electrophilic metab-
For instance, studies from a number of labora-           olites of PAH) is associated with an increased
tories have indicated that 7b,8a-dihydroxy-9a,           risk of cigarette smoking–induced lung cancer.15
10aepoxy-7,8,9,10-tetrahydrobenzo(a)pyrene is                Cigarette smoke contains other toxic chemi-
an ultimate mutagenic and carcinogenic metab-            cals in addition to PAH. One of the most deadly
olite of benzo(a)pyrene. The evidence is strong          is a carcinogenic nitrosamine, nitrosamine 4-
that the analogous metabolites of other PAHs that        (methylnitrosamino)-1-(3-pyridyl)-1-butanone
are similar to benzo(a)pyrene are also the ultimate      (NNK), which is metabolized in several steps
carcinogens of these compounds.                          by a cytochrome P-450, cyclooxygenase, or li-
   It should be noted that although a number of          poxygenase to produce metabolites that bind
interactions of chemical carcinogens with DNA            DNA.16 One of the DNA adducts formed is an
and other cellular macromolecules have been              06-methylguanine that causes a GC-AT transi-
observed, there is still no formal proof that the        tional base mispairing that has been associated
major reaction products detected in cells ex-            with an activating point mutation in the K-ras
posed to these agents are the ones actually in-          oncogene. This mutation has been observed in
volved in carcinogenesis. It could be that some          NNK-induced pulmonary adenocarcinomas in
minor or as-yet undetected reaction is the cru-          mice and rats.17
AFB1                         O                                                               O                                               endo
                                    O                                                                      O                                                     O       O
                         O                                                O
                     H                                                                  O                                                    O
                                                  CYP3A4                            H                                                                        O
                                                                                                                  AFB1 exo-8,9-oxide
            O                                                             O
            H    O               OCH3                                     H    O                  OCH3                                               O               OCH 3

AAF                                                                                               OH                                                      OSO3
                         NH             CYP1A2                                                   N                SULT                                   N
                         O                                                                   O                                                           O
                                                                                                                                                         N -Sulphoxy-AAF
                                  CYP1A1, 1B1                                                                         CYP1A1,                    O
                                  mEH                                                                                 1B1, 3A4

                                                                     HO                                                                     HO
                                                                              OH R,R-dihydrodiol                                                     OH ( )-Anti -BPDE

            Cl                                                                                                 -Glu
                         GST                                  -Glu
                                                                                                   S        Cys        GSH episulphonium ion
      Cl                                     Cl         S    Cys
TCE                                                                           -GT
            Cl                                     Cl                                                  Cl
                         GST                                                  DP                                        -Lyase
                                                                                                                                             C S
       Cl   Cl                          Cl         S    Cys                                 Cl         S       Cys                     Cl
                                                                                                                                                 Chloro thioketene

                     Figure 2–7. Enzymatic conversion of some selected human carcinogens to-
                     ward their ultimate DNA-reactive metobolites. Activation of aflatoxin B1
                     (AFB1), 2-acetylaminofluorene (AAF), and benzo[a]pyrene (BP) requires the
                     activity of cytochrome P450–dependent monooxygenases (CYPs). CYP3A4
                     activates AFB, at its 8,9-bond, resulting in the AFB, exo-8,9-oxide. The endo-
                     diasteromer is not formed by CYP3A4, but might be formed in small amounts
                     by CYP1A2. AAF is converted by CYP1A2 into N-hydroxy-AAF, which
                     subsequently might undergo sulphotransferase (SULT)-catalysed esterifica-
                     tion into the ultimate genotoxic form, the N-sulphoxy-AAF, BP is initially
                     converted mainly by CYP1A1 or CYP1B1 into the 7,8-epoxide. This epoxide
                     is a substrate of microsomal epoxide hydrolase (mEH), which produces the
                     7,8-dihydrodiol. Both reactions together stereoselectively form the R,R-
                     dihydrodiol. Further epoxidation at the vicinal double bond catalyzed by
                     CYP1A1, CYP1B1, and CYP3A4 generates the ultimate genotoxic diol-
                     epoxide of BP (BPDE). Of the four possible resulting diastereomers, the
                     (þ)-anti-BPDE is formed at the highest levels. 1,2-Dichloroethane (DCE) is
                     activated by glutathione-S-transferases (GSTs) into glutathione (GSH) half-
                     mustard and GSH episulphonium electrophiles, which can bind directly to
                     DNA. GST-catalyzed conjugation of trichloroethylene (TCE) produces GSH
                     adducts. Cleavage of the terminal amino acids by g-glutamyltransferase (g-
                     GT) and cysteinylglycine dipeptidase (DP) activity give rise to cysteine (Cys)
                     adducts that can be converted into genotoxic thioketenes by the kidney-
                     specific cysteine conjugate b-lyase. The red arrows point to the position of the
                     nucleophile (DNA, protein, GSH) attack. GSH conjugates of AFB, oxide, or
                     PAH diol-epoxides are detoxification products. (From Luch,7 reprinted by
                     permission from Macmillan Publishers Ltd.)
26                                                                                    CANCER BIOLOGY

   Another important enzyme in the carcinogen             The potential biological consequences of
activation pathway is the microsomal epoxide           DNA base–adduct formation by chemical car-
hydrolase that catalyzes the stereoselective hy-       cinogens are several. In some cases, it may stabi-
dration of alkene and arene oxides to trans-           lize an intercalation reaction in which the flat
dihydrodiols. This enzyme is inducible by various      planar rings of a polycyclic hydrocarbon are in-
xenobiotics, including some chemical carcino-          serted between the stacked bases of double-
gens and phenobarbital, and its level is increased     helical DNA and distort the helix, leading to a
in hepatic nodules and hepatomas induced by            frame-shift mutation during DNA replication
chemical carcinogens.18                                past the point of the intercalation.19 Alkylated
                                                       bases in DNA can mispair with the wrong base
                                                       during DNA replication—for example, O6
DNA Adduct Formation
                                                       methylguanine pairs with thymine instead of
Since most chemical carcinogens react with             cytosine during DNA replication, leading to a
DNA and are mutagenic, interactions with DNA           base transition (i.e., GC?AT) type of mutation
have been viewed as the most important reac-           during the next round of DNA replication.20
tions of these agents with cellular macromole-         Many of the base adducts formed by carcino-
cules. Reaction of chemical carcinogens with           gens involve modifications of N-3 or N-7 posi-
DNA is the simplest mechanism that explains            tions on purines that induce an instability in the
the induction of a heritable change in a cell          glucosidic bond between the purine base and
leading to malignant transformation; thus many         deoxyribose, resulting in loss of the base and
investigators view this as the most plausible          creation of an apurinic site in DNA.21 This ‘‘open’’
mechanism for initiation of carcinogenesis. Rep-       apurinic site can then be filled by any base, but
resentative agents from virtually all classes of       most commonly by adenine, during subsequent
chemical carcinogens have been shown to affect         DNA replication. This substitution can result in a
DNA in some way, and a number of distinct              base transition (purine–pyrimidine base change,
biochemical-reaction products have been iden-          but in the same orientation, e.g., GC?AT) or a
tified after treatment of cells in vivo or in culture   base transversion (inverted purine–pyrimidine
with carcinogenic agents.                              orientation, e.g., GC?TA). Finally, interaction
   The principal reaction products of the nitro-       with some carcinogens has been shown to favor
samines and similar alkylating agents with DNA         a conformational transition of DNA from its
are N-7 and O6 guanine derivatives. However, the       usual double-helical B form to a Z-DNA form.
extent of O6 alkylation of DNA guanine residues        This could alter the transcribability of certain
correlates better with mutagenic and carcino-          genes, since B?Z conformational transitions
genic activity than the quantitatively greater         are thought to be involved in regulating chro-
N-7 alkylation of guanine residues (see below).        matin structure.
Reactions also occur with other DNA bases, and            Another interesting point is that interaction of
these may be important in subsequent muta-             chemical carcinogens with DNA or chromatin
genic or carcinogenic events. Aflatoxin forms           does not appear to be a random process. For
adducts of guanine at the N-7 position after           example, when the ultimate carcinogen of ben-
metabolic activation. The principal reaction           zo[a]pyrene, that is, its diol epoxide metabolite,
product of AAF with cellular DNA is the C-8            is reacted with cloned chicken b-globin DNA, it
position ofguanine, justasitisforRNA. Othercar-        preferentially binds in a 300–base pair sequence
cinogenic aromatic amines, such as N-methyl-4-         immediately 50 to the RNA cap site.22 Since this
aminoazobenzene, also produce C-8 substituted          region is thought to contain sequences involved
guanine residues as their major nucleic acid           in regulating gene transcription, its alteration by
reaction product (adduct). Polycyclic aromatic         a chemical carcinogen could change the func-
hydrocarbons, after activation, also react with        tion of genes downstream from the regulatory
DNA and RNA, forming adducts involving the             sequences. Moreover, treatment of the large
2-amino group of guanine, but other reaction           polythene chromosomes of Chironomus with the
products derived from guanine, adenine, and            ultimate carcinogen benzo(a)pyrene diol epox-
cytosine have been observed as well.7                  ide in vitro or administration of the parent
CAUSES OF CANCER                                                                                     27

unmetabolized compound in vivo to Chir-              are the most highly conserved domains of these
onomus larvae demonstrates that the carcinogen       exons. In human colon tumors, the majority of
binds preferentially to areas most active in gene    the mutations are G?A transitions (just as for
transcription.23 DNA in transcribing regions         ras); however, other types of base alterations of
associated with the nuclear matrix also appears      p53 are seen in other human cancers.
to be a preferential target for carcinogen bind-        Perhaps the most interesting observation is
ing. Taken together, these data indicate that the    the finding of a high incidence of p53 point
specificity of carcinogen binding is determined       mutations in hepatocellular carcinomas in pa-
to some extent by the base sequence of DNA, its      tients from parts of China and southern Africa
location within the nucleus (e.g., association       where exposure to aflatoxin B1 is endemic.25,26
with nuclear matrix), and the structure of chro-     Most of these are at a single site, the third
matin, with active, ‘‘open’’ sites being favored.    base of codon 249, and are G?T transversions.
                                                     Moreover, hepatocellular carcinomas from low-
                                                     aflatoxin exposure areas appear to only rarely
Interaction of Chemical Carcinogens
                                                     have this mutation.
with Oncogenes and Tumor
Suppressor Genes
Cellular oncogenes and tumor suppressor genes
                                                     Epigenetic Changes
are two of the critical DNA targets for chemical
carcinogens, leading to activation of oncogenes      Even though the application of Ockham’s (or
and the inactivation of suppressor genes. This       Occam’s) razor to the effects of chemical car-
will be discussed further in Chapter 5, but a few    cinogens leads to the concept that the genotoxic
examples will be given here.                         results of carcinogen-DNA binding are the sim-
   Carcinogens can activate cellular oncogenes       plest, most straightforward explanation for their
(proto-oncogenes) by a variety of mechanisms         carcinogenicity, a number of important epigenetic
including base substitution (point) mutations,       effects are also observed. For example, changes in
chromosomal translocations, and gene amplifi-         gene expression patterns caused by carcinogen-
cation. One fairly common example is the acti-       induced epigenetic alterations such as changes in
vation of ras proto-oncogenes by chemical and        DNA methylation or histone acetylation have
physical carcinogens in both cultured mamma-         been observed after exposure of cells to carcin-
lian cells and animal models (reviewed in Ref.       ogens. This pattern has been observed, for ex-
24). H-ras and K-ras proto-oncogene mutations,       ample, during cells’ exposure to the carcinogenic
for example, have been observed in rodent mod-       metals nickel, cadmium, or arsenic.7 The carci-
els of skin, liver, lung, and mammary carcino-       nogenic effects of nickel have been linked to
genesis. The observed mutations in the tumors        DNA hypermethylation and histone deacetyla-
correlate with expected base adducts formed by       tion, both of which can alter chromatin structure
the carcinogen: G?A base transitions with al-        and cause epigenetic silencing of tumor suppres-
kylating agents (e.g., NMU and MNNG), G?T            sor genes (see Chapter 5).
transversions for benzo(a)pyrene, A?T trans-
versions for 7,12 dimethylbenzanthracene,
                                                     Tumor Initiation, Promotion,
G?T transversions and G?A transitions for
                                                     and Progression
aflatoxin B1.24 These mutations appear to reflect
similar base substitution mutations in human         The idea that development of cancer is a mul-
tumors.                                              tistage process arose from early studies of virus-
   The best documented example of a tumor            induced tumors and from the discovery of the
suppressor gene being inactivated during car-        cocarcinogenic effects of croton oil. Rous and
cinogenesis is the p53 gene. Mutations of the        colleagues found that certain virus-induced skin
p53 gene have been observed in animal tumors         papillomas in rabbits regressed after a period of
and in a wide variety of human cancers. Most of      time and that papillomas could be made to
the mutations are point mutations involving ‘‘hot    reappear if the skin was stressed by punching
spots’’ in exons 5 through 8. Interestingly, these   holes in it or by applying such irritant substances
28                                                                                    CANCER BIOLOGY

as turpentine or chloroform. These findings led        cellular damage inflicted by initiation to be
Rous and his associates to conclude that tumor        propagated, and the initiated cells are clonally
cells could exist in a latent or dormant state and    expanded. This propagation of damaged cells
that the tumor induction process and subse-           in which genetic alterations have been produced
quent growth of the tumor involved different          leads to the production of more genetic alter-
mechanisms, which they called ‘‘initiation and        ations. This genetic instability is the hallmark
promotion.’’27 The term cocarcinogen was coined       of the progression phase of carcinogenesis and
by Shear, who discovered that a basic fraction of     leads to the chromosomal translocations and
creosote oil enhanced the production of mouse         aneuploidy that are frequently seen in cancer
skin tumors by benzo(a)pyrene.28 In 1941, Be-         cells.34 Such alterations in the genome of the
renblum29 reported that among mice receiving a        neoplastic cell during the progression phase
single skin painting of a carcinogen, such as         lead to the increased growth rate, invasiveness,
methylcholanthrene, only a small number of            and metastatic capability of advanced neo-
animals developed papillomas, but if the same         plasms. Some of the gene expression alterations
area of skin was later painted repeatedly with        that occur during tumor initiation and promo-
croton oil, which by itself is not carcinogenic,      tion are shown in Figure 2–9 (see color insert).
almost all the animals developed skin carcino-           Evidence for multistage induction of malig-
mas. Taken together, the data of these investi-       nant tumors has also been observed for mam-
gators suggested a multistage mechanism for           mary gland, thyroid, lung, and urinary bladder
carcinogenesis.                                       and in cell culture systems (reviewed in Refer-
   Studies of the events involved in the initiation   ence 9), thus it seems to be a general phenom-
and promotion phases of carcinogenesis were           enon. This experimental evidence is consistent
greatly aided by the identification of agents that     with the observed clinical history of tumor de-
have primarily an initiating activity, such as ure-   velopment in humans after exposure to known
thane or a low dose of a ‘‘complete’’ carcinogen      carcinogens—that is, initial exposure to a known
(see below), and by the purification of the com-       chemical or physical carcinogen, a long lag pe-
ponents of croton oil that have only a promoting      riod during which exposure to promoting agents
activity. Diesters of the diterpene alcohol phor-     probably occurs, and finally the appearance of
bol were isolated from croton oil and found to be     a malignant tumor.
the tumor-promoting substances.30,31 Of these,           Several characteristics of tumor initiation,
12-O-tetradecanoylphorbol-13-acetate (TPA) is         promotion, and progression provide some in-
the most potent promoter.32                           sight into the mechanisms involved in these pro-
   A scheme used to study the initiation–             cesses. Initiation can occur after a single, brief
promotion phases of mouse skin carcinogenesis         exposure to a potent initiating agent. The actual
is depicted in Figure 2–8. Typically, tumor ini-      initiation events leading to transformation into a
tiation is brought about by the single application    dormant tumor cell appear to occur within one
of an initiator, such as urethane, or a subcarci-     mitotic cycle, or about 1 day for the mouse skin
nogenic dose of an agent with both initiating and     system.32 Furthermore, initiation appears to be
promoting activity, such as the polycyclic hydro-     irreversible; the promoting agent can be given
carbon benzo(a)pyrene; promotion is carried out       for up to a year later and a high percentage of
by repeated application of a phorbol ester, such      tumors will still be obtained. Thus, the initiation
as TPA (e.g., three times a week).31,33 Benign        phase only requires a small amount of time, it is ir-
papillomas begin to appear at 12 to 20 weeks          reversible, and it must be heritable because the
and by about 1 year, 40% to 60% of the animals        initiated cell conveys the malignant alteration to
develop squamous cell carcinomas. If the pro-         its daughter cells. All these properties are con-
moting agent is given alone, or before the initi-     sistent with the idea that the initiation event
ating agent, usually no malignant tumors occur.       involves a genetic mutation, although other ‘‘epi-
   The progression stage of carcinogenesis is an      genetic’’ explanations are possible (see above).
extension of the tumor promotion stage and            The promotion phase, by contrast, is a slow,
results from it in the sense that the cell prolif-    gradual process and requires a more prolonged
eration caused by promoting agents allows the         exposure to the promoting agent. Promotion
CAUSES OF CANCER                                                                                       29

               Figure 2–8. Scheme of initiation–promotion phases of induction of carci-
               nogenesis in mouse skin. Initiation is caused by the single application of a
               subcarcinogenic dose of an agent such as 7,12-dimethylbenz[a]anthracene,
               benzo[a]pyrene, or urethane. Promotion is carried out by repeated applica-
               tion (e.g., k three times a week) of an agent such as the phorbol ester TPA.
               Papillomas develop within 12 to 20 weeks, squamous carcinomas in about 1
               year. Solid lines indicate continual application of agent; dotted lines indicate
               the duration of time without exposure to agents. Note that promoter may be
               added up to 1 year after a single application of the initiating agent and tumors
               still occur. I, initiator; P, promoter. (Used with permission.)

occupies the greater part of the latent period of          The later events in the tumor progression
carcinogenesis, is at least partially reversible,       phase are also thought to be irreversible because
and can be arrested by certain anticarcinogenic         of the pronounced changes in the genome that
agents (see Chapter 9). Tumor promotion is a            have occurred leading into this phase. Agents that
cell proliferation phase that propagates the ini-       are ‘‘pure’’ progression-causing agents are hard
tiated damage and leads to the emergence of an          to identify, but the free radical–generating agent
altered clone of cells. Most promoting agents           benzoylperoxide appears to be a progression-
are mitogens for the tissue in which promotion          inducing agent during experimental epidermal
occurs. Tumor progression requires continued            carcinogenesis.35
clonal proliferation of altered cells, during              It should be noted that some potent carcin-
which a loss of growth control and an escape            ogens are ‘‘complete carcinogens’’ in that at cer-
from host defense mechanisms become                     tain doses they can by themselves induce a
predominant phenotypic traits. This process al-         cancer. Such agents include polycyclic aromatic
lows growth to progress to a clinically detectable      hydrocarbons, nitrosamines, certain aromatic
tumor.                                                  amines, and aflatoxin B1. When these agents are
Figure 2–9. Tumor promotion and tumor initiation. Genotoxic carcinogens
can induce damage in tumor suppressors or oncogenes in different ways, all
of which contribute to the transformation of normal cells into tumor cells—
this is known as the tumor initiation stage in carcinogenesis. Some chemical
carcinogens are also capable of promoting the outgrowth of those trans-
formed cell clones and of contributing to the generation of visible tumor cell
masses—this is known as the tumor promotion stage in carcinogenesis. a.
Chemical compounds such as 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD)
or benzo[a]pyrene (BP) result in tumor promotion through arylhydrocarbon
receptor (AhR)–mediated signal transduction. Binding of TCDD or BP to
AhR leads to activation and translocation of the complex into the nucleus.
After heterodimerization with the AhR nuclear translocator (ARNT), the
complex binds to xenobiotic-responsive elements (XREs) and induces the
expression of a variety of different genes involved in carcinogen metabolism,
including CYP forms 1A1, 1B1, and 1A2. It also changes the expression
pattern of several factors involved in cellular growth and differentiation, such
as plasminogen-activator inhibitor type 1 (PAI1), metallothionein II (MT-II),
human enhancer of filamentation 1 (HEF1), guanine nucleotide exchange
factor (GEF), COT, and KRAS. Pro-apoptosis factors such as tumor necrosis
factor (TNF; superfamilies 3, 6, 8, 9, 10) and heat-shock protein 40 (HSP40)
are down-regulated, and cell cycle genes can either be up-regulated (such as
cyclin B2) or down-regulated (such as NEK2). The factors shown here are
only a few examples of the great number of factors that have been shown by
gene expression analysis to be altered following carcinogen exposure. COT
and NEK2 are serine/threonine kinases.70,71 b. Tumor initiation occurs
through DNA adduct–derived mutations in cancer susceptibility genes. DNA
binding by genotoxic carcinogens such as activated BP leads to the induction
of base pair or frameshift mutations in cancer susceptibility genes such as
TP53 or RAS. The mutagenic potency of such polycyclic aromatic hydro-
carbon diol-epoxide-DNA adducts can be increased because of inhibition
of nucleotide excision repair (NER) by metal ions (Meþ; for example, Ni2þ),
or as a result of NER factor immobilization at repair-resistant DNA-
adduct sites, also known as decoy adducts. (From Luch,7 reprinted by per-
mission from Macmillan Publishers Ltd.)
CAUSES OF CANCER                                                                                    31

given in sufficient dose to animals during cancer-       chemicals. Using this system, together with
causing protocols, they can cause DNA damage            a liver microsomal fraction that has the
and produce tissue necrosis, which is itself            capacity to activate most chemical carcin-
enough to stimulate several rounds of cell pro-         ogens metabolically, Ames and colleagues
liferation in response to the tissue damage. In         have shown that about 90% of all carcino-
this situation, the promotion–progression pha-          gens tested are also mutagenic.36 More-
ses are often collapsed in time, resulting in           over, few noncarcinogens show significant
the production of aneuploid malignant cells.34          mutagenicity in this test system. Malignant
                                                        transformation can be induced in a variety
                                                        of cultured mammalian cells by agents that
     Mechanisms of Tumor Initiation                     are mutagenic for the same cells. For ex-
                                                        ample, carcinogenic polycyclic hydrocar-
Initiation of malignant transformation of normal
                                                        bons cause mutations, as measured by
cells by a carcinogenic agent involves a perma-
                                                        induction of resistance to 8-azaguanine,
nent, heritable change in the gene expression of
                                                        ouabain, or elevated temperature, in Chi-
the transformed cell. This could come about by
                                                        nese hamster V79 cells if the cells are
either direct genotoxic or mutational events, in
                                                        cocultured with lethally irradiated rodent
which a carcinogenic agent reacts directly with
                                                        cells that can metabolize the hydrocarbons
DNA, or by indirect or ‘‘epigenetic’’ events that
                                                        to their electrophilic, active metabolite.37,38
modulate gene expression without directly re-
                                                        In these studies, mutagenicity was obtained
acting with the base sequence of DNA. Most
                                                        with the carcinogenic hydrocarbons 7,12-
investigators favor the mutational theory of
                                                        dimethylbenz(a)anthracene, benzo(a)pyr-
carcinogenesis—that is, that the initiating events
                                                        ene, and 3-methylcholanthrene. There was
involve a direct action on the genome.
                                                        no mutagenicity with a noncarcinogenic
   The mutational theory depends on three kinds
                                                        hydrocarbon, and the degree of mutage-
of evidence:
                                                        nicity was related to the degree of carci-
   1. Agents that damage DNA are frequently             nogenicity of the chemicals in vivo.
      carcinogenic. As discussed previously,         3. Incidence of cancer in patients with DNA-
      chemical carcinogens are usually activated        repair deficiencies is increased. In indi-
      to form electrophilic agents that form spe-       viduals with certain recessively inherited
      cific reaction products with DNA. The ex-          disorders, the prevalence of cancer is sig-
      tent of formation of some of these reaction       nificantly higher than in the general pop-
      products, for example, alkyl-O6-guanine,          ulation.39 The connecting link between
      has been shown to correlate with muta-            these disorders is the inability to repair cer-
      genicity and carcinogenicity of certain           tain kinds of physical or chemical damage
      chemical agents. Ultraviolet and ionizing         to DNA. The high incidence of cancer in
      radiation also interact with DNA at doses         these diseases constitutes the best avail-
      that are carcinogenic.                            able evidence for a casual relationship
   2. Most carcinogenic agents are mutagens.            between mutagenicity and carcinogenicity
      A number of in vitro test systems using           in humans.
      mutational events in microorganisms have             One example of xeroderma pigmento-
      been developed to rapidly screen the muta-        sum (XP) is characterized by extreme
      genic potential of various chemical agents.       sensitivity of the skin to sunlight and is the
      One of the best known of these, the Ames          most widely studied of the repair-deficient
      test, is based on certain characteristics of      human diseases. Virtually 100% of affected
      specially developed strains of the bacterium      individuals will eventually develop some
      Salmonella typhimurium. The tester strain,        form of skin cancer. In addition, hetero-
      a mutant line that requires exogenous histi-      zygotes who carry the XP gene but do not
      dine for its growth (hisauxotroph), has a         have the disease appear to have a higher
      poor excision repair mechanism and an in-         incidence of nonmelanoma skin cancer.40
      creased permeability to exogenously added         All individuals with XP are defective in
32                                                                                  CANCER BIOLOGY

      repair of ultraviolet damage to DNA, and        has been shown to produce renal adenocarcino-
      most of them have a defect in the excision      mas in male hamsters. The estrogen-dependent
      repair pathway (see below). The repair          induction of these tumors is inhibited by the
      defect ranges from 50% to 90% repair            simultaneous administration of testosterone or
      efficiency in cells from different patients,     progesterone.42 Malignant neoplasms arise when
      and there is good correlation between the       rat ovarian tissue is transplanted into normal rat
      severity of the molecular defect and the        spleen, presumably because of the hormonal
      extent of the disease. The defect in most       imbalance thus induced.43 Tumors can also be
      patients appears to be at the nicking or        induced in rats by the insertion of plastic or
      incision step of excision repair, although      metal films, depending only on the physical state
      patients in one complementation group           (i.e., solid versus porous or fibrous form) of the
      have normal excision repair and are de-         inserted material. In any of these cases, it is
      fective in postreplication repair.39 The        unlikely that mutational events resulting from
      XP cells are also less efficient at repair-      interaction with nuclear DNA could have pro-
      ing chemically induced damage to their          duced the tumors, although it has been shown
      DNA.                                            that an oxidative metabolite of DES can induce
         Other examples of enhanced suscepti-         SCE in cultured hamster embryo cells.44 SCE
      bility to cancer in individuals with DNA        was also observed in human hepatoma cells that
      repair deficiencies are ataxia telangiectasia    metabolize DES but not in cell lines that do
      (AT), Fanconi’s anemia (FA), and Bloom’s        not.45 Presumably, this phenomenon occurs
      syndrome (BS). Cell lines derived from AT       as the result of the formation of phenoxy radi-
      patients are defective in repair replication    cal intermediates from DES metabolites by a
      following exposure to irradiation or carci-     peroxidase-mediated reaction and the DNA
      nogenic chemicals.40 Patients with AT are       strand breakage that is produced by these oxy-
      more prone to develop leukemia and some         gen radicals.
      other cancers. FA cells have a defect in the       In cases in which direct interaction with DNA
      repair of cross-linked bases in DNA; they       does not seem to occur, it is likely that regula-
      also appear to have a slight deficiency in       tion of DNA expression is changed by epigenetic
      the repair of g-ray- or ultraviolet-induced     alterations that result in malignant transforma-
      damage, and patients with FA are at in-         tion stemming from chromatin conformational
      creased risk to develop cancer.39 Lympho-       changes. These changes can be induced by DNA
      blastoid cell lines from patients with a type   methylation or histone modifications as noted
      of BS characterized by a high rate of sister    above or by other mechanisms, such as (1) in-
      chromatid exchange (SCE) are more highly        teraction of carcinogenic agents with membrane
      tumorigenic after brief exposure in vitro to    proteins that regulate cell surface receptors for
      4-nitroguinoline-N-oxide and N-methyl-N0 -      growth factors or that control feedback regula-
      nitro-N-nitrosoguanidine (MNNG) in the          tion of cell proliferation in the cell’s microen-
      nude mouse assay than are lymphoblastoid        vironment; (2) reaction with RNA molecules
      cells from normal individuals.41                involved in translation of proteins; (3) binding to
                                                      regulatory proteins that control gene transcrip-
   Although the mutational theory is the simplest     tion; or (4) interaction with proteins involved in
explanation of a heritable change in a cell that      the cell’s mitotic apparatus. These changes
could produce a cancer-initiating event, not all      could be ‘‘heritable,’’ at least for several cell
initiating agents are mutagenic in the test sys-      generations, because each daughter cell re-
tems used, and some do not react directly with        ceives a certain complement of the parent cell’s
DNA. For example, malignant tumors can be             RNA and protein. In any case, the ultimate re-
produced by agents that do not interact with          sult is a cell that is genetically unstable, since
DNA or by certain cells placed in abnormal            tumor progression leads to the appearance of
tissue locations. Prolonged administration of the     cells that are genetically different from cells of
estrogenic compound diethylstilbestrol (DES)          the tumor’s tissue of origin.
CAUSES OF CANCER                                                                                         33

                                                            If at least two mutations are required (e.g., as
     Endogenous Carcinogenesis
                                                         the Knudson model predicts), the required mu-
An important question that arises is, what is the        tation rate would be the square of 1.4 Â 10À10.
source of mutations in the human genome that             In this case, the expected number of spontane-
leads to cancer? One might argue that the answer         ously arising cancer cells in an individual’s life-
is obvious. We live in a sea of carcinogens: PAHs        time would be about 300,000. Even this number
from automobile exhaust, industrial pollution,           seems high. And even if it were true, most of
pesticide residues in foods, chlorinated organic         these cells would die or be eliminated by im-
compounds in drinking water, etc. Furthermore,           mune mechanisms. If more than two mutations
epidemiologists argue that almost 30% of human           are required to produce a cancer, then the num-
cancers are related to cigarette smoking. Yet, a         ber becomes much smaller, e.g., 5.5 cancers per
significant amount of cancers occur in people             100,000 individuals for three mutations. Thus, if
with no clear evidence of exposure to clearly            this latter assumption is correct, i.e., that three
defined carcinogens. For a number of cancers of           or more mutations are required to produce a
the pancreas, ovary, kidney, and breast, for ex-         cancer, then the spontaneous mutation rate
ample, there are in most cases no clear geo-             could not by itself explain the number of cancers
graphic or genetic risk factors (although heritable      in the human population. Moreover, the type
genetic changes may account for 5% to 10% of             and distribution of spontaneous mutations differ
some cancers such as breast cancer). Thus, if            from that of chemically induced mutations in
cancer is initiated through a mutation or a series       cells and from that of a number of mutations
of mutations, how might these arise?                     found in human cancers.47 Thus, although
   One possibility is that ‘‘spontaneous’’ muta-         spontaneous mutations may contribute to the
tions arising from an inherent error rate in the         causation of human cancer, they are unlikely by
fidelity of DNA replication and /or repair could          themselves to cause the initiation and progres-
give rise to mutations, some of which by chance          sion events that lead to most invasive, metastatic
could be in key genes involved in regulation of          neoplasms.
cell proliferation and differentiation. The sponta-         Several potential mechanisms exist for sponta-
neous or background mutation rate in human               neous mutations in human cells. These include
somatic and germline cells has been estimated to         depurination, deamination, damage to DNA by
be 1.2 to 1.4 Â 10À10 mutations per base pair per        oxygen radicals, and errors in DNA replication.
cell division6 (i.e., one mistake in 10 billion base     Depurination is the most common potentially
pairs per each cell division). Since there are about     mutagenic event, occurring at a rate of about
1014 cells in the adult human (with a genome of          10,000 depurination events per cell per day.48
3 Â 109 base pairs) and they undergo an esti-            This results from breakage of the N-glycosidic
mated 1016 cell division cycles in a normal life         bond connecting a purine base to the deoxyribose-
span, about 3.6 Â 1015 single base (point) muta-         phosphate backbone of DNA and creates a gap in
tions could arise in a lifetime.6 If a single mutation   the base sequence. When DNA polymerase en-
could produce a cancer, this would lead to               counters such a gap during DNA replication, it
3.6 Â 1015 spontaneously arising cancer cells in a       may insert the wrong base, usually an adenine, in
lifetime, a highly unlikely proposition. Several         place of the missing base. Obviously, this couldn’t
considerations moderate this wildly excessive            happen very often or the mutation rate in the
number. For example, most base changes are               human genome would be higher than it is.
repaired; not all base changes are in coding re-            Deamination of cytidine to uridine occurs at
gions and some are silent (not producing an al-          1/500 the rate of depurination, or about 20
tered protein); not all mutations produce a cancer       events per cell per day. Since uridine base pairs
cell (they may not occur in key oncogenes or tu-         with adenine during DNA replication, this could
mor suppressor genes); and more than one mu-             lead to a G?A transition. Also, deamination of
tation is necessary to produce a cancer cell (in one     methylcytosine can occur producing thymidine,
study, 25% to 50% of human colon cancers, for            which if not repaired, could produce a G?A
example, contained nine or more mutations).46            transition.
34                                                                                    CANCER BIOLOGY

   The rate of damage to DNA produced by               rapid accumulation of errors that favor their
oxygen radicals, which are continuously gener-         survival. One way this could occur is through
ated in cells by normal metabolism, isn’t clear,       alterations in DNA polymerases involved in
but it could be as high or higher than depuri-         DNA replication and repair, although evidence
nation. Measurements in human urine of 8-              for such altered enzymes is not conclusive.
hydroxydeoxyguanosine and thymine glycol,                 One thing that is clear, however, is that can-
which are oxidative breakdown products of DNA,         cer is, in general, a disease of aging. As noted
suggest that 10,000 oxygen radical–induced al-         above, the average age at time of diagnosis of a
terations in DNA could occur per cell per day.49       malignant tumor is about 65. Moreover, the
However, cells have stringent mechanisms to            incidence of a number of adult solid cancers
protect themselves against free radicals generated     increases with the fourth to sixth power of age.52
by cellular metabolism. These include superox-         This finding, plus the fact that aneuploidy and
ide dismutase (SOD), catalase, and glutathione         other genetic alterations increase during tumor
generating systems. Oxygen radical damage, if un-      progression, which may occur over many years
repaired, can produce single-strand and double-        (up to 15 or 20 years for some cancers), supports
strand breaks in the DNA backbone.                     the notion that it is the accumulation of genetic
   Errors in DNA replication or repair could result    errors over time that is most dangerous for the
from any of the mechanisms described above. In         human genome.
addition, errors introduced by the DNA replicat-
ing machinery itself can occur. DNA polymerases,
                                                            Mechanisms of Tumor Promotion
though usually incrediblyaccurate, can make some
                                                            and Progression
mistakes. Using a f bacteriophage DNA rep-
lication system, Kunkel and Loeb50,51 have             Tumor-initiating agents most likely act by in-
determined the error rate of mammalian DNA             teracting with DNA to induce mutations, gene
polymerases. DNA pol a, the major DNA repli-           rearrangements, or gene amplification events
cating polymerase in eukaryotic cells, has an error    that produce a genotypically altered cell. What
rate of 1/30,000 to 1/200,000 bases, depending on      happens next is that the initiated cells undergo a
the method of purification; pol b, a major repair       clonal expansion under the influence of pro-
enzyme has a 1/5000 error rate; and pol d, a           moting agents that act as mitogens for the trans-
polymerase with proof reading ability, has an er-      formed cell type. As will be discussed later,
ror rate of 1/500,000. The most frequent error for     these promoting actions appear to be mediated
all polymerases are single base substitutions and      by cell membrane events, although a direct ac-
‘‘minus one’’ base frameshifts. These error rates,     tion of promoters on DNA has also been pro-
determined in vitro, probably overestimate the         posed. It is important to note that multiple
error rate in vivo, however, because purified en-       clones of cells are likely to be initiated by a DNA-
zymes were used, and repair enzymes that are           damaging agent in vivo and that, through a rare
part of the intracellular DNA replication complex      second event, one or a small number of these
may have been lost. Nevertheless, errors in DNA        clones progresses to malignant cancer.
replication are another potential source for spon-        It may be useful to think of the promotion
taneous mutations leading to cancer.                   phase as the stage of cell proliferation and clonal
   The facts that the actual cancer rate in the        expansion induced by mitogenic stimuli and of
population can’t be readily explained by the           the progression phase as the gradual evolution
background spontaneous mutation rate and that          of genotypically and phenotypically altered cells
genetic instability increases with tumor pro-          that occurs due to genetic instability of the
gression have led to the hypothesis that malig-        progressing cells. This process leads to the de-
nant cells have a way to increase their error rate     velopment of cell heterogeneity within a tumor,
and that they gain some selective advantage            an idea first described by Foulds53 and later
from this. The term mutator phenotype has              expanded by Nowell.54 During the progression
been used to describe this phenomenon,6 and            phase, which can take many years in humans,
refers to the ability of tumor cells to direct their   individual tumors develop heterogeneity with
own mutation rate or, more precisely, to allow a       respect to their invasive and metastatic charac-
CAUSES OF CANCER                                                                                        35

teristics, antigentic specificity, state of cellular    actually be the same genes involved in a selec-
differentiation, and responsiveness to hormones,       tive growth advantage for these cells. These cells
drugs, and immune-modulating agents. Presum-           may be lurking even in early-stage cancers. That
ably, some powerful selection process goes on to       is, some cancers are predestined almost from
favor the growth of one progressing cell type          the beginning to evolve into invasive, metastatic
over another. This preferential selection may be       tumors and some are not. This possibility has
due to a certain cell type developing a growth         huge implications for cancer screening, diag-
advantage in the host’s tissues over its peers, as     nosis, and choice of therapy. Numerous women
proposed by Nowell, or to the host’s immuno-           receive a diagnosis of ductal carcinoma in situ of
logic defense system being able to recognize and       the breast based on mammography screening,
destroy some cell types better than others, thus       and many men receive a diagnosis of prostate
providing the selection pressure for expansion         cancer based on a prostate-specific antigen
of one clone over another, or to a combination of      (PSA) test and subsequent biopsy. And yet many
these factors. Experimental evidence supports          of these patients have indolent tumors that
such a selection of tumor cells growing in vivo.       would not affect their overall life expectancy,
For example, Trainer and Wheelock55 have               and they still often undergo significant surgical
shown that during the growth of L5178Y lym-            and drug treatments. The problem is that we are
phoma cells in mice, a continual selection of          only beginning to be able to tell (e.g., by gene
cells with a decreasing ability to be killed by        expression arrays) which of these so-called
cytolytic T lymphocytes (CTL) ‘‘armed’’ against        early-stage cancers will be lethal and which ones
the tumor occurs, until an ‘‘emergent pheno-           won’t.
type’’ appears that is highly resistant to the CTL         Another point of the Weinberg theory is that
cells.                                                 the genetic alterations that occur during tumor
                                                       progression do not necessarily occur in a given
central dogma of tumor                                 sequence and are probably different for differ-
progression                                            ent cancers.56 One might even suggest that they
The standard concept of how cancer starts is           may be different in different patients who have
that malignant tumors arise from a single cell         the same histological tumor type. Ultimately,
transformed by a chemical carcinogen, onco-            however, these genetic and phenotypic changes
genic virus, radiation damage, endogenous ge-          lead to a similar loss of cell proliferation control
netic damage caused by oxidative insult to DNA,        and expression of a panoply of genes (maybe not
or any of a host of other potential ways (e.g.,        the identical ones) that make some tumors in-
chronic infections with a bacteria such as H.          vasive and metastatic.
pylori or with a parasite such as schistosomiasis,         There are clinical data supporting some of
or hormonal imbalance). Once the initiated cell        these concepts. In a study by van de Vijver
starts to undergo clonal expansion, it undergoes       et al.,58 it was determined that the gene ex-
multiple genetic changes, due to genetic insta-        pression profile of breast cancers was a much
bility, leading to an invasive metastatic cancer.      better predictor of disease outcome in patients
This progression is thought to occur sequen-           with breast cancer than standard clinical and
tially, as exemplified by the work of Vogelstein        histopathological staging. Indeed, they could re-
and colleagues on colon cancer.46 The idea here        stratify patients listed as low risk or high risk by
is that colon cancer goes through a series of ‘‘evo-   clinical staging into a more accurate prognostic
lutionary’’ changes from hyperplasia, to early-        outcome category (based on actual metastasis-
stage adenoma, to late-stage adenoma, to car-          free survival) through gene expression arrays. In
cinoma, and finally to metastatic cancer.               addition, Al-Hajj et al.59 were able to identify
    There is, however, another point of view           and isolate the more tumorigenic cells from a
proposed by Weinberg and colleagues.56,57 This         heterogeneous population of breast tumors in
hypothesis, for which there are supportive clini-      eight of nine patients. These more aggressive cell
cal data (see below), states that the genes involved   types were identified by their cell surface mark-
in driving invasiveness and metastasis may be          ers and by repeated passage in nude mice. Each
expressed early in the progression pathway and         time the more aggressive cells were injected
36                                                                                CANCER BIOLOGY

into nude mice they produced tumors, whereas        same receptor or, if not, through the same bio-
the marker-negative cells did not grow. These       chemical steps? The answer is not known, but
data suggest that the aggressive tumorigenic        the list of potential promoters in the human
cells can be prospectively identified in initial     environment is so large that it seems unlikely
tumor biopsies containing mixed populations of      that they would all act by means of the same
cells and can be used to discriminate patients      proximal (‘‘receptor’’) mechanism. More likely,
with potentially more aggressive tumors.            they act through different steps in a cascade lead-
                                                    ing to the same end point—namely, clonal ex-
mechanisms of tumor-promoting                       pansion of initiated cells and progressive selec-
agents                                              tion of genetically variant populations of tumor
The terms tumor promotion, tumor progression,       cells.
and multistage carcinogenesis are overlapping          Tumor-promoting phorbol esters produce a
and somewhat redundant. Some people use             wide variety of biochemical changes in cells. A
these terms interchangeably and some use them       number of these changes may be related to the
to define discreet steps in the carcinogenesis       ability of these agents to promote the growth of
process. Mechanistically, tumor promotion and       initiated tumor cells in vivo. Many of the cellular
progression are a continuum, even though they       changes induced by phorbol esters are reminis-
appear to be ‘‘multistage.’’ Promotion involves a   cent of characteristics of the transformed phe-
clonal expansion (proliferative phase), and pro-    notype (see Chapter 4). The effects of phorbol
gression usually refers to the genetic alteration   esters on cultured cells include (1) induction of
phase. But as was noted above, the genes in-        ornithine decarboxylase, 50 -nucleotidase, AT-
volved in these steps are overlapping or similar.   Pase, and plasminogen activator activities; (2)
Nevertheless, studies of chemical carcinogenesis    stimulation of sugar transport, DNA synthesis,
models have been used to define and discrimi-        and cell proliferation; and (3) alteration of cell
nate initiation events and promotion or progres-    morphology with a loss of cell surface fibronectin
sion events, and these studies have been useful     and the appearance of a diffuse pattern of actin-
in determining the genetic and biochemical steps    containing cytoskeletal elements (reviewed in
involved in these steps, as well as providing       Reference 60). In addition, phorbol esters stim-
targets for drug therapy and chemoprevention.       ulate anchorage-independent growth of adeno-
   The isolation and characterization of tumor-     virus-transformed cells61 and inhibit the termi-
promoting agents have provided the tools to         nal differentiation of chicken myoblasts62 and
study the mechanisms of tumor promotion in          chondroblasts,63 murine lipocytes,64 erythro-
vitro and in vivo. The reader is reminded that      leukemia cells,65 and neuroblastoma cells.66
these agents are primarily defined by their          Tumor-promoting phorbol esters also transform
ability to promote skin carcinogenesis in the       mouse embryo fibroblasts treated with ultravi-
mouse skin-painting assay, and the mechanisms       olet light67 and enhance the transformation of
by which they do this may or may not be rele-       human lymphocytes by Epstein-Barr virus.68
vant to the mechanism of tumor promotion and        These cell culture effects are exerted by low
progression during carcinogenesis in other or-      concentrations (nonomolar range) of phorbol
gans in experimental animals or in humans.          esters, and there is generally a correlation be-
Nevertheless, the study of these compounds has      tween the potencies of phorbol esters for the
been extremely useful in determining the bio-       cell culture effects and their potencies as pro-
chemical actions of tumor promoters. Of the         moters in mouse skin carcinogenesis. Phorbol
promoting agents examined, the phorbol esters       esters share a number of biological properties
have been the most widely studied. Still, one       with epidermal growth factor (EGF) and may
must ask: what the ‘‘phorbol esters’’ are in hu-    act by mechanisms similar to EGF.
man carcinogenesis. Most likely they are factors       An interesting observation suggests that TPA
to which we are continually exposed through our     can induce neoplastic transformation of fibro-
diet, cigarette smoke, and other kinds of envi-     blasts from humans genetically predisposed to
ronmental agents. This answer leads to a second     cancer.69 In these experiments, fibroblasts de-
question: Do all these agents act through the       rived from individuals with familial adenomatosis
CAUSES OF CANCER                                                                                     37

of the colon and rectum were treated with TPA        cell membranes and can indirectly damage chro-
in culture and then injected into athymic mice.      matin by generating oxygen radicals.
Cultures treated with TPA produced tumors in            TPA and other tumor-promoting phorbol
the mice, whereas untreated cultures did not.        esters stimulate Naþ outside/Hþ inside exchange
These results indicate that the fibroblasts from      across cell membranes, leading to a transient
adenomatosis patients exist in an ‘‘initiated’’      intracellular alkalinization that precedes mito-
state due to the dominant mutation that pro-         genic events and appears to be coupled to at
duces the disease, and that this dominantly in-      least some of the phorbol ester–induced chan-
herited trait can be induced to undergo malig-       ges in gene expression in target cells.71–73 TPA-
nant progression by treatment with promoting         induced Naþ/Hþ exchange also produces cell
agents alone. This observation supports the idea     swelling,74 and this could lead to further chan-
that initiation of cancer is a mutagenic event and   ges in cell functions and integrity. Other mem-
has profound implications for human cancer.          brane effects of tumor-promoting phorbol es-
For example, if the promoting agents present in      ters are inhibition of gap-junctional intercellular
our environment could be identified and expo-         communication;75 phosphorylation of cell surface
sure to them eliminated or significantly dimin-       receptors for EGF, insulin-like growth factor 1
ished, could human cancer be prevented? This         (IGF-1), insulin, and transferrin, leading to de-
approach could conceivably be more effective         creased ligand–receptor binding and increased
than eliminating exposure to initiating agents,      receptor internalization;76 reorganization of ac-
since exposure to them need be only very short       tin and vinculin elements in the cytoskeleton;77
and is irreversible. Completely preventing ex-       and structural rearrangement of the nuclear
posure to initiating agents over a lifetime is not   matrix–intermediate filament scaffold.78
practical; however, if the promotion phase takes        Tumor-promoting phorbol esters also appear
15 to 20 years, expanding it to 30 to 40 years       to be able to alter cellular gene expression by
would mean that most individuals could have a        indirectly altering DNA structure and chromo-
life expectancy approaching normal before they       somal proteins by generating oxygen radicals
developed a fatal cancer.                            (reviewed in Reference 79). TPA has been
    Chronic application of TPA to mouse skin         shown to induce chromosomal alterations in a
indicates that a number of target cells may be       variety of human cell types, and these effects are
involved in the tumor promotion phase of skin        inhibited by the addition of antioxidants. TPA
carcinogenesis.70 When TPA is applied twice          also stimulates poly-ADP-ribosylation of chro-
weekly to the skin of mice for several weeks, cell   mosomal proteins in human monocytes, an ef-
damage, edema, and acute inflammation occur           fect frequently stimulated by DNA strand break-
in both the epidermis and dermis during the          age and one that could modify gene expression.
first week. By 3 weeks, epidermal hyperplasia         Further evidence for the role of oxygen radicals
occurs, accompanied by chronic inflammation           in tumor promotion comes from the obser-
in the dermis and hyperplasia of the hair folli-     vations that O2À, H2O2, and certain organic
cles. These features remain until TPA treatment      hydroperoxides promote carcinogenesis in
ceases. Although many of these changes regress       chemical- or irradiation-initiated cells. In con-
within 2 weeks after TPA application ceases, an      trast, antioxidants such as butylated hydroxy-
increased number of hair follicles, capillary        toluene and butylated hydroxyanisole inhibit
vessels, and mast cells in the dermis and an in-     transformation of initiated, TPA-treated mouse
creased dermal thickness remain. This finding         cells. Some of the changes in gene expression
indicates that the target cells for TPA are not      induced by TPA may be due to oxygen-radical
confined to the epidermis and suggests that           generation, since induction of ornithine dec-
chronic tissue irritation, as evidenced by the       arboxylase by TPA is blocked by the antioxidant
hyperplastic and inflammatory responses, plays        enzymes catalase and superoxide dismutase.
a role in tumor promotion. Although the mech-        These DNA-damaging effects of tumor promot-
anism of this damage is not totally clear, it has    ers would be expected to induce chromosomal
been demonstrated that tumor-promoting               breaks and gene rearrangements. It has been
phorbol esters produce a number of changes in        demonstrated that during a single-step selection
38                                                                                    CANCER BIOLOGY

assay for methotrexate (MTX) resistance in              released by tumor cells would foster degrada-
cultured mouse fibroblasts, TPA causes a 100-            tion of the growth-limiting basal lamina, release
fold increase in the incidence of MTX-resistant         of such enzymes may be another way in which
colonies, an effect shown to be due to MTX-             tumor promoters foster tumor expansion and
gene amplification in these cells.80 Thus, tumor         ultimately invasion into underlying tissues. TPA
promoters could also alter gene expression by           also induces angiogenesis, at least in vitro, as
favoring gene rearrangements and gene ampli-            evidenced by its ability to cause cultured en-
fication events in initiated cells that already have     dothelial cells to infiltrate into an underlying
damaged DNA and a propensity for genetic                collagen matrix and form an extensive network
instability.                                            of capillary-like structures.82
   The DNA-damaging effects of tumor-                       Many of the effects of tumor-promoting
promoting agents seem to be incompatible with           phorbol esters are thought to be due to their
the view that the tumor-promotion phase of car-         ability to activate a calcium-dependent protein
cinogenesis is at least partially reversible. How-      kinase known as protein kinase C (PKC). The
ever, most cells have mechanisms to protect             mechanism for TPA activation of PKC has been
themselves against the generation of oxygen rad-        worked out (reviewed in References 83 and 84).
icals, and the ability of agents like TPA to produce    Interaction of TPA with its receptor kinase fa-
oxygen radical–mediated damage in normal cells          vors binding of the inactive cytosolic form to the
may be relatively low. This effect would be ex-         cell membrane, where it is activated. TPA acts
pected to be increased in cells whose DNA               as diacylglycerol (DAG) does, and can substitute
was already damaged or in cells whose oxy-              for it by increasing the affinity of PKC for Ca2þ
gen radical–scavenging mechanisms are compro-           and phosphatidylserine, thereby fostering the
mised. Moreover, the effects of oxygen radical–         translocation of PKC from cytosol to plasma
induced damage may be cumulative over time,             membrane and causing its activation. Part of TPA
thus explaining the long duration of the tumor          chemically resembles DAG. TPA thus acts syn-
promotion and progression phase. It could also          ergistically with Ca2þ-mobilizing agents, such as
explain, at least in part, why cancer is a disease of   those that activate the inositol phospholipid
aging, since aged individuals would have accu-          turnover cascade (see Chapter 4). Unlike DAG,
mulated many more ‘‘hits’’ on their genetic ma-         TPA and similar phorbol esters have a long half-
terial over time, and there is some evidence that       life in cellular membranes, which may explain
the ability to scavenge free radicals decreases in      how they can provide a prolonged signal for cell
senescent cells. It is interesting that cells from      proliferation, unregulated by the normal feed-
patients with hereditary diseases, such as ataxia       back mechanism provided by turnover of DAG
telangiectasia, Fanconi’s anemia, and Bloom’s           and the subsequent inactivation of PKC.
syndrome, which are all characterized by in-
creased cancer incidence, are hypersensitive to
                                                        Experimental Models for the
damage by agents that induce oxygen-radical
                                                        Study of Carcinogenesis
formation. For example, increased oxygen tension
causes an excessive amount of chromosomal aber-         A number of models for the study of carcino-
rations in cells from patients with Fanconi’s ane-      genesis have been developed over the years.
mia.79 Moreover, the serum of patients with ataxia      Historically, two of the most useful ones have
telangiectasia or Bloom’s syndrome contains             been the initiation-promotion model of mouse
DNA-breaking (‘‘clastogenic’’) factors; this effect,    skin carcinogenesis (the ‘‘skin-painting’’ model)
which can be observed when this serum is added          and the induction of liver cancers in rats.
to cultures of normal human cells, is inhibited by         The classic model of carcinogenesis is the
addition of superoxide dismutase to the cultures.       single application of an initiating agent such as a
   In addition to stimulating cell proliferation        polycyclic aromatic hydrocarbon followed by the
and altered gene expression, phorbol ester tumor        continuous application of a promoting agent like
promoters induce the secretion of plasminogen           TPA to the backs of shaved mice. Much of what
activator and type IV collagenase by human fi-           we know about tumor initiation, promotion, and
broblasts.81 Because proteases and collagenases         progression has come from this model system.
CAUSES OF CANCER                                                                                     39

   Initiation and promotion during mouse skin        the carcinogenic damage and to allow for the
carcinogenesis produce multiple benign squa-         initial proliferation of a damaged clone of cells.
mous papillomas. A few squamous cell carci-          Once the damaged clone is present, it could
nomas eventually arise from the papillomas over      undergo alteration due to its genetic instability
many months. However, malignant conversion           and gradually progress to a detectable malignant
can be speeded up by exposure of papilloma-          tumor. This idea is supported by the experi-
bearing mice to mutagens, which activates on-        ments of Pitot et al.,91 who treated rats with a
cogenes such as H-ras and causes loss of tumor       single dose of diethylnitrosamine by intubation
suppressor genes such as p53, as noted above.        24 hours after partial hepatectomy (partial re-
   The mouse skin carcinogenesis model is also a     moval of the liver), which stimulates DNA syn-
useful one in which to study the role of diet and    thesis and cell proliferation in the remaining
chemopreventive agents in carcinogenesis (see        tissue. If the animals were then treated, starting
also Chapter 9). For example, calorie-restricted     8 weeks later, with phenobarbital in the diet
diets have been shown to reduce the number           for 6 months, many small, phenotypically het-
and size of papillomas during and following          erogeneous foci characterized by glucose-6-
promotion with TPA in DMBA-initiated SEN-            phosphatase–deficient areas, ATPase-deficient
CAR mice.85 Furthermore, the latency period          areas, and g-glutamyltranspeptidase-containing
for occurrence of carcinomas was increased and       areas developed in the liver. Many of these
the total number of carcinomas was decreased.        animals also had hepatomas, for which the
Applicationofapigenin,aplantalkaloid,86 retinoic     enzyme-altered foci appear to represent the
acid,87 and prostratin, a nonpromoting phorbol       early stage of neoplastic development. Thus in
ester88 have been shown to inhibit the promo-        this case, phenobarbital appears to have stimu-
tion phase (appearance of papillomas) of mouse       lated the replication of dormant initiated cells,
skin carcinogenesis.                                 which, in the absence of the promoter, would
   Multistage carcinogenesis has also been ob-       not have proliferated. If each enzyme-altered
served for liver tissue. For example, Peraino        focus observed in these experiments were a
et al.89 observed that a 3-week exposure of rats     clone derived from a single cell, about 104 to 105
to AAF in the diet produced only a small num-        cells in the liver were ‘‘initiated’’ by diethylni-
ber of hepatomas after several months, but if the    trosamine, and a very small number of these
animals were subsequently treated with phe-          subsequently underwent clonal proliferation
nobarbital for several months after carcinogen       during phenobarbital feeding.91 Thus the con-
feeding was discontinued, a high incidence of        version of these abnormal foci, or early nodules,
hepatomas was noted. Similar results have been       as they have been called, to a malignant neo-
obtained by Kitagawa et al.,90 who fed rats a        plasm is a rare event.
nonhepatocarcinogenic dose of 2-methyl-N,N-             Newer models of carcinogenicity have involved
dimethyl-4-aminoazobenzene for 2 to 6 weeks,         the use of knock-out or knock-in rodent models,
and then a dietary administration of phenobar-       in which various oncogenes, tumor-suppressor
bital for 70 weeks. By 72 weeks, many large          genes, or susceptibility genes have been engi-
hepatocellular carcinomas had developed in the       neered into or out of rodent embryos (usually
phenobarbital-treated animals, whereas only a        mice). This process has enabled the definition of
few small tumor nodules were observed in the         some of the genes that are key to various steps in
rats not given phenobarbital. Thus, the action of    the tumor-initiation promotion and progression
phenobarbital appears to be analogous to that of     steps. These tumor models are now being super-
TPA in the mouse skin system—that is, it ‘‘fixes’’    ceded by conditional genetic knock-out models in
the damage to cells induced by an initiating         mice that allow for the controlled expression of
agent and causes a clone of cells arising from a     oncogenes or tumor suppressor genes in a way
damaged cell to proliferate. However, whereas        that more closely mimics ‘‘spontaneously’’ arising
TPA stimulates DNA synthesis and hyperplasia         human cancers (Table 2–2).
in skin, phenobarbital produces only a transient        Conditional gene expression in the mouse has
and relatively small increase in DNA synthesis       been achieved by mutations induced by FLP/
in liver. Perhaps that is all that is needed to fix   FRT or Cre/lox P site-specific recombination
40                                                                                                         CANCER BIOLOGY

           Table 2–2. Conditional and Inducible Mouse Tumor Models
           Tumor Type                                   Conditional or Inducible Gene*
           Colorectal adenomas                          ApcIoxP
           Mammary adenocarcinomas                      Brca1IoxP þ Trp53þ/À
           Mammary adenocarcinomas                      Brca2IoxP þ Trp53IoxP
           Mammary adenocarcinomas                      Brca2IoxP
           Schwannomas                                  Nf 2IoxP
           Lung adenocarcinomas                         StopIoxP þKrasG12D
           Pituitary tumours                            Rb IoxP or RbFRT
           Medulloblastomas                             RbIoxP þ Trp53IoxP
           Liver haemangiomas                           VhlIoxP
           Lung tumors, thymic lymphomas,               KrasG12D
             skin papillomas
           GENE DELIVERY
           Gliomas                                      ErbB2 þ Cdk4 or ErbB2 þ Cdkn2aÀ/À
           Gliomas                                      Pdfg or Pdfg þ Cdkn2aÀ/À
           Gliomas                                      PyV-mT
           Glioblastomas                                KrasG12D þ Akt
           Ovarian carcinomas                           Combinations of KrasG12D, Myc, and Akt in wildtype
                                                          or Trp53À/À background
           B-cell leukemia                              Bcr-Abl1
           Skin tumors                                  ErbB2
           Lung hyperplasias                            Fgf 7
           Papillomatosis                               Myc
           T-cell lymphomas, AML                        Myc
           T-cell lymphomas                             Myc
           Melanomas                                    HrasV12G þ Cdkn2a
           Lung adenocarcinomas                         KrasG12D þ Cdkn2aÀ/À or KrasG12D þ Trp53À/À
           Salivary gland hyperplasia                   SV40 Tag
           *Combinations with conventional tumor suppressor gene knockouts (for example, Cdkn2aÀ/À or Trp53À/À)
           are included. Apc, adenomatous polyposis coli; Bcr-Abl1, breakpoint cluster region Abelson 1; Brca1, breast
           cancer gene 1; Brca2, breast cancer gene 2; Cdk4, cyclin-dependent kinase 4; Cdkn2a, cyclin-dependent
           kinase inhibitor 2a (which encodes the Ink4a and Arf tumor suppressors); ErbB2, avian erythroblastic
           leukemia viral oncogene homologue 2 (which encodes an epidermal growth factor receptor homologue);
           Fg f 7, fibroblast growth factor 7; Hras, Harvey rat sarcoma viral oncogene homologue; Kras, Kirsten rat
           sarcoma viral oncogene homologue; Myc, avian myelocytomatosis viral oncogene homologue; Nf 2, neuro-
           fibromatosis type 2; Pdg f, platelet-derived growth factor; Pyv-mT, polyomavirus middle T antigen; Rb,
           retinoblastoma; SV40 TAg, simian virus 40 large T antigen; Trp53, transformation-related protein 53; Vhl,
           Von Hippel-Landau. (From Jonkers and Berns,92 reprinted by permission from Macmillan Publishers Ltd.)

systems, regulatable oncogene expression, and                     the various stages of tumor development. These
retroviral gene transfer in transgenic mice that                  models also provide a way to validate various
express an avian retroviral receptor (reviewed in                 targets for anticancer drug development.
Reference 92). These models provide for the
induction of somatic mutations in tissue-specific
                                                                  Validity of Tests for Carcinogenicity
and time-sequenced way. These models more
closely mimic human cancer development, which                     There is quite a bit of debate among scientists
involves the activation of oncogenes and inacti-                  and regulatory agencies about how to assess the
vation of tumor suppressor genes over time.                       carcinogenic hazards of chemicals, both man-
They enable investigators to determine the                        made and natural, in our environment. Much of
contribution that individual mutations make to                    this debate has spilled over into the media,
CAUSES OF CANCER                                                                                        41

generating a sort of ‘‘carcinogen-of-the-month         The problem is that (1) humans would have to
club’’ and much confusion among the public.            consume 25 kilograms of sodium saccharin a day
Indeed, as one observer put it, ‘‘Cancer news is a     to achieve the cancer-causing dose in male rats,
health hazard.’’93                                     and (2) humans don’t have the same excretion
   For many years, the prevailing view among           patterns as those of male rats.97 It was later found
cancer epidemiologists has been that 60% to 90%        that male rats are considerably more susceptible
of human cancers are attributable to environ-          to the saccharin-caused bladder cancer than fe-
mental and lifestyle factors, including cigarette      male rats, mice, hamsters, monkeys, and, ac-
smoking, diet, ultraviolet irradiation, sexual prac-   cording to epidemiological studies, humans. The
tices, parasitic and viral infections, industrial      main difference among these species is that male
pollution, and, more recently, pesticides.94,95        rats excrete large amounts of protein in the urine,
   The implication of this attribution is that most    unlike the other species, and silicate microcrys-
cancers are preventable. Thus the prevailing           tals form in the male rats’ urinary bladder as a
view, adopted by federal regulatory agencies, is       result. This leads to a chronic irritation of the
that, as much as possible, all carcinogens should      uroepithelium, increased cell proliferation, and a
be eliminated from the environment. One out-           hyperplasia eventually leading to carcinoma.97
come of this view is the famous (or infamous,          These effects are not observed in humans. The
depending on your point of view) Delaney               realization of this species and dose-reality dif-
clause to the Food and Drug Act.96 The Delaney         ference eventually resulted in the re-release of
clause interdicts the use of any food additive         saccharin as a sweetener, albeit with warnings
in processed foods (interestingly, it pays no at-      being required on the package.
tention to pesticide residue on non-processed             The example of saccharin and certain other
foods) that is found to be carcinogenic at any         chemical carcinogens raises several important
dose in one or more animal species. Methods to         questions about carcinogenicity testing. What is
detect carcinogens have since become a major           an appropriate dose and time frame to use in
issue for the food and pharmaceutical indus-           animalcarcinogenicityassays?Whateffectsmight
tries. The question, then, is how can carcinogens      lower dose but lifetime exposure to a chemical
be identified before they are found retrospec-          have on humans? Is there a ‘‘threshold’’ dose for
tively to cause human cancer?                          a given agent, below which human exposure is
   The classic approach has been long-term             safe? Are there cheaper, more accurate, faster
studies in rodents, exposing them to a ‘‘maximum       ways to find out if a chemical is carcinogenic?
tolerated dose’’ (MTD) for the life span of the        How does one evaluate the relative risk of ex-
animal. An MTD is defined as the highest dose           posure to natural chemicals in food and water
that can be given without causing severe weight        compared to man-made chemicals? Is there an
loss or other signs of life-threatening toxicity.      acceptable risk–benefit ratio for economically
This kind of testing is very expensive (the esti-      important chemicals? Can society afford the
mated range is up to $1 million to $2 million per      cost of removing every trace of a potentially
compound) and time consuming (3 to 4 years). In        carcinogenic substance from the environment?
addition, the doses used in these tests are usually    Are there dietary or other factors that can be
orders of magnitude higher than those that most        used to supplement diets to prevent or delay
humans would ever be exposed to. Moreover,             cancer even if one is exposed to known or un-
these tests fail to take into account the differ-      known carcinogens? Can highly susceptible in-
ences in pharmacokinetics, drug metabolism, and        dividuals be identified so that they can be ad-
excretion mechanisms between mice or rats and          vised to avoid certain employment or activities?
humans. This situation has led to some fascinat-       Can exposure levels be determined that could
ing snafus. Saccharin is a good example. Sodium        lead to removal of substances from an environ-
saccharin, the artificial sweetener, was shown to       ment and/or careful follow-up of individuals so
increase the incidence of bladder cancer in rats       exposed?
when administered beginning at conception or at           These are all difficult questions to answer,
birth and continuing through an animal’s life-         and there is considerable debate about the an-
time. Hence, it was banned as a food additive.         swers, but some approaches have been taken to
42                                                                                    CANCER BIOLOGY

answer some of them. The Ames test for muta-           of examples of genotoxic chemicals, e.g., 2-
genesis in bacteria Salmonella is one of the           acetylaminofluorene (2-AAF), that cause differ-
widely used short-term tests that have been            ent effects in different tissues depending on the
developed. It is fast, inexpensive, and reasonably     proliferative response they produce in a given
accurate. However, the Ames test doesn’t detect        tissue. The authors also cite non-genotoxic chem-
certain kinds of chemical carcinogens, e.g.,           icals that can act at high enough doses to induce
those that don’t bind to DNA such as chloro-           cell damage and a mitogenic response that in-
form and dibromochloromethane, and it over-            creases the genetic error rate. In the example of 2-
predicts carcinogenicity for others. In one sur-       AAF, even though the dose–response effect may
vey of 224 chemicals, the Ames test had a              depend on the chemical’s effect on cell prolifer-
sensitivity of only 54% (percentage of true car-       ation in a given tissue, a threshold for the carci-
cinogens identified) and a specificity of 70%            nogenic effect is unlikely. In the case of non-
(true negatives, i.e., chemicals correctly identi-     genotoxic chemicals, a threshold effect is likely.
fied as noncarcinogens).98                                 Weinstein and others100,101 have argued that
   Other short-term tests include induction of         there is definitive evidence that spontaneous
resistance to antimetabolites in cultured cell         mutation or endogenous DNA damage is car-
lines, and chromosomal breakage in exposed             cinogenic and that there is no consistent corre-
cultured cells. Using the data from short-term         lation between the inherent growth fraction
tests, particularly the Ames test, in combination      (percent proliferating cells) in a tissue and
with determination of a ‘‘chemically alerting’’        cancer incidence in that tissue. It seems, how-
structure (i.e., a chemical containing structural      ever, that cell proliferation has to occur for a
elements known to be carcinogenic, such as             cancer to develop and that the self-renewal stem
electrophile-producing side chains), Ashby and         cells of a tissue are the most likely targets for
Tennant99 showed a high correlation between            carcinogens. Unfortunately, there is no well-
those compounds that were structurally alerting        defined way to simulate whole-organism carci-
and those that were mutagenic. When taken to-          nogenicity testing other than in whole organ-
gether, a reasonable correlation with carcino-         isms, where the parameters of absorption, tissue
genicity was predicted, based on carcinogenicity       distribution, metabolism (activation and deacti-
in animal tests.                                       vation), pharmacokinetics, and excretion can be
   One of the problems with the use of the MTD         evaluated. Furthermore, it can be argued that a
approach to carcinogenicity testing in animals is      full dose-range, including doses that stimulate
that in addition to being a dose often an order of     cell proliferation, of potential carcinogens should
magnitude more than a dose to which humans             continue to be tested in animals.101 The diffi-
are likely to be exposed, the MTD is a dose that is    culty comes in translating these data to realistic
often sufficient to kill cells and induce a prolif-     human exposure levels. Thus, computers will
erative, tissue-repair response in target organs.      still not be able to replace good judgment and
This increased cell proliferation puts cells at risk   common sense.
for propagating an unrepaired lesion produced             The debate about linear versus nonlinear
by the spontaneous background mutation rate            models for estimating the human carcinogeni-
(see above) or by exposure to an environmental         city of chemicals continues.102 The linear, no-
agent. This has led Cohen and Ellwein97 to             threshold model (Fig. 2–10) assumes that all
propose a biological model for carcinogenesis          carcinogens act similarly and that there is no
based on whether a chemical is genotoxic or non-       threshold for their carcinogenic action. The
genotoxic. The key feature of this model is that       logical conclusion from this assumption is that
an agent can increase the incidence of cancer          some risk of carcinogenicity exists for any dose
by either damaging a cell’s DNA (genotoxic) or         of a carcinogen. This notion defies what is
stimulating cell proliferation (non-genotoxic),        known from all other toxicological and phar-
increasing the likelihood for a spontaneous            macological events, i.e., that there is a dose below
genetic error to occur during DNA replication.         which no toxicity or drug response occurs (Fig.
They provide as evidence for this a number             2–11). It also implies that cancer is a one-hit
CAUSES OF CANCER                                                                                                                  43

                                           Measured cancer risk                                            Measured cancer risk
                                           at high dose                                                    at high dose

                                                                  Increasing incidence of tumors
Increasing incidence of tumors

                                 Increasing dose                                                   Increasing dose

                                                                  Figure 2–11. Nonlinear threshold. (From America’s
Figure 2–10. Linear curve. Curves with this appear-
                                                                  War on ‘‘Carcinogens’’: Reassessing the Use of Ani-
ance are not usually found experimentally in dose–
                                                                  mal Tests to Predict Human Cancer Risk, p. 53, with
response assays, and the idea that a dose–response
curve could take such a form is now considered obso-
lete. (From America’s War on ‘‘Carcinogens’’: Re-
assessing the Use of Animal Tests to Predict Human
Cancer Risk, p. 52, with permission.)
                                                                  mutagens, teratogens, and clastogens (DNA-
                                                                  damaging agents).104 Thus the authors con-
                                                                  cluded that natural and synthetic chemicals are
event. The no-threshold risk calculation is de-                   equally likely to be positive in animal cancer
rived from the extrapolation of high–dose ex-                     tests and that at the low doses of most human
posure situations. For example, the incidence of                  exposures, ‘‘the comparative hazards of synthetic
lung cancer in deep mineworkers exposed to                        pesticide residues are insignificant.’’103
radon is plotted versus radon exposure linearly
back through 0/0 on the plot (Fig. 2–10). In
other words, some infinitismal exposure to ra-                     IRRADIATION CARCINOGENESIS
don will still cause some cancers. From this sort
of extrapolation, it is concluded that no level of                A number of the points made about chemical
exposure to radon is safe. This model is still the                carcinogenesis are also true for radiation-induced
one most commonly used for cancer risk assess-                    carcinogenesis. Both X-rays and ultraviolet (UV)
ment by government agencies and its use has                       radiation, for example, produce damage to DNA.
led to the propagation of a number of ‘‘cancer                    As with chemical carcinogens, this damage in-
myths’’ (see Chapter 3).                                          duces DNA repair processes, some of which are
   Ames and colleagues103,104 have argued that                    error prone and may lead to mutations. The de-
natural chemicals are as likely as synthetic chem-                velopmentofmalignanttransformationincultured
icals to be carcinogenic in various tests and that                cells after irradiation requires cell proliferation to
on the basis of actual levels of human exposure                   ‘‘fix’’ the initial damage into a heritable change and
(other than industrial workers or farmers, for                    then to allow clonal proliferation and expression
example, who might have high exposure rates to                    of the typical transformed phenotype.105 Fixation
industrial chemicals or pesticides), natural che-                 appears to be complete after the first postirradi-
micals are at least as dangerous as synthetic                     ation mitotic cycle. In the case of mouse C3H /10
ones. In high-dose tests, 30% to 50% of both                      T 1/2 cells, expression of radiation-induced
natural and synthetic chemicals are carcinogens,                  transformation requires an additional 12 rounds
44                                                                                  CANCER BIOLOGY

of cell division. Thus, as in the case of chemical   cause cancer was not adopted until the 1950s,
carcinogenesis, a promotion phase is required for    when data from atomic bomb survivors in Japan
full expression of the initiated malignant alter-    and certain groups of patients treated with X-
ation. Moreover, when low doses of chemical          rays for noncancerous conditions, such as en-
carcinogens and X-rays are used together, these      larged thyroids, were analyzed. These and other
two types of agents act synergistically to produce   data led to the concept that the incidence of
malignant transformation.105                         radiation-induced cancers might increase as a
   When cells are exposed to UV light in the 240     linear, nonthreshold function of dose. Thus the
to 300 nm range, the bases acquire excited en-       debate about whether there is a safe threshold
ergy states, producing photochemical reactions       pertains to radiation carcinogenesis, just as it
between DNA bases (reviewed in Reference             does to chemical carcinogenesis.
106). The principal products in DNA at bio-             In radiation carcinogenesis, the damage to
logically relevant doses of UV light are cyclo-      DNA, and hence its mutagenic and carcinogenic
butane dimers formed between two adjacent            effect, is due to the generation of free radicals as
pyrimidine bases in the DNA chain. Both              the radiation passes through tissues. The amount
thymine–thymine and thymine–cytosine dimers          of radical formation and ensuing DNA damage
are formed. That formation of these dimers is        depend on the energy of the radiation. In gen-
linked to mutagenic events (see below).              eral, X-rays and gamma rays have a low rate of
   Heavy exposure to sunlight induces similar        linear energy transfer, generate ions sparsely
changes in human skin, and the degree of ex-         along their tracks, and penetrate deeply into
posure to sunlight is closely related to the inci-   tissue. This profile contrasts with that of charged
dence of skin cancer (see Chapter 3). Whether        particles, such as protons and a particles, which
continuing exposure to UV rays in sunlight is the    have a high linear energy transfer, generate
promoting agent in skin cancer or additional         many more radical ions locally, and have low
promoting events are required is not clear, but it   penetration through tissues. The damage to DNA
seems that UV irradiation is a complete carcin-      can include single- and double-strand breaks,
ogen, just as some chemicals are—that is, it has     point mutations due to misrepair deletions, and
both initiating and promoting activities. Patients   chromosomal translocations.107–109 The molec-
who cannot efficiently repair UV-induced dam-         ular genetic events that follow radiation damage
age, such as those with xeroderma pigmento-          to cells include (1) induction of early-response
sum, have a much higher risk of developing           genes such as c-jun and Egr-1; (2) induction
malignant skin tumors.                               of later-response genes such as tumor necro-
                                                     sis factor-a (TNF-a), fibroblast growth factor
                                                     (FGF), and platelet-derived growth factor-a
Ionizing Radiation
                                                     (PDGF-a); (3) activation of interleukin-1 (IL-1)
The history of radiation carcinogenesis goes         PKC110; and (4) activation of oncogenes such as
back a long way (reviewed in Reference 107).         c-myc and K-ras.111 Induction of these genes
The harmful effects of X-rays were observed          may be involved in the cellular responses to ir-
soon after their discovery in 1895 by W. K.          radiation and in the longer-range effects that
Rontgen. The first observed effects were acute,       lead to carcinogenesis. At any rate, the pro-
such as reddening and blistering of the skin         duction of clinically detectable cancers in hu-
within hours or days after exposure. By 1902, it     mans after known exposures generally occurs
became apparent that cancer was one of the           after long latent periods. Estimates of these la-
possible delayed effects of X-ray exposure.          tent periods are 7 to 10 years for leukemia, 10–
These cancers, which included leukemia, skin         15 years for bone, 27 years for brain, 20 years for
cancers, lymphomas, and brain tumors, were           thyroid, 22 years for breast, 25 years for lung,
usually seen in radiologists only after long-term    26 years for intestinal, and 24 years for skin
exposure before adequate safety measures were        cancers.
adopted, thus it was thought that there was a           A more recent example of nuclear fallout
safe threshold for radiation exposure. The hy-       leading to environmental exposure to radiation
pothesis that small doses of radiation might also    is the Chernobyl accident, which happened on
CAUSES OF CANCER                                                                                    45

April 26, 1986. A steam explosion blew the lid        the fallout, and inhabitants of parts of the
off the reactor. The graphite core caught fire         Ukraine and Belarus are still exposed to low
and over 1019 becquerels (Bqs) of radioisotopes       levels of radioactive cesium. The long-term ef-
were released, producing a fallout that covered       fects, if any, of such exposure is not yet clear.
much of Belarus, Northern Ukraine, and part
of the Russian Federation. Estimates are that
10–20 million people were exposed to significant       Ultraviolet Radiation
fallout. There were some deaths due to acute
                                                      Ultraviolet radiation–induced lesions, generated
radiation sickness from high levels of exposure.
                                                      by UV-B (280–320 nm wavelength) or UV-A
However, the long-term effects are still being
                                                      (320–400 nm wavelength), result from DNA
recorded. So far, the reliable reports of increases
                                                      damage, which is converted to mutations during
in cancer incidences are mostly limited to thy-
                                                      cellular repair processes. UB-B and UV-A gen-
roid cancer.112 This finding is in contrast to
                                                      erate different types of DNA damage and DNA
cancer incidence among atomic bomb survivors
                                                      repair mechanisms (reviewed in Reference
in Japan, some of whom developed cancers of
                                                      113). Irradiation with UV-B produces cyclobu-
various types, including cancers of the thyroid,
                                                      tane pyrimidine dimers that are repaired by
breast, lung, stomach, esophagus, bladder, leu-
                                                      nucleotide excision repair. If left unrepaired,
kemia, and lymphoma (although the incidence
                                                      C?T and CC?TT base transitions occur. UV-
of cancers in Japanese atomic bomb survivors
                                                      A-induced DNA damage produces mostly oxi-
was less than would have been predicted by
                                                      dative lesions via photosensitization mecha-
radiation exposure). The reason for this dis-
                                                      nisms and is repaired by base excision repair.
crepancy is most likely that those exposed to the
                                                      UV-B and UV-A also produce different effects
Chernobyl fallout received primarily dosage from
                                                      on the immune system and elicit different tran-
b-emitters, mostly isotopes of iodine, which con-
                                                      scriptional and inflammatory responses. While
centrates in the thyroid. Atomic bomb survivors,
                                                      the specific mechanisms by which UV radiation
by contrast, received whole-body irradiation
                                                      induces basal cell or squamous cell carcinomas
from neutrons and gamma rays.
                                                      or melanoma are not clear, a number of signal
   Another interesting point about the Cherno-
                                                      transduction pathways are affected that can ei-
byl survivors is that the type of thyroid cancer
                                                      ther lead to apoptosis or to increased cell pro-
they developed, mostly among those under 2
                                                      liferation (Fig. 2–12). UV irradiation activates
years of age if they were exposed, were 98%
                                                      receptor tyrosine kinases and other cell surface
papillary, many with an unusual morphology,
                                                      receptors. It also enhances phosphorylation by
whereas in non-exposed populations, only 67%
                                                      ligand-independent mechanisms via inhibition
of childhood thyroid cancers are papillary.112
                                                      of protein tyrosine phosphatase activity. Ligand-
   Expression of two families of oncogenes, the
                                                      dependent cell surface receptor activation can
c-ret and ras families, has been shown to be
                                                      also occur by activation of autocrine or para-
involved in papillary thyroid cancers. The on-
                                                      crine release of growth factors from keratino-
cogene c-ret is a receptor tyrosine kinase acti-
                                                      cytes, melanocytes, or neighboring fibroblasts. It
vated by gene rearrangement, and two of these,
                                                      is clear, however, that better animal models are
ret-ptc 1 and ret-ptc 3, are activated in papillary
                                                      needed to clearly define the mechanisms by
carcinomas. Since c-ret is activated by rear-
                                                      which UV light causes human cancer.
rangement, the high proportion of double-
strand DNA breaks seen in radiation-induced
papillary carcinomas of the thyroid may explain
its activation.                                       OXYGEN FREE RADICALS,
   Since the thyroid is not the only tissue that      AGING, AND CANCER
concentrates iodine, malignancies of other tis-
sues that also concentrate iodine, such as the        The diseases of aging include cardiovascular
breast, salivary gland, and stomach, may appear       disease, decline in function of the immune sys-
in higher incidence as time goes on. Moreover,        tem, brain dysfunction, and cancer. People liv-
other isotopes including cesium were present in       ing in the United States who are 65 or older have
46                                                                                        CANCER BIOLOGY

                                                         decreases with age. There is a fair amount of
                                                         circumstantial evidence to support this latter
                                                         hypothesis (reviewed in References 114–116).
                                                         Oxidative damage to DNA, proteins, lipids, and
                                                         other macromolecules accumulates with age.
                                                         Oxidation products formed during normal
                                                         metabolic processes in cells include superoxide
                                                         (O2À), hydrogen peroxide (H2O2), and hydroxyl
                                                         radical (OH). These are also produced in cells
                                                         by radiation, and they are capable of damaging
                                                         DNA and producing mutagenesis. Of these, the
                                                         hydroxyl radical appears to be the primary
                                                         DNA-damaging species, but it has a short half-
                                                         life and high reactivity, so it must be generated
                                                         in close proximity to DNA.117 This may occur in
                                                         the cell nucleus by an interaction of H2O2 with
                                                         chromatin-bound metals such as Feþ2 by the
                                                         following reaction:
                                                              Feþ2 þ H2 O2 ! Feþ3 þ :OH þ OH:À
Figure 2–12. Immediate and long-term effects of
ultraviolet (UV) radiation on skin cells and their ge-
                                                         Singlet oxygen, which is produced by lipid
nomes are complex. Immediate cellular responses to       peroxidation or by the respiratory bursts from
UV can occur through stress pathways (p38, mitogen-      neutrophils, is also mutagenic and has a much
activated protein kinase [MAPK] and Jun N-terminal       longer half-life than the hydroxyl radical. Lipid
kinase [JNK]), cell surface receptors (receptors ty-     peroxidation can also give rise to mutagenic
rosine kinase [RTKs]) and direct DNA damage. The
response to DNA damage is largely mediated by p53,
                                                         products such as lipid epoxides, hydroperoxides,
which can arrest growth and facilitate DNA repair        alkanyl and peroxyl radicals, and a, b unsatu-
or, if the damage is too extensive, induce apoptosis.    rated aldehydes.114
Mutation-bearing melanocytes that escape these              Cells have multiple mechanisms to protect
fates and survive are the seeds of potential future      themselves from oxidative damage, including
melanomas. Proteins of the Rb pathway, such as
p16Ink4a and cyclin-dependent kinase 6 (Cdk6), are
                                                         superoxide dismutase, catalase, glutathione per-
important biological targets of UV. However, the         oxidase, and glutathione-S-transferases. In addi-
genetic changes observed (loss of p16Ink4a and am-       tion, DNA damaged by oxidants is subject to re-
plification of Cdk6 [green]) are not characteristic of    pair (see below), oxidized proteins are degraded
UV-induced mutagenesis and, hence, these conse-          by proteases, and lipid peroxides are destroyed by
quences are probably indirect, with stochastic mu-
tations selected for in cells that go on to become
                                                         glutathione peroxidase. Nevertheless, some oxi-
melanomas. Arrows do not necessarily represent di-       dative damage and misrepair may persist, and the
rect interactions. ATM, ataxia telengiectasia mutated;   ability to carry out these repair mechanisms de-
ATR, ataxia telengiectasia and Rad3 related. (From       creases with aging. It is estimated that the human
Merlino and Noonan,113 with permission.)                 genome suffers about 10,000 ‘‘oxidative hits’’ to
                                                         DNA per cell per day.114 Mutations accumulate
                                                         with age in the rat so that an ‘‘old’’ rat (2 years old)
10 times the risk of those under age 65 for de-          has twice as many DNA lesions per cell as a young
veloping cancer.                                         rat. Furthermore, the frequency of somatic muta-
   Part of the increase in cancer incidence with         tions found in human lymphocytes is about nine-
aging could be due to an accumulation of dam-            fold higher in the aged than in neonates.118 How
age to DNA over a lifetime of exposure to car-           much of this mutation frequency is due to oxida-
cinogenic substances. Another, perhaps more              tive damage of DNA isn’t clear, but a number of
likely, possibility is that cellular damage pro-         altered bases have been observed in cells under-
duced by endogenous oxidants accumulates over            going oxidative stress. These include hydroxy-
time and the body’s ability to repair this damage        methyl uracil, thymine glycol, 8-hydroxyguanine,
CAUSES OF CANCER                                                                                       47

8-hydroxyadenine, and formamido derivatives of        that one of the major sources of exogenous oxi-
altered purines.119 Some of these products ap-        dant exposure is the oxides of nitrogen found in
pear in the urine and may be an index of oxidative    cigarette smoke.114
damage. They are also produced by exposure of
DNA to ionizing radiation and oxygen-radical
generators. 8-Hydroxyguanine appears to be the        GENETIC SUSCEPTIBILITY
most frequently altered base to result from oxi-      AND CANCER
dative damage to DNA, and if this base is left
unrepaired in DNA it produces G?T trans-              As was noted above, there are a number of in-
version. 119                                          herited cancer susceptibility gene mutations,
   Endogenous oxidants can also damage pro-           such as xeroderma pigmentosum, Fanconi’s ane-
teins and lipids. Oxygen free-radicals catalyze       mia, and ataxia telangiectasia. These types of
the oxidative modification of proteins leading to      inherited defects that lead to cancer are gener-
an age-related increase in carbonyl content of        ally caused by a deficiency in DNA repair path-
cellular proteins.115 For example, there is a         ways. Almost certainly we have only scratched
significant increase in carbonylated proteins in       the surface of inherited cancer susceptibility
human erythrocytes from older individuals, and        genes that make an individual more prone to
the carbonyl content of proteins in cultured          developing cancer. Other susceptibility genes
human skin fibroblasts increases exponentially         may include alterations in the metabolic en-
with the age of the fibroblast donor.115 In ad-        zymes that metabolize drugs and environmental
dition, the protein carbonyl content of fibro-         toxins, polymorphisms in genes that regulate
blasts from individuals with premature aging          utilization of certain essential nutrients such as
(progeria and Werner’s syndrome) is higher            folic acid, or inherited mutations in tumor sup-
than that in age-matched controls. There is also      pressor genes.
an age-related decrease in neutral alkaline pro-         The completion of the Human Genome Pro-
tease activity that degrades oxidized proteins.       ject allows a systematic approach to discovering
The end result is an increased retention of           the genetic alterations that make individuals prone
damaged proteins with aging. The degree to            to developing various diseases. The Environmen-
which oxidative protein damage contributes to         tal Genome Project is producing a catalogue of
diseases of aging and cancer isn’t clear, but         variation in genes involved in catabolizing toxins,
treatment of gerbils with the radical trapping        nutrient metabolism, and DNA repair.121 These
agent tert-butyl-a-phenylnitrone inhibits age-        data, which will be largely generated by detection
related increases in oxidized protein in the brain    of single nucleotide polymorphisms (SNPs), will
and blocks age-related memory loss (as mea-           enable toxicologists and cancer biologists to pre-
sured by a radial-arm maze test).120 Oxygen-          dict individual susceptibility to diseases triggered
radical damage to lipids leads to fluorescent li-      or promoted by environmental pollutants, diet,
pid oxidation products that appear to result from     and other lifestyle factors. Some examples of this
cross-links between proteins and lipid peroxi-        SNP analysis approach are the increased sus-
dation products, and these also increase with         ceptibility of individuals with altered folate me-
age.114                                               tabolism genes to develop leukemia after ben-
   Caloric or protein restriction in the diet slows   zene exposure and the ethnic variation in the
oxidative damage to proteins and DNA and              BRCA1 gene SNPs that affect susceptibility to
decreases the rate of formation of neoplasms in       breast cancer.
rodents. Similar results are seen by dietary
supplementation with antioxidants such as to-
copherol (vitamin E), ascorbate (vitamin C), and      MULTIPLE MUTATIONS IN CANCER
carotenoids such as b-carotene, leading to the
hypothesis, supported by epidemiological data         In most cases, it takes years for a full-blown
in humans, that dietary intake of such sub-           invasive, metastatic cancer to develop from a
stances could decrease the incidence of human         small clone of initiated cells. This process might
cancer (see Chapter 9). It should also be noted       take 20 years or more, during which time an
48                                                                                   CANCER BIOLOGY

initiated clone of cells undergoes clonal expan-      DNA REPAIR MECHANISMS
sion via multiple cell doublings. As these clones
expand, various cells in the population accumu-       Not all interactions of chemicals and irradiation
late multiple genetic alterations, some of which      with DNA produce mutations. In fact, all cells
facilitate dysregulated cell proliferation and        have efficient repair mechanisms that repair such
some of which lead to cell death. These genetic       lesions. DNA repair mechanisms include sets of
alterations can include point mutations, chro-        enzymes that survey DNA for specific kinds of
mosomal translocations, gene deletions, gene          damage, remove the altered portion of DNA, and
amplifications, loss of genetic heterozygosity         then restore the correct nucleotide sequence.
(LOH), and loss of genetic imprinting (LOI).          The important role of DNA repair in human
These will be discussed in detail in Chapter 5.       cancer has been established by the finding that a
This accumulation of genetic defects that occurs      number of inherited defects in DNA repair sys-
during clonal expansion of transformed cells is       tems predispose individuals to getting cancer.
due to ‘‘genetic instability.’’ The cause of this     These diseases include xeroderma pigmentosum,
genetic instability is not clearly understood, but    ataxia telangiectasia, Fanconi’s anemia, Bloom’s
it includes defects in cell replication checkpoint    syndrome, Cokayne’s syndrome, and hereditary
controls and decreased ability to repair DNA          retinoblastoma.125
damage.                                                  There are several types of DNA repair sys-
    There is evidence for the accumulation of         tems, a number of which have been preserved
thousands of mutations in cancer cells derived        from bacteria to humans. These include125–127
from human tumors. For example, examination of        (1) abnormal precursor degradation, e.g., the
the colon tumor–derived DNA from patients with        hydrolysis of the oxidized nucleotide triphos-
hereditary non-polyposis colon cancer (HNPCC)         phate 8-hydroxy-dGTP to its nucleotide 8-OH-
reveals that as many as 100,000 repetitive DNA        dGMP, preventing incorporation into DNA; (2)
sequences are altered from the mismatch DNA           a visible light-activated photoreactivation repair
repair defects that these patients’ cells harbor      mechanism for removal of UV-induced cyclo-
(reviewed in Reference 122). Mismatch repair          butane pyrimidine dimmers; (3) strand break
defects have also been noted in ‘‘sporadic’’ (not     repair via an action of DNA ligase, exonuclease,
known to be hereditary) cancers.                      and polymerase activities; (4) base excision re-
    As noted earlier, one hypothesis explaining       pair that recognizes simple base alterations such
the genetic instability of transformed cells is       as cytosine deamination to uracil and requires
the mutator phenotype hypothesis, championed          the action of (a) a purine or pyrimidine glyco-
by Loeb and colleagues.122 This hypothesis            sylase that breaks the deoxyribose-base bond,
states that an ‘‘initial mutator [gene] mutation      (b) an endonuclease to cleave at the abasic site,
generates further mutations including muta-           (c) a phosphodiesterase to clip away the ‘‘naked’’
tions in additional genetic stability genes, re-      abasic site, (d) DNA polymerase, and (e) DNA
sulting in a cascade of mutations throughout          ligase to refill and reclose the site; (5) nucleotide
the genome.’’ The molecular defect that could         excision repair that recognizes bulky DNA base
provide this phenotype could be a mutation            adducts, pyrimidine dimers, and base cross-
in DNA polymerases that leads to error-prone          links and requires the concerted action of
DNA replication. The mutator phenotype would          enzymes and recognition factors (see below);
have to be generated early in tumorigenesis for       and (6) 06-alkyguanine-DNA alkyltransferase
this hypothesis to be valid. There are a number       that recognizes and removes small alkyl adducts
of arguments against this idea, such as obser-        from DNA. In mammalian cells, key repair
vations that there is not necessarily an increased    mechanisms are base excision repair, nucleotide
mutation rate in cancer cells over that of normal     excision repair, transcription-coupled repair,
cells123 and that a similar ‘‘evolution’’ of genet-   homologous recombination and end joining, and
ically altered cancer cells could arise by clonal     mismatch repair.128
selection followed by clonal expansion of cells          Excision repair is the most general DNA repair
with a genetic alteration that provides a prolif-     mechanism in higher organisms. Base excision re-
erative advantage.124                                 pair removes damage such as deaminated bases,
CAUSES OF CANCER                                                                                      49

oxidized or ring-opened bases generated by           correcting repair defects in humans. The RAD2
hydroxyl or superoxide radicals, and abnormally      gene product has been shown to have the ability
methylated bases such as 3-methyladenine.126         to act as a single-stranded DNA endonuclease,
Nucleotide excision repair requires sequential       directly implicating the RAD2 gene and its XPG
steps of (1) preincision recognition of damage;      human homologue as an important component
(2) incision of the damaged DNA strand at or         in the incision of a damaged DNA strand during
near the damaged site; (3) excision of the dam-      excision repair.131 Two other human DNA re-
aged site and local removal of nucleotides in        pair defect diseases, Cockayne’s syndrome (CS)
both directions from the defect in the affected      and PIBIDS, which is a photosensitive form
DNA strand; (4) repair replication to replace        of the brittle-hair disease trichothiodystrophy
the excised region, using the undamaged strand       (TTD), also have genetic defects that may cor-
as a template; and (5) ligation to join the re-      relate to genes in yeast. It is peculiar, however,
paired sequence of nucleotides at its 30 end to      that patients with PIBIDS have all of the
the contiguous DNA strand.125                        symptoms of CS as well as some of those of
   DNA repair is usually very accurate, but if       TTD, and patients with CS and PIBIDS do not
repair cannot occur prior to or during DNA           appear to have a higher than expected incidence
replication it may be error prone. This error-       of cancer. This is one of the curious examples of
prone, post-replication repair seems to be           a remarkable clinical heterogeneity among pa-
brought into play by certain types of agents or      tients with the same apparent genetic defects,
when a cell is overwhelmed by damage that it         that is, the same mutation in different individ-
cannot handle by excision repair before the cell     uals giving rise to different clinical syndromes.
enters S phase during the next round of cell         Such heterogeneity has been observed in cystic
division. In this case, the new DNA is synthe-       fibrosis as well. In the case of NER defects,
sized on templates that still contain damaged        mutations in one gene may give rise to symp-
bases, leading to mispairing or recombinational      toms of XP, combined XP and CS, or PI-
events that transfer damaged bases to daughter       BIDS.130 These kinds of results indicate that
strands. For example, in mammalian cells, 5% to      identifying mutations is only a first step in un-
30% of UV-induced thymidine dimers are               derstanding the mechanism of the disease pro-
transferred from parental to daughter strands        cess. A mutant gene may act differently in one
during postreplication repair.129                    cell type than in another cell type, depending on
   Nucleotide excision repair (NER) of DNA in        its interaction with cell-specific transcription
eukaryotic cells requires several gene products.     factors or cis-regulatory elements, interaction of
Some of these gene products appear to be iden-       its gene product with other gene products ex-
tical or highly homologous in yeast, rodents,        pressed at different levels in different cell types,
and humans.130,131 A number of defects in the        or post-transcriptional and post-translational
NER system have been found by studying mu-           mechanisms in different cell types that regulate
tations in cells from patients with xeroderma        the expression and /or function of the protein
pigmentosum, in whom at least nine different         coded for by the mutant gene in question. One
kinds of mutations (i.e., nine different comple-     candidate for this latter point is the way protein
mentation groups) have been found.125 Some of        folding is regulated. Proteins must fold into
these XP genes have been cloned and found to         particular conformations to have biological ac-
be highly homologous to yeast RAD genes that         tivity, and some mutations observed in human
are required for excision repair in Saccharo-        genetic diseases appear to affect protein folding
myces cerevisiae.130–133 Some of the cloned hu-      and intracellular translocations processes. Such
man genes also correct repair defects in mutant      defects are seen in certain forms of cystic fi-
rodent cells and are called excision repair cross-   brosis, a1-anti-trypsin deficiency, and certain gly-
complementing (ERCC) genes.                          cogen storage diseases.
   The RAD2 gene in S. cerevisiae shares remark-        It has become increasingly important to un-
able sequence homology to the XP-G gene of           derstand the mechanisms of DNA repair and
the G complementation group of xeroderma,            the defects associated with repair processes,
which is the same as the ERCC 5 gene for             because gene mutations identified in association
50                                                                                                              CANCER BIOLOGY

with hereditary forms of various cancers have                         factor TFIIH) and RNA polymerase II.130 For
been found to be mutations in DNA repair en-                          example, 25 or more proteins participate in the
zymes (Table 2–3). This intriguing discovery                          machinery involved in nucleotide excision repair.
highlights the importance of DNA repair pro-                          These are assembled in step-wise fashion at the
cesses in maintaining the integrity of the human                      site of the lesion and then disassembled after the
genome and in protecting the organism from                            repair event is completed.128 A second point is
genetic alterations that can lead to cancer. It                       that actively transcribed genes are repaired more
also highlights the importance of studying basic                      rapidly than inactive genes. For example, repair
biochemical and genetic functions in lower                            of lesions in the active dihydrofolate reductase
phylogenetic systems, because often that is where                     (DHFR) gene that are induced by UV damage
the important functions of genes are first dis-                        or alkylating agents is done much more rapidly
covered. The discovery of the human hereditary                        and completely than overall genome repair in
non-polyposis colon cancer gene is a case in                          the same cells.125 Moreover, actively transcribed
point. Without the knowledge of how DNA                               proto-oncogenes or tumor suppressor genes, al-
mismatch repair genes work in yeast and                               though perhaps more subject to damage by car-
bacteria and a somewhat serendipitous finding                          cinogens, are also avidly repaired on the tran-
based on a gene bank search, identification of                         scribing strand. Thus mutations may accumulate
the function of the colon cancer genes as mu-                         on the non-transcribed strand. Mutations in
tated DNA repair genes may have taken many                            proto-oncogenes that are not actively transcribed
months or years longer.134                                            in a given cell type may not become evident until
   Two other points about DNA repair systems                          such time as they become expressed. This delay
should be made here. One is that the DNA dam-                         might explain, in part, the long latency time of
age recognition and repair complex contains pro-                      certain forms of human cancer.
teins that function in concert with transcription                        The last repair mechanism mentioned above,
factors (e.g., the repair protein complex XPBC/                       namely O6-alkylguanine-DNA alkyltransferase
ERCC3 interacts with the basal transcription                          (AGT), is important for the repair of alkyl

Table 2–3. Human Syndromes with Defective Genome Maintenance
                                  Affected Maintenance               Main Type of Genome
Syndrome                          Mechanism                          Instability                           Major Cancer Predisposition
Xeroderma pigmentosum             NER (±TCR)                         Point mutations                       UV-induced skin cancer
Cockayne syndrome                 TCR                                Point mutations                       None*
Trichothiodystrophy               NER, TCR                           Point mutations                       None*
Ataxia telangiectasia (AT)        DSB response and repair            Chromosome aberrations                Lymphomas
AT-like disorder                  DSB response and repair            Chromosome aberrations                Lymphomas
Nijmegen breakage                 DSB response and repair            Chromosome aberrations                Lymphomas
BRCA 1 and BRCA 2                 HR                                 Chromosome       aberrations          Breast (ovarian) cancer
Werner syndrome                   HR? TLS?                           Chromosome       aberrations          Various cancers
Bloom syndrome                    HR?                                Chromosome       aberrations          Leukemia, lymphoma,
                                                                       (SCE:)                                others
Rothmund-Thomson                  HR?                                Chromosome       aberrations          Osteosarcoma
Ligase IV deficiency{              EJ                                 Recombination fidelity                 Leukemia(?)
HNPCC                             MMR                                Point mutations                       Colorectal cancer
Xeroderma pigmentosum             TLS{                               Point mutations                       UV-induced skin cancer
*Defect in transcription-coupled repair triggers apoptosis, which may protect against UV-induced cancer.
One patient with leukemia and radiosensitivity was described with active-site mutation in ligase IV.
Specific defect in relatively error-free bypass replication of UV-induced cyclobutane pyrimidine dimmers.
Abbreviations: BER, base-excision repair; DSB, double-strand break; EJ, end joining; HNPCC, hereditary non-polyposis colorectal cancer;
HR, homologous recombination; MMR, mismatch repair; NER, nucleotide-excision repair; SCE, sister-chromatid exchange; TCR,
transcription-coupled repair; TLS, translesion synthesis; UV, ultraviolet. (From Heijmakers128)
(Reprinted by permission from Macmillan Publishers Ltd. Nature 411:366–374)
CAUSES OF CANCER                                                                                       51

adducts in chemical carcinogen–damaged               1911, Rous138 induced sarcomas in chickens by
DNA.127,135 AGT is both an alkyltransferase and      filtrates obtained by passing tumor extracts
an alkyl acceptor protein that transfers, for ex-    through filters that were impermeable to cells
ample, a methyl group from O6-methylguanine to       and bacteria. These findings remained dormant
an internal cysteine, forming S-methylcysteine       for two decades until Shope showed, in 1933,
in the protein and regenerating unalkylated G        that the common cutaneous papillomas of wild
in the DNA strand. This transferase can repair       rabbits in Kansas and Iowa were caused by a
O6-methyl G and O4-methyl T and is inhibited         filterable agent.139 It was later found that when
by O6-benzylguanine.136 This inhibitory effect       these tumors were transplanted subcutaneously
has been employed to show the importance             they became invasive squamous cell carcino-
of O6-alkylations in carcinogen-induced DNA          mas.140 In 1934, Lucke observed that kidney
damage and mutations and to augment the              carcinomas commonly found in frogs in New
cytoxicity of alkylating anticancer drugs such       England lakes could be transmitted by lyophi-
as bischloroethylnitrosourea (BCNU). AGT is          lized cell-free extracts.141 Two years later, Bittner
most active in the repair of smaller alkyl ad-       demonstrated the transmission of mouse mam-
ducts; for example, O6-methyl G is repaired          mary carcinoma through the milk of mothers to
about three times faster than O6-ethyl G, and it     offspring.142 This was the first documented ex-
is likely that even larger adducts are repaired      ample of transmission of a tumor-inducing virus
primarily by excision repair.127 The importance      from one generation to another.
of AGT in preventing chemical carcinogenesis            Drawing on the experiments of Bittner, Gross
in vivo has been shown by the prevention of N-       postulated that mouse leukemia was also caused
methyl-N-nitrosourea-induced thymic lympho-          by a virus and that occurrence of the disease
mas in transgenic mice bearing the human AGT         in successive generations of mice was due to
gene.135                                             transmission of virus from parents to offspring.143
                                                     The proof of this hypothesis eluded Gross for
                                                     a number of years until he was prompted, by
VIRAL CARCINOGENESIS                                 evidence based on transmission of Coxsackie
                                                     viruses to newborn mice, to attempt inoculation
                                                     of mice less than 48 hours old. Using this ap-
Historical Perspectives
                                                     proach, he successfully transmitted mouse leu-
It has long been suspected that various forms        kemia by injecting filtered extracts prepared from
of cancer, particularly certain lymphomas and        organs of inbred AK or C58 mice, which have a
leukemias, are caused or at least ‘‘co-caused’’ by   high incidence of ‘‘spontaneous’’ leukemia, or
transmissible viruses. This theory has had its ups   from embryos of these mice, into newborn C3H
and downs during the first half of this century,      mice, which have a very low incidence of leu-
and it was not generally accepted until the 1950s    kemia. These experiments demonstrated for the
that viruses can cause malignant tumors in ani-      first time that mouse leukemia is caused by a
mals. The known carcinogenic effects of certain      virus and that the virus is transmitted in its la-
chemicals, irradiation, chronic irritation, and      tent form through embryos. This led to the
hormones did not fit with the idea of an infec-       isolation of a mouse leukemia virus.144 The
tious origin of cancer. In early experiments, the    isolated virus was also found to induce leuke-
basic assay to determine whether cancer could        mias and lymphomas in inbred strains of mice.
be induced by a transmissible agent involved         Electron-microscopic studies145 showed that
transmititng malignant disease by inoculation of     the mouse leukemia virus is spheroid, has a di-
filtered extracts prepared from diseased tissues.     ameter of about 100 nm, and contains a dense,
If the disease occurred in animals inoculated        centrally located ‘‘nucleus’’ separated from the
with such filtrates, it was assumed to be caused      external envelope by a clear circular zone. The
by a virus. In 1908, Ellermann and Bang137           Gross mouse leukemia virus was classified as a
transmitted chicken leukemia by cell-free, fil-       type C virus, a term now used to describe a wide
tered extracts and thus were among the first to       variety of RNA-containing oncogenic viruses of
demonstrate the viral etiology of this disease. In   similar morphology.
52                                                                                 CANCER BIOLOGY

    The RNA oncoviruses have been classified by       a silent form. The widespread distribution of
morphological criteria. Intracytoplasmic type A      gene sequences homologous to those of oncov-
particles were initially observed in early embryos   iruses throughout the animal kingdom suggests
of mice and in certain murine tumors. These A        that these sequences are evolutionarily very
particles are noninfectious, bud into intracellu-    old. One of the most important observations of
lar membranes rather than through the plasma         Bittner and Gross was that an oncogenic virus
membrane, and thus stay within the cell. They        could be transmitted vertically from parent to
have an active reverse transcriptase and exist as    offspring. In the case of mouse leukemia, it be-
a proviral form in chromosomal DNA. Type B           came apparent that a mouse born to AK or C58
viruses have spikes on their outer envelope, bud     parents receives at birth the genetic information
from cells, and have been identified primarily in     for malignant disease. These findings suggest
murine species, mouse mammary tumor virus            that later in the life of a carrier animal, the ex-
(MMTV) being an example. Type C viruses have         pression of virally coded genes, perhaps trig-
been found widely distributed among birds and        gered by exposure to chemicals, irradiation,
mammals, can induce leukemias, sarcomas, and         hormone imbalance, or chronic irritation, be-
other tumors in various species, and have cer-       comes activated, causing leukemia. In the case
tain gene sequences that are homologous to           of the mouse (and perhaps of humans), the ac-
‘‘transforming’’ sequences isolated from various     tivation of latent oncogenic viruses may not oc-
human tumors (see below). Another subgroup,          cur during the life span of the carrier animal,
type D RNA oncoviruses, has been isolated            and the animal may remain disease-free, even
from primate species but their oncogenic po-         though it carries and transmits the viral genome
tential is not well established. The subtypes of     to its progeny. The way in which this type of
RNA tumor viruses, known as Retroviridiae,           vertical transmission could occur is explained
share a genetically related genome containing a      by later findings that the genome of onco-
gag-pol-env gene sequence coding for virus in-       genic viruses is integrated into the host cells’
ternal structural proteins, the special type of      genome prior to cellular transformation (see
RNA-directed DNA polymerase called reverse           Chapter 5).
transcriptase and viral envelope proteins, re-          Another oncogenic virus was discovered in
spectively. Thus, they most likely share a com-      the tissue extracts of leukemic AK mice after it
mon evolutionary heritage.146 However, distinct      was noted that when newborn C3H mice were
subclasses of retrovirus evolution, based on pol     injected with such extracts, some of the C3H
gene sequence homologies, have been found;           mice developed parotid gland tumors rather
one major pathway gives rise to mammalian type       than leukemia.143 Some of these mice also de-
C viruses and a second to A, B, D, and avian         veloped cutaneous sarcomas, mammary carcino-
type C oncoviruses.146 A more recent addition        mas, and other malignancies. Eddy, Stewart, and
to the retrovirus classification is the human T-      colleagues found that virus produced by cul-
cell leukemia virus (HTLV), isolated from pa-        tured cells after infection with extracts from
tients with certain forms of adult T-cell leuke-     leukemic AK mice organs caused parotid gland
mias (discussed later). The pol gene of HTLV         tumors and a variety of other neoplasms in mice,
appears to have evolved from a progenitor            hamsters, and rats. They suggested that this
common to the types A, B, D, and avian C on-         multiple-tumorigenic agent be called polyoma
coviruses rather than from the mammalian C           virus.147
type.146 If true, this would be unusual because         Since the early pioneering work in this field, a
most mammalian type C viruses share antigenic        number of other oncogenic viruses have been
determinants among several gag, pol, and env         identified and characterized. These include (1)
gene products, suggesting a common progenitor        feline leukemia virus, shown by inoculation of
for this subclass of retroviruses.                   cell-free extracts from leukemic cats into new-
    Unlike most infectious viruses, oncoviruses      born kittens; (2) SV40 virus, shown to be latent
can be transmitted through the germline of           and harmless in the rhesus monkey but to in-
animal species, and thus these viral genes can be    duce leukemias and sarcomas after inoculation
passed from one generation to the next, often in     into newborn hamsters; (3) adenoviruses, which
CAUSES OF CANCER                                                                                                           53

cause the common cold in humans and induce              Table 2–4. Examples of Oncogenic Viruses
sarcomas in newborn hamsters and rats; and (4)          Virus                                        Species of Isolation
such herpesviruses as Herpes saimiri, which is
                                                        I. Oncogenic RNA viruses
indigenous in the New World squirrel monkey
                                                             A. Acute-acting type*
and may induce lymphosarcomas and leukemias                     Rous sarcoma                         Chickens
when inoculated into certain species of mon-                    Fujinami sarcoma                     Chickens
keys. Table 2–4 lists some of the different types               Retculoendotheliosis                 Chickens and turkeys
                                                                Avian erythroblastosis               Chickens
of RNA and DNA oncogenic viruses.                               Avian myeloblastosis                 Chickens
                                                                Avian myelocytomatosis               Chickens
                                                                Moloney sarcoma                      Mice
Role of Viruses in the Causation                                Abelson leukemia                     Mice
of Human Cancer                                                 FBJ osteosarcoma                     Mice
                                                                Harvey/Kirsten sarcoma{              Rat
To prove a causal relationship between a putative               Rat sarcoma                          Rat
                                                                Feline sarcoma                       Cat
cancer-causing virus and human cancer is not a                  Woolly monkey sarcoma                Woolly monkey
simple task. Such proof relies on evidence that is           B. Chronic type{
to a fair extent circumstantial. This evidence in-              Avian leucosis                       Chickens
                                                                Mouse leukemia**                     Mice
cludes (1) epidemiological data showing a cor-                  Feline leukemia                      Cat
relation between living in an area of endemic                   Bovine leukemia                      Cow
viral infection and a type of cancer; (2) serological           Gibbon ape leukemia{{                Gibbon ape
                                                                Mouse mammary tumor                  Mice
evidence of antibody titers to viral antigens in                Human T lymphotropic                 Human
patients with a given cancer type; (3) evidence for               viruses
insertion of viral DNA into a cancer-bearing            II. Oncogenic DNA viruses
host’s cell genome; (4) evidence for a consistent            A. Papovaviruses                        Rabbit, man, dog,
chromosomal translocation, particularly those in-                                                    cow, and others
volving an oncogene, in virally infected patients;              Papilloma                            Mouse
                                                                Polyoma                              Monkey
(5) data showing that viral infection of cells in               SV40                                 Human
culture or transfection of viral genes into cells               JC                                   Human
causes cell transformation and the ability of such              BK
                                                             B. Adenoviruses                         Human,{{ monkey,
cells to produce tumors in nude mice; and (6)                                                          birds, cow
development of cancers of the suspected target               C. Herpes viruses
organ in transgenic mice produced by embryonic                  Epstein-Barr                         Human
                                                                Lucke carcinoma                      Frog
gene transfer of viral genes.                                   Marek’s disease                      Chicken
   On the basis of this sort of evidence, some
                                                        *These viruses are acute, transform cells in vitro, have rapid disease
human cancers are considered to be caused by            induction in vivo, and carry ‘‘transforming onc gene’’ related to cell
viral infection either directly or indirectly. By       gene. Most are replication defective but can be isolated free of helper
‘‘directly,’’ I mean that the viral gene(s) can         {
                                                         These hybrid viruses, created experimentally, contain mouse helper
themselves cause cells to become malignant              virus and rat ‘‘src’’ sequences.
(sometimes also requiring the loss of a tumor           {
                                                         These viruses are chronic, have no transformation in vitro, long la-
suppressor gene). By ‘‘indirectly,’’ I mean that        tency period in vivo, and no evidence of transforming gene.
viral infection may simply cause the progression        These all appear to be horizontally transmitted; in some cases, re-
                                                        lated sequences are found in cell DNA.
of malignant cell growth by producing an                **The Friend leukemia virus complex contains a defective genome,
immunodeficiency state (e.g., the occurrence of          codes for a small envelope glycoprotein not incorporated into virions,
                                                        does not transform cells in vitro, and perhaps should be placed in a
non-Hodgkin’s lymphoma in HIV-infected pa-              separate category.
tients) or by stimulating the proliferation of al-      {{
                                                          Viruses show a distant relationship to mouse DNA, but not that of
ready transformed cells. Sometimes viral infec-         primates; this indicates ‘‘ancient’’ horizontal transmission.
tion acts in concert with other infectious agents        There are 31 members of the human adenovirus group and at least
                                                        12 induce tumors in newborn animals and/or transform cells in vitro.
or chemical carcinogens. Such is the case for
malarial infection of Epstein-Barr virus (EBV)–
infected patients and for aflatoxin exposure of          human cancer thought to be caused by viral
individuals bearing the hepatitis B viral genome        infection and the strength of epidemiological
in their liver cells (see below). The types of          associations are shown in Table 2–5.148
54                                                                                                     CANCER BIOLOGY

         Table 2–5. Human Cancer Viruses and Associated Cancers, Strength of Association, and
         Necessary Preconditions
                                                         Strength of Epidemiologic
         Virus         Cancer                            Association                        Required Precondition
         HBV           Hepatocellular carcinoma          Strong                             None
         HTLV-I        T-cell lymphoma                   Strong                             None
         EBV           Burkitt’s lymphoma                Strong                             Chronic malaria
         EBV           High-grade lymphoma               Strong                             HIV
         HPV           Cervical cancer                   Consistent                         ?None
         EBV           Nasopharyngial carcinoma          Inconsistent                       —
         HSV-2         Cervical cancer                   Inconsistent                       —
         EBV, Epstein-Barr virus; HBV, hepatitis B virus; HSV-2, herpes simplex virus 2; HTLV-1, human T-lymphotropic
         virus 1. (From Henderson148)

     Association of Epstein-Barr Virus                          though the exact role of EBV remains to be
     and Human Cancers                                          elucidated.149
Epstein-Barr virus has been linked to four dif-
ferent types of human cancer: Burkitt’s lym-
                                                                        Hepatitis Virus and Hepatocellular
phoma (BL), nasopharyngeal carcinoma (NPC),
B-cell lymphomas in immunosuppressed indi-
viduals such as HIV-infected patients, and some                 Epidemiological evidence strongly points to a
cases of Hodgkin’s lymphoma.149 The evidence                    link between chronic hepatitis B virus (HBV)
is strongest for an association with BL and NPC.                infection and hepatocellular carcinoma (HCC).
   Infection with EBV does not by itself cause                  In areas where HBV infection is endemic, such
cancer. On average, across the world, about 90%                 as Taiwan, Senegal, South Africa, Hong Kong,
of the population may be infected by the time                   China, and the Philippines, the incidence of
they reach adulthood. In some endemic areas,                    HCC is much higher than in countries where
the incidence rate approaches 100%. In devel-                   HBV infection is less common (reviewed in Ref-
oping countries, EBV infection often occurs in                  erence 149). Hepatocellular carcinoma usually
young childhood. In more affluent societies,                     appears after decades of chronic liver involve-
EBV infection tends to occur as the ‘‘kissing age’’             ment due to HBV-induced liver cell damage and
of adolescence or young adulthood is reached,                   regeneration. HBV DNA can be found in the
and manifests itself as infectious mononucleosis.               majority of liver cancers from patients in high-
In developing countries, particularly in equa-                  risk areas and a specific piece of HBV DNA
torial Africa, concomitant or subsequent infec-                 called HBx, which encodes a transcription fac-
tion with the malarial parasite induces B-cell                  tor, is found in HCC cells and can induce liver
proliferation and an immunodeficiency state that                 tumors in transgenic mice.150 Some chromo-
leads to malignant transformation and progres-                  somal modifications are also observed in HCC.
sion. There is a consistent chromosomal translo-                These include alterations of the short arm of
cation involving immunoglobulin genes, usually                  chromosome 11, deletions in the long arm of
on chromosome 14, and sequences within or                       chromosome 13, and point mutations of the p53
adjacent to the c-myc gene locus on chromo-                     gene on chromosome 17. These latter mutations
some 8 (see Chapters 5 and 7).                                  are particularly interesting because such muta-
   The role of EBV in NPC is less well charac-                  tions are seen in areas where there is concomi-
terized, but the evidence for an association in-                tant exposure to aflatoxins in foods, suggesting a
cludes high serum antibody titers against EBV                   joint role of HBV infection and chemical car-
antigens and the presence of EBV DNA in NPC                     cinogens causing HCC.
cells. Similar evidence suggests an association                    There is also some evidence that chronic in-
between EBV infection and induction of some                     fection with hepatitis C virus, which may cause
B-cell lymphomas and some Hodgkin’s disease                     chronic liver injury and regeneration, may be
cases in immunosuppressed individuals, al-                      a causative agent for HCC and be a factor
CAUSES OF CANCER                                                                                    55

particularly in HBV-negative cases. Currently        from patients with adult T-cell leukemia-
there are a number of countries in which wide-       lymphoma (ATLL), named HTLV-I, were found
spread vaccination against HBV is occurring,         to have several characteristics in common. In
and time will tell whether such vaccination          1982, a new subgroup of HTLV, called HTLV-
lowers the incidence of HCC in these areas.          II, was identified.155 The DNA sequence of the
                                                     two subgroups is clearly different, but there is
                                                     significant homology. Moreover, the two types
     Papillomaviruses and
                                                     of virus genomes encode a very similar p24 core
     Cervical Cancer
                                                     protein and share a common mechanism of gene
A large class of papillomaviruses that are patho-    activation, indicating that they are members
genic for humans (HPV) have been identified.          of the same family. HTLV-I and -II are retro-
More than 60 genotypic subtypes have been            viruses with reverse transcriptase and an RNA
isolated.148 Only two HPV subtypes have been         genome of about 9 kilobases. DNA sequences
closely associated with cervical cancer, HPV 16      homologous to HTLV DNA are not found in
and HPV 18. The evidence for this association is     the genome of normal human cells; thus, they
the following (reviewed in Reference 149): (1)       are exogenous, not endogenous, genomic se-
viral DNA is found in about 90% of cervical          quences, in contrast to the c-onc genes that
cancers; (2) in most cases a specific piece of the    are homologous to v-onc genes of retroviruses.
viral DNA is integrated into the host’s genome;      HTLVs do not appear to carry their own onc
(3) the vast majority of all HPV-positive cervical   gene. In this respect they appear to be similar to
cancers contain cells that express two specific       the chronic-acting retroviruses such as avial
gene transcripts, E6 and E7; (4) E6 and E7           leukosis virus, and they have an LTR-gag-pol-
genes of high-risk HPV (e.g., HPV 16 and 18),        env-LTR gene arrangement typical of other
but not of low-risk HPV subtypes, immortal-          retroviruses. HTLV-I and -II can transform
ize human cells in culture; (5) E6- and E7-          normal human T cells in culture and the trans-
expressing cells frequently undergo a progres-       formed cells contain at least one proviral DNA
sion to aneuploidy and gene amplification in          copy, transcribe viral DNA, and make low levels
culture; (6) the E6 and E7 oncoproteins bind         of viral proteins.155
to and inactivate or degrade the p53 and RB             In addition to the genes common to all re-
tumor suppressor gene proteins, respectively;        troviruses, the HTLV-I genome contains a 1.6
and (7) uterine cervical dysplasia can be induced    kilobase sequence at the 30 terminal region that
in mice by inoculation of a recombinant retro-       encodes at least two trans-acting regulator pro-
virus bearing the E6 and E7 genes of HPV 16          teins: a 40,000 Dalton protein product of a gene
into the vagina of mice.151                          called tax and a 27,000 Dalton protein product
                                                     of a gene called rex. Transgenic mice bearing
                                                     the tax gene develop multiple mesenchymal
     HTLV-I and Adult T-Cell Leukemia
                                                     tumors at about 3 months of age.156
The retrovirus HTLV-I was first identified in             A high percentage of patients with ATLL
interleukin-2 (IL-2)-stimulated T lymphocytes        and certain T-cell lymphomas have antigens to
from two patients, one with a T-cell lymphoma        HTLV-I proteins in their serum, and their tumor
and the other with T-cell leukemia.152,153           cells contain one or more copies of the HTLV
This was the first demonstration that a retrovirus    genome.154,157,158 But patients with childhood
could cause malignancy in humans, although it        cancers, non–T-cell leukemias and lymphomas,
had long been suspected that retroviruses might      myeloid leukemias, Hodgkin’s disease, and solid
do so, based on numerous examples in animals.        tumors do not have evidence of HTLV anti-
   Since the original isolation of a human T-cell    bodies in their serum, and healthy individuals in
leukemia and lymphoma virus, several similar         nonendemic areas are also antibody negative.157
isolates have been made from patients with T-        However, almost 50% of relatives of ATLL pa-
cell neoplasms in different parts of the world,      tients and about 12% of healthy blood donors in
including Japan, Africa, the Caribbean basin,        endemic areas have been reported to be anti-
England, and the Netherlands.154 The isolates        body positive. These data indicate the T-cell
56                                                                                   CANCER BIOLOGY

specificity of the neoplastic transformation pro-     tion of African Old World primates with HTLV-
cess induced by HTLV and suggest horizontal          like viruses, it has been proposed that the origin
spread of the virus among people. Major clus-        of HTLV was in Africa and that spread to other
ters of HTLV-related T-cell cancers are found in     countries may have occurred by means of ex-
areas of high endemic infection with HTLV, such      plorers who introduced infected primates or had
as southwestern Japan, in the Caribbean basin,       contact with infected primates in Africa.159
and in certain areas of South America and               In conclusion, it is fair to say that infections
Africa.154 As is the case with Burkitt’s lymphoma,   with oncogenic viruses are clearly associated
however, only certain people infected with the       with certain kinds of human cancer. However,
virus get ATLL. Thus, other predisposing fac-        even in those cases in which viral infection ap-
tors must exist.                                     pears to be a predisposing factor, viral infection
   Horizontal transmission among individuals         itself is insufficient to cause cancer. In all cases,
appears to require prolonged and intimate con-       there are other contributing factors, which in-
tact with an HTLV-positive person. Cell-to-cell      clude cell type–specific mitogenic stimulation,
transmission can also be demonstrated in cell        suppression of the immune response, and, pos-
culture systems when HTLV-producing cells are        sibly, genetic factors. It is also clear, though, that
cocultured with normal T lymphocytes.154 The         a combination of infection with certain onco-
HTLV-infected, transformed cells often pro-          genic viruses, chronic mitogenic stimulation of
duce infectious HTLV, but virus-nonproducing         the virus-infected cells, and a concomitant im-
transformed cells, containing the integrated viral   mune deficiency state have a high propensity to
genome as a provirus, are also observed.             induce the cancerous process in human beings.
   Only about 1 out of 25 to 30 infected individ-       It should be noted that the human immuno-
uals will eventually develop ATLL, and HTLV-I        deficiency virus (HIV) that causes AIDS is also a
DNA is consistently demonstrated in the ATL          slow infectious retrovirus virus (lentivirus) that
cells from these patients. There appears to be a     is a T-cell lymphotropic virus with some geno-
latency period between primary infection and         mic similarities to HTLV. HIV infects CD4þ
leukemia development of several decades.149          cells and causes disease by its immunosuppres-
   From these findings, the following question        sive effects. Patients with AIDS are at high risk
obviously arises: Where did these viruses come       to develop Kaposi’s sarcoma and non-Hodgkin’s
from and are they a recently evolved class of        lymphoma. These cancers most likely arise be-
viruses? Although, there is no definitive evi-        cause of the immunosuppressive effects of the
dence for HTLV infections in humans before           AIDS virus rather than the direct transforming
the late 1970s, HTLV viruses have probably           activity of the virus, although some direct cell-
been around for a long time, perhaps hundreds        transforming effects have been observed.
or thousands of years. Most likely, there was an
animal vector originally, and recently the virus
may have undergone some evolutionary change          References
that made it more infectious or more cytopathic        1. E. G. Luebeck and S. H. Moolgavkar: Multistage
for humans. Analysis of serum samples from                carcinogenesis and the incidence of colorectal
subhuman primates, for example, shows that                cancer. Proc Natl Acad Sci USA 99:15095, 2002.
several Old World monkey species, including            2. P. C. Nowell and D. A. Hungerford: A minute
Japanese and Chinese macaques, African green              chromosome in human chronic granulocytic leu-
                                                          kemia. Science 132:1497, 1960.
monkeys, and baboons, are seropositive for             3. J. D. Rowley: A new consistent chromosomal
HTLV.159 In Japan, where HTLV is endemic in               abnormality in chronic myelogenous leukemia
humans in certain areas, the distribution of virus        identified by quinacrine fluorescence and Giemsa
in primates is much more widespread, suggest-             staining. Nature 243:290, 1973.
ing independent entry of HTLV-like virus into          4. A. G. Knudson: Two genetic hits (more or less)
                                                          to cancer. Nat Rev Cancer 1:157, 2001.
the primate and human populations and arguing          5. C. Lengaur, K. W. Kinzler, and B. Vogelstein:
against current transmission between primates             Genetic instabilities in human cancers. Nature
and humans. Because of the widespread infec-              396:643, 1998.
CAUSES OF CANCER                                                                                         57

 6. L. A. Loeb: Mutator phenotype may be required     21. L. A. Loeb: Apurinic sites as mutagenic interme-
    for multistage carcinogenesis. Cancer Res 51:         diates. Cell 40:483, 1985.
    3075, 1991.                                       22. T. C. Boles and M. E. Hogan: Site-specific
 7. A. Luch: Nature and nuture—Lessons from               carcinogen binding to DNA. Proc Natl Acad Sci
    chemical carcinogenesis. Nat Rev Cancer 5:            USA 81:5623, 1984.
    113, 2005.                                        23. P. D. Kurth and M. Bustin: Site-specific carcin-
 8. P. D. Lawley: Carcinogenesis by alkylating            ogen binding to DNA in polytene chromo-
    agents. In C. E. Searle. ed.: Chemical Carcin-        somes. Proc Natl Acad Sci USA 82:7076, 1985.
    ogens, Monograph No. 173. Washington, DC:         24. C. C. Harris: Chemical and physical carcino-
    American Chemical Society, 1976, pp. 83–244.          genesis: Advances and perspectives for the
 9. E. C. Miller: Some current perspectives on            1990s. Cancer Res 51:5023, 1991.
    chemical carcinogenesis in humans and exper-      25. I. C. Hsu, R. A. Metcalf, T. Sun, J. Welsh, N. J.
    imental animals: Presidential address. Cancer         Wang, and C. C. Harris: p53 gene mutational
    Res 38:1479, 1978.                                    hotspot in human hepatocellular carcinomas
10. E. C. Miller and J. A. Miller: Searches for           from Qidong, China. Nature 350:427, 1991.
    ultimate chemical carcinogens and their reac-     26. B. Bressac, M. Kew, J. Wands, and M. Ozturk:
    tions with cellular macromolecules. Cancer            Selective G to T mutations of p53 gene in he-
    47:2327, 1981.                                        patocellular carcinoma from southern Africa.
11. E. Boyland: The biological significance of me-         Nature 350:429, 1991.
    tabolism of polycyclic compounds. Biochem Soc     27. W. F. Friedewald and P. Rous: The initiating
    Symp 5:40, 1950.                                      and promoting elements in tumor production:
12. A. Pullman and B. Pullman: Electronic struc-          An analysis of the effects of tar, benzpyrene,
    ture and carcinogenic activity of aromatic mol-       and methylcholanthrene on rabbit skin. J Exp
    ecules. Adv Cancer Res 3:117, 1955.                   Med 80:101, 1944.
13. G. Kellerman, C. R. Shaw, and M. Luyten-          28. R. D. Sall and M. J. Shear: Studies in carcino-
    Kellerman: Aryl hydrocarbon hydroxylase in-           genesis. XII. Effect of the basic fraction of cre-
    ducibility and bronchogenic carcinoma. N Engl         osote oil on the production of tumors in mice by
    J Med 289:934, 1973.                                  chemical carcinogens. J Natl Cancer Inst 1:45,
14. T. L. McLemore, S. Adelberg, M. C. Liu, N. A.         1940.
    McMahan, S. J. Yu, W. C. Hubbard, et al.: Ex-     29. I. Berenblum: The mechanism of carcinogenesis:
    pression of CYP1A1 gene in patients with lung         A study of the significance of carcinogenic ac-
    cancer: Expression for cigarette smoke–induced        tion and related phenomena. Cancer Res 1:807,
    gene expression in normal lung tissue and for         1941.
    altered gene regulation in primary pulmonary      30. E. Hecker: Isolation and characterization of the
    carcinomas. J Natl Cancer Inst 82:1333, 1990.         cocarcinogenic principles from croton oil. Meth-
15. K. Nakachi, K. Imai, S.-I. Hayashi, and K.            ods Cancer Res 6:439, 1971.
    Kawajiri: Polymorphisms of the CYP1A1 and         31. B. L. Van Duuren: Tumor-promoting agents in
    glutathione S-transferase genes associated with       two-stage carcinogenesis. Prog Exp Tumor Res
    susceptibility to lung cancer in relation to          11:31, 1969.
    cigarette dose in a Japanese population. Cancer   32. I. Berenblum: Sequential aspects of chemical
    Res 53:2994, 1993.                                    carcinogenesis: Skin. In F. F. Becker, ed.: Can-
16. H. M. Schuller: Mechanisms of smoking-related         cer: A Comprehensive Treatise. New York: Ple-
    lung and pancreatic adenocarcinoma develop-           num Press, 1975, pp. 323–344.
    ment. Nat Rev Cancer 2:455, 2002.                 33. R. K. Boutwell: Some biological aspects of skin
17. S. A. Belinsky, T. R. Devereux, and M. W.             carcinogenesis. Prog Exp Tumor Res 4:207, 1964.
    Anderson: Role of DNA methylation in the acti-    34. H. c. Pitot and Y. P. Dragan: Facts and theories
    vation of proto-oncogenes and the induction of        concerning the mechanisms of carcinogenesis.
    pulmonary neoplasia by nitrosamines. Mutat            FASEB J 5:2280, 1991.
    Res 233:105, 1990.                                35. J. F. O’Connell, A. J. P. Klein-Szanto, D. M.
18. V. D.-H. Ding, R. Cameron, and C. B. Pickett:         DiGiovanni, J. W. Fries, and T. J. Slaga: En-
    Regulation of microsomal, xenobiotic epoxide          hanced malignant progression of mouse skin
    hydrolase messenger RNA in persistent hepa-           tumors by the free-radical generator benzoyl
    tocyte nodules and hepatomas induced by               peroxide. Cancer Res 46:2863, 1986.
    chemical carcinogens. Cancer Res 50:256, 1990.    36. J. McCann, E. Choi, E. Yamasaki, and B. N. Ames:
19. M. E. Hogan, N. Dattagupta, and J. P. Whitlock,       Detection of carcinogens as mutagens in the
    Jr.: Carcinogen-induced alteration of DNA             Salmonella/microsome test: Assay of 300 chem-
    structure. J Biol Chem 256:4504, 1981.                icals. Proc Natl Acad Sci USA 72:5135, 1975.
20. J. S. Eadie, M. Conrad, D. Toorchen, and M. D.    37. E. Huberman and L. Sachs: Cell-mediated
    Topal: Mechanism of mutagenesis by O6-                mutagenesis of mammalian cells with chemical
    methylguanine. Nature 308:201, 1984.                  carcinogens. Int J Cancer 13:326, 1974.
58                                                                                        CANCER BIOLOGY

38. E. Huberman and L. Sachs: Mutability of dif-           53. L. Foulds: The experimental study of tumor pro-
    ferent genetic loci in mammalian cells by                  gression: A review. Cancer Res 14:327, 1954.
    metabolically activated carcinogenic polycyclic        54. P. C. Nowell: The clonal evolution of tumor cell
    hydrocarbons. Proc Natl Acad Sci USA 73:188,               populations. Science 194:23, 1976.
    1976.                                                  55. D. L. Trainer and E. F. Wheelock: Character-
39. R. B. Setlow: Repair deficient human disorders              ization of L5178 Y cell phenotypes isolated dur-
    and cancer. Nature 271:713, 1978.                          ing progression of the tumor-dormant state in
40. M. Swift and C. Chase: Cancer in families in               DBA/2 mice. Cancer Res 44:2897, 1984.
    xeroderma pigmentosum. J Natl Cancer Inst              56. D. Hanahan and R. A. Weinberg: The hall-
    62:1415, 1979.                                             marks of cancer. Cell 100:57, 2000.
41. Y. Shiraishi, T. H. Yosida, and A. A. Sandberg:        57. R. Bernards and R. A. Weinberg: A progression
    Malignant transformation of Bloom syndrome                 puzzle. Nature 418:823, 2002.
    B-lymphoblastoid cell lines by carcinogens.            58. M. J. van de Vijver, Y. D. He, L. J. van’t Veer,
    Proc Natl Acad Sci USA 82:5102, 1985.                      H. Dai, et al.: A gene-expression signature as a
42. H. Kirkman: Estrogen-induced tumors of the                 predictor of survival in breast cancer. N Engl
    kidney in the Syrian hamster. National Cancer              J Med 347:1999, 2002.
    Institute Monograph No. 1, 1959, p. 137.               59. M. Al-Hajj, M. S. Wicha, A. Benito-Hernandez,
43. M. S. Biskind and G. S. Biskind: Development               S. J. Morrison, and M. F. Clarke: Prospective
    of tumors in the rat ovary after transplantation           identification of tumorigenic breast cancer
    into the spleen. Proc Soc Exp Biol Med 55:176,             cells. Proc Natl Acad Sci USA 100:3983, 2003.
    1944.                                                  60. D. B. Rifkin, R. M. Crowe, and R. Pollack:
44. T. Tsutsui, G. H. Degan, D. Schiffman, A.                  Tumor promoters induce changes in the chick
    Wong, H. Maizumi, J. A. McLachlan, and J. C.               embryo fibroblast cytoskeleton. Cell 18:361,
    Barrett: Dependence on exogenous metabolic                 1979.
    activation for induction of unscheduled DNA            61. P. B. Fisher, J. H. Bozzone, and I. B. Weinstein:
    synthesis in Syrian hamster embryo cells by di-            Tumor promoters and epidermal growth factor
    ethylstilbestrol and related compounds. Cancer             stimulate anchorage-independent growth of
    Res 44:184, 1984.                                          adenovirus-transformed rat embryo cell. Cell
45. S. K. Buenaventura, D. Jacobson-Kram, K. L.                18:695, 1979.
    Dearfield, and J. R. Williams: Induction of sister      62. R. Cohen, M. Pacifici, N. Rubernstein, J. Biehl,
    chromatid exchange by diethylstilbestrol in                and H. Holtzer: Effect of tumor promoter on
    metabolically competent hepatoma cells but                 myogenesis. Nature 226:538, 1977.
    not in b/fibroblasts. Cancer Res 44:3851, 1984.         63. M. Lowe, M. Pacifici, and H. Holtzer: Effects
46. E. R. Fearon and B. Vogelstein: A genetic model            of phorbol-12-myristate-13-acetate on the phe-
    for colorectal tumorigenesis. Cell 61:759, 1990.           notypic program of cultured chrondroblasts and
47. B. S. Strauss: The origin of point mutations in            fibroblasts. Cancer Res 38:2350, 1978.
    human tumor cells. Cancer Res 52:249, 1992.            64. L. Diamond, T. O’Brien, and A. Rovera: Inhi-
48. L. A. Loeb: Endogenous carcinogenesis: Molec-              bition of adipose conversion of 3Y3 fibroblasts
    ular oncology into the twenty-first century—                by tumor promoters. Nature 269:247, 1977.
    Presidential address. Cancer Res 49:5489, 1989.        65. E. Fibach, R. Gambari, P. A. Shaw, G. Maniatis,
49. C. Richter, J-W. Park, and B. N. Ames: Normal              R. C. Reuben, S. Sassa, R. A. Rifkind, and P. A.
    oxidative damage to mitochondrial and nuclear              Marks: Tumor promoter–mediated inhibition of
    DNA is extensive. Proc Natl Acad Sci USA                   cell differentiation: Suppression of the expres-
    85:6465, 1988.                                             sion of erythroid functions in murine erythro-
50. T. A. Kunkel and L. A. Loeb: Fidelity of mam-              leukemia cells. Proc Natl Acad Sci USA 76:
    malian DNA polymerase. Science 213:765,                    1906, 1979.
    1981.                                                  66. D. Ishii, E. Fibach, H. Yamasaki, and I. B.
51. T. A. Kunkel: The mutational specificity of                 Weinstein: Tumor promoters inhibit morpho-
    DNA polymerase-b during in vitro DNA syn-                  logical differentiation in cultured mouse neu-
    thesis: production of frameshift, base substitu-           roblastoma cells. Science 200:556, 1978.
    tion and deletion mutations. J Biol Chem 260:          67. S. Mondal and C. Heidelberger: Transforma-
    5787, 1985.                                                tion of C3H/10T 1/2 CL8 mouse embryo
52. B. N. Ames, R. L. Saul, E. Schwiers, R.                    fibroblasts by ultraviolet irradiation and a phor-
    Adelman, and R. Cathcart: Oxidative DNA                    bol ester. Nature 260:710, 1976.
    damage as related to cancer and aging: The             68. N. Yamamoto and H. zur Hausen: Tumor
    assay of thymine glycol, thymidine glycol, and             promoter TPA enhances transformation of hu-
    hydroxymethyluracil in human and rat urine. In             man leukocytes by Epstein-Barr virus. Nature
    R. S. Sohal, L. S. Birnbaum, and R. G. Cutler, eds.:       280:244, 1979.
    Molecular Biology of Aging: Gene Stability and         69. L. Kopelovich, N. E. Bias, and L. Helson:
    Gene Expression. New York: Raven Press, 1985,              Tumor promoter alone induces neoplastic
    pp. 137.                                                   transformation of fibroblasts from human
CAUSES OF CANCER                                                                                               59

      genetically predisposed to cancer. Nature 282:             basement membrane (type IV) collagen-
      619, 1979.                                                 degrading metalloproteinase by human fibro-
70.   C. M. Aldaz, C. J. Conti, I. B. Gimenez, T. J.             blasts. J Biol Chem 260:8526, 1985.
      Slaga, and A. J. P. Klein-Szanto: Cutaneous          82.   R. Montesano and L. Orci: Tumor-promoting
      changes during prolonged application of 12-O-              phorbol esters induce angiogenesis in vitro. Cell
      tetradecanoylphorbol-13-acetate on mouse skin              42:469, 1985.
      and residual effects after cessation of treatment.   83.   Y. Nishizuka: The role of protein kinase C in
      Cancer Res 45:2753, 1985.                                  cell surface signal transduction and tumor pro-
71.   P. M. Rosoff, L. F. Stein, and L. C. Cantley:              duction. Nature 308:693, 1984.
      Phorbol esters induce differentiation in a pre-      84.   J. E. Niedel, L. J. Kuhn, and G. R. Vandenbark:
      B-lymphocyte cell line by enhancing Naþ/Hþ                 Phorbol diester receptor copurifies with protein
      exchange. J Biol Chem 259:7056, 1984.                      kinase C. Proc Natl Acad Sci USA 80:36, 1983.
72.   J. M. Besterman, W. S. May Jr., H. LeVine III,       85.   D. F. Birt, J. C. Pelling, L. T. White, K.
      E. J. Cragoe Jr., and P. Cuatrecasas: Amiloride            Dimitroff, and T. Barnett: Influence of diet and
      inhibits phorbol ester–stimulated Naþ/Hþ ex-               calorie restriction on the initiation and promo-
      change and protein kinase C: An amiloride                  tion of skin carcinogenesis in the SENCAR
      analog selectively inhibits Naþ/Hþ exchange.               mouse model. Cancer Res 51:1851, 1991.
      J Biol Chem 260:1155, 1985.                          86.   H. Wei, L. Tye, E. Bresnick, and D. F. Birt:
73.   I. Sussman, R. Prettyman, and T. G. O’Brien:               Inhibitory effect of apigenin, a plant flavonoid,
      Phorbol esters and gene expression: The role               on epidermal ornithine decarboxylase and skin
      of rapid changes in Kþ transport in the                    tumor promotion in mice. Cancer Res 50:499,
      induction of ornithine decarboxylase by 12-O-              1990.
      tetradecanoyl-13-acetate in BALB/c 3T3 cells         87.   A. Leder, A. Kuo, R. D. Cardiff, E. Sinn, and
      and a mutant cell line defective in NaþKþC1-               P. Leder: v-Ha-ras transgene abrogates the ini-
      cotransport. J Cell Biol 101:2316, 1985.                   tiation step in mouse skin tumorigenesis: Effects
74.   S. Grinstein, S. Cohen, J. D. Goetz, A.                    of phorbol esters and retinoic acid. Proc Natl
      Rothstein, and E. W. Gelfand: Characterization             Acad Sci USA 87:9178, 1990.
      of the activation of Naþ/Hþ exchange in lym-         88.   Z. Szallasi and P. M. Blumberg: Prostratin, a
      phocytes by phorbol esters: Change in cyto-                nonpromoting phorbol ester, inhibits induction
      plasmic pH dependence of the antiport. Proc                by phorbol 12-myristate 13-acetate of ornithine
      Natl Acad Sci USA 82:1429, 1985.                           decarboxylase, edema, and hyperplasia in CD-1
75.   C.-C. Chang, J. E. Trosko, H.-J. Kung, D.                  mouse skin. Cancer Res 51:5355, 1991.
      Bombick, and F. Matsumura: Potential role of         89.   C. Peraino, R. J. M. Fry, E. Staffeldt, and
      the src gene product in inhibition of gap-                 W. E. Kisielski: Effects of varying the exposure
      junctional communication in NIH/3T3 cells.                 to phenobarbital on its enhancement of 2-
      Proc Natl Acad Sci USA 82:5360, 1985.                      acetylaminofluorene-induced hepatic tumori-
76.   W. S. May, N. Sahyoun, S. Jacobs, M. Wolf, and             genesis in the rat. Cancer Res 33:2701, 1973.
      P. Cuatrecasas: Mechanism of phorbol diester–        90.   T. Kitagawa, H. C. Pitot, E. C. Miller, and J. A.
      induced regulation of surface transferring re-             Miller: Promotion by dietary phenobarbital of he-
      ceptor involves the action of activated protein            patocarcinogenesis by 2-methyl-N, N-dimethyl-
      kinase C and an intact cytoskeleton. J Biol Chem           4-aminoazobenzene in the rat. Cancer Res 39:
      260:9419, 1985.                                            112, 1979.
77.   M. Schliwa, T. Nakamura, K. R. Porter, and U.        91.   H. C. Pitot, L. Barsness, T. Goldsworthy, and T.
      Euteneuer: A tumor promoter induces rapid and              Kitagawa: Biochemical characterization of stages
      coordinated reorganization of actin and vinculin           of hepatocarcinogenesis after a single dose of
      in cultured cells. J Cell Biol 99:1045, 1984.              diethylnitrosamine. Nature 271:456 1978.
78.   E. G. Fey and S. Penman: Tumor promoters             92.   J. Jonkers and A. Berns: Conditional mouse
      induce a specific morphological signature in the            models of sporadic cancer. Nat Rev Cancer
      nuclear matrix–intermediate filament scaffold               2:251, 2002.
      of Madin-Darby canine kidney (MDCK) cell col-        93.   S. R. Lichter: Why cancer news is a health haz-
      onies. Proc Natl Acad Sci USA 81:4409, 1984.               ard. Wall Street Journal November 12, 1993.
79.   P. A. Cerutti: Pro-oxidant states and tumor          94.   J. Higginson: Changing concepts in cancer
      promotion. Science 227:375, 1985.                          prevention: Limitations and implications for
80.   J. Barsoum and A. Varshavsky: Mitogenic hor-               future research in environmental carcinogene-
      mones and tumor promoters greatly increase the             sis. Cancer Res 48:1381, 1988.
      incidence of colony-forming cells bearing am-        95.   F. Perera and P. Boffetta: Perspectives on
      plified dihydrofolate reductase genes. Proc Natl            comparing risks of environmental carcinogens.
      Acad Sci USA 80:5330, 1983.                                J Natl Cancer Inst 80:1282, 1988.
81.   T. Solo, T. Turpeenniemi-Hujanen, and K.             96.          ¨
                                                                 K. Muster, E. M. Whelan, G. L. Ross, and
      Tryggvason: Tumor-promoting phorbol esters                 A. N. Stimola, eds.: Animal cancer testing in
      and cell proliferation stimulate secretion of              laws and regulations. In America’s War on
60                                                                                         CANCER BIOLOGY

       ‘‘Carcinogens’’. New York: American Council         113. G. Merlino and F. P. Noonan: Modeling gene–
       on Science and Health, 2005, pp. 107–115.                environment interactions in malignant mela-
 97.   S. M. Cohen and L. B. Ellwein: Genetic errors,           noma. Trends Mol Med 9:102, 2003.
       cell proliferation, and carcinogenesis. Cancer      114. B. N. Ames, M. K. Shigenaga, and T. M.
       Res 51:6493, 1991.                                       Hagen: Oxidants, antioxidants, and the degen-
 98.   A. Hay: Testing times for the tests. Nature 350:         erative diseases of aging. Proc Natl Acad Sci
       555, 1991.                                               USA 90:7915, 1993.
 99.   J. Ashby and R. W. Tennant: Definitive rela-         115. E. R. Stadtman: Protein oxidation and aging.
       tionships among chemical structure, carcino-             Science 257:1220, 1992.
       genicity and mutagenicity for 301 chemicals         116. L. D. Youngman, J.-Y. K. Park, and B. N. Ames:
       tested by the U.S. NTP. Mutat Res 257:229,               Protein oxidation associated with aging is
       1991.                                                    reduced by dietary restriction of protein or
100.   I. B. Weinstein: Mitogenesis is only one factor          calories. Proc Natl Acad Sci USA 89:9112, 1992.
       in carcinogenesis. Science 251:387, 1991.           117. H. Joenje: Genetic toxicology of oxygen. Mutat
101.   R. L. Melnick: Does chemically induced hepa-             Res 219:193, 1989.
       tocyte proliferation predict liver carcinogene-     118. S. A. Grist, M. McCarron, A. Kutlaca, D. R.
       sis? FASEB J 6:2698, 1992.                               Turner, and A. A. Morley. In vivo human so-
102.   K. Meister, E. M. Whelan, G. L. Ross, and A. N.          matic mutation: Frequency and spectrum with
       Stimola, eds.: Limitations and results of animal         age. Mutat Res 266:189, 1992.
       carcinogen testing. In America’s War on ‘‘Car-      119. T. M. Reid and L. A. Loeb: Mutagenic spe-
       cinogens’’. New York:: American Council on               cificity of oxygen radicals produced by human
       Science and Health, 2005, pp. 45–57.                     leukemia cells. Cancer Res 52:1082, 1992.
103.   B. N. Ames, M. Profet, and L. S. Gold: Dietary      120. J. M. Carney, P. E. Starke-Reed, C. N. Oliver,
       pesticides (99.99% all natural) Proc Natl Acad           R. W. Lundum, M. S. Cheng, et al.: Reversal of
       Sci USA 87:7777, 1990.                                   age-related increase in brain protein oxidation,
104.   B. N. Ames, M. Profet, and L. S. Gold: Nature’s          decrease in enzyme activity, and loss in tempo-
       chemicals and synthetic chemicals: Comparative           ral and spatial memory by chronic adminis-
       toxicology (carcinogens/mutagens/teratofens/             tration of the spin-trapping compound N-
       dioxin). Proc Natl Acad Sci USA 87:7782, 1990.           tertbutyl-alpha-phenylnitrone. Proc Natl Acad
105.   J. B. Little: Radiation carcinogenesis in vitro:         Sci USA 88:3633, 1991.
       Implications for mechanisms. In H. H. Hiatt,        121. J. Kaiser: Tying genetics to the risk of environ-
       J. D. Watson, and J. A. Winsten, eds.: Origins of        mental diseases. Science 300:563, 2003.
       Human Cancer. Cold Spring Harbor, NY: Cold          122. L. A. Loeb, K. R. Loeb, and J. P. Anderson:
       Spring Harbor Laboratory, 1977, pp. 923–939.             Multiple mutations and cancer. Proc Natl Acad
106.   W. A. Haseltine: Ultraviolet light repair and            Sci USA 100:776, 2003.
       mutagenesis revisited. Cell 33:13, 1983.            123. I. Tomlinson and W. Bodmer: Selection, the
107.   A. C. Upton: The biological effects of low-level         mutation rate and cancer: Ensuring that the tail
       ionizing radiation. Sci Am 246:41, 1982.                 does not wag the dog. Nat Med 5:11, 1999.
108.   A. J. Grosovsky, J. G. De Boer, P. J. De Jong,      124. P. C. Nowell: The clonal evolution of tumor cell
       E. A. Drobetsky, and B. W. Glickman: Base                populations. Science 194:23, 1976.
       substitutions, frameshifts, and small deletions     125. V. A. Bohr, D. H. Phillips, and P. C. Hanawalt::
       constitute ionizing radiation-induced point mu-          Heterogeneous DNA damage and repair in the
       tations in mammalian cells. Proc Natl Acad Sci           mammalian genome. Cancer Res 47:6426, 1987.
       USA 85:185, 1988.                                   126. T. Lindahl: Instability and decay of the primary
109.   L. H. Lutze, R. A. Winegar, R. Jostes, F. T.             structure of DNA. Nature 362:709, 1993.
       Cross, and J. E. Cleaver: Radon-induced dele-       127. A. E. Pegg: Mammalian O6-alkylguanine-DNA
       tions in human cells: Role of nonhomologous              alkyltransferase: Regulation and importance in
       strand rejoining. Cancer Res 52:5126, 1992.              response to alkylating carcinogenic and thera-
110.   R. R. Weichselbaum, D. E. Hallahan, Z. V.                peutic agents. Cancer Res 50:6119, 1990.
       Sukhatme, A. Dritschilo, M. L. Sherman, et al.:     128. J. H. J. Hoeijmakers: Genome maintenance
       Review: Biological consequences of gene regu-            mechanisms for preventing cancer. Nature 411:
       lation after ionizing radiation exposure. J Natl         366, 2001.
       Cancer Inst 83:480, 1991.                           129. R. B. Setlow, F. E. Ahmed, and E. Grist:
111.   S. J. Garte, F. J. Burns, T. Ashkenazi-Kimmel,           Xeroderma pigmentosum: Damage to DNA
       M. Felber, and M. J. Sawey: Amplification of              is involved in carcinogenesis. In H. H. Hiatt,
       the c-myc oncogene during progression of                 J. D. Watson, and J. A. Winsten, eds.: Origins
       radiation-induced rat skin tumors. Cancer Res            of Human Cancer. Cold Spring Harbor, NY:
       50:3073, 1990.                                           Cold Spring Harbor Laboratory, pp. 889–202,
112.   D. Williams: Cancer after nuclear fallout:               1977.
       Lessons from the Chernobyl accident. Nat Rev        130. D. Bootsma and J. H. J. Hoeijmakers: Engage-
       Cancer 2:543, 2002.                                      ment with transcription. Nature 363:114, 1993.
CAUSES OF CANCER                                                                                              61

131. Y. Habraken, P. Sung, L. Prakash, and S.             147. B. E. Eddy, S. E. Stewart, and W. Berkeley:
     Prakash: Yeast excision repair gene RAD2 en-              Cytopathogenicity in tissue cultures by a tumor
     codes a single-stranded DNA endonuclease.                 virus from mice. Proc Soc Exp Biol Med 98:848,
     Nature 366:365, 1993.                                     1958.
132. D. Scherly, T. Nouspikel, J. Corlet, C. Ucla, A.     148. B. E. Henderson: Establishment of an associa-
     Bairoch, et al.: Complementation of the DNA               tion between a virus and a human cancer. J Natl
     repair defect in xeroderma pigmentosum group              Cancer Inst 81:320, 1989.
     G cells by a human cDNA related to yeast             149. H. zur Hausen: Viruses in human cancers. Sci-
     RAD2. Nature 363:182, 1993.                               ence 254:1167, 1991.
133. E. C. Friedberg: Xeroderma pigmentosum,              150. C.-M. Kim, K. Koike, I. Saito, T. Miyamura,
     Cockayne’s syndrome, helicases, and DNA re-               and G. Jay: HBx gene of hepatitis B virus
     pair: What’s the relationship? Cell 71:887, 1992.         indices liver cancer in transgenic mice. Nature
134. J. E. Cleaver: It was a very good year for DNA            351:317, 1991.
     repair. Cell 76:1, 1994.                             151. T. Sasagawa, M. Inoue, H. Inoue, M. Yutsudo,
135. L. L. Dumenco, E. Allay, K. Norton, and S. L.             O. Tanizawa, et al.: Induction of uterine cervical
     Gerson: The prevention of thymic lymphomas                neoplasias in mice by human papillomavirus
     in transgenic mice by human O6-alkylguanine-              type 16 E6/E7 genes. Cancer Res 52:4420, 1992.
     DNA alkyltransferase. Science 259:219, 1993.         152. B. J. Poiesz, F. W. Ruscetti, A. F. Gazdar,
136. S. M. O’Toole, A. E. Pegg, and J. A. Swenberg:            P. A. Bunn, J. D. Minna, et al.: Detection and
     Repair of O6-methylguanine and O4-methylthy-              isolation of type-C retrovirus particles from
     midine in F344 rat liver following treatment              fresh and cultured lymphocytes of a patient
     with 1,2-dimethylhydrazine and O6-benzylgua-              with cutaneous T-cell lymphoma. Proc Natl
     nine. Cancer Res 53:3895, 1993.                           Acad Sci USA 77:7415, 1980.
137. V. Ellermann and O. Bang: Experimentelle             153. B. J. Boiesz, F. W. Ruscetti, M. S. Reitz, V. S.
     Leukamie bei Huhnern. Zentralbl f Bacteriol 46:
                       ¨                                       Kalyanaraman, and R. C. Gallo: Isolation of a
     595, 1908.                                                new type-C retrovirus (HTLV) in primary
138. P. Rous: A sarcoma of the fowl transmissible by                                                    ´
                                                               uncultured cells of a patient with Sezary T-
     an agent separable from the tumor cells. J Exp            cell leukemia. Nature 295:268, 1981.
     Med 13:397, 1911.                                    154. R. C. Gallo and F. Wong-Staal: Current
139. R. E. Shope: Infectious papillomatosis of rab-            thoughts on the viral etiology of certain human
     bits. J Exp Med 58:607, 1933.                             cancers. Cancer Res 44:2743, 1984.
140. P. Rous and J. W. Beard: The progression to          155. V. S. Kalyanaraman, M. G. Sarngadharan, M.
     carcinoma of virus-induced rabbit papillomas              Robert-Guroff, D. Blayney, D. Golde, et al.: A
     (Shope). J Exp Med 62:523, 1935.                          new subtype of human T-cell leukemia virus
141. B. A. Lucke A neoplastic disease of the kidney of         (HTLC-II) associated with a T-cell variant of
     the frog, Rana pipiens. Am J Cancer 20:352, 1934.         hairy cell leukemia. Science 215:571, 1982.
142. J. J. Bittner: Some possible effects of nursing on   156. M. Nerenberg, S. H. Hinrichs, R. K. Reynolds,
     the mammary gland tumor incidence in mice.                G. Khoury, and G. Jay: The tat gene of human T-
     Science 84:162, 1936.                                     lymphotropic virus type I induces mesenchymal
143. L. Gross: Viral etiology of cancer and leukemia:          tumorsin transgenic mice.Science 237:1324, 1987.
     A look into the past, present, and future—G. H.      157. R. C. Gallo, V. S. Kalyanaraman, M. G.
     A. Clowes memorial lecture. Cancer Res                    Sarngadharan, A. Sliski, E. C. Vonderheid, et al.:
     38:485, 1978.                                             Association of the human type C. retrovirus
144. L. Gross: Development and serial cell-free pas-           with a subset of adult T-cell cancers. Cancer
     sage of a highly potent strain of mouse leukemia          Res 43:3892, 1983.
     virus. Proc Soc Exp Biol Med 94:767, 1957.           158. F. Wong-Staal, B. Hahn, V. Manzari, S.
145. L. Dmochowski: Viruses and tumors in the light            Colombini, G. Franchini, et al.: A survey of hu-
     of electron microscope studies: A review. Can-            man leukemias for sequences of a human retro-
     cer Res 20:997, 1960.                                     virus. Nature 302:626, 1983.
146. I.-M. Chiu, R. Callahan, S. R. Tronick, J.           159. H.-G. Guo, F. Wong-Staal, and R. C. Gallo:
     Schlom, and S. A. Aaronson: Major pol gene                Novel viral sequences related to human T-cell
     progenitors in the evolution of oncoviruses.              leukemia virus in T-cells of a seropositive ba-
     Science 223:364, 1984.                                    boon. Science 223:1195, 1984.

The Epidemiology
of Human Cancer

TRENDS IN CANCER INCIDENCE                                    The U.S. age-adjusted cancer death rates for
AND MORTALITY                                              selected cancer types from 1930 through 2001,
                                                           the latest year for which complete data are avail-
                                                           able, were discussed in Chapter 1. A number of
U.S. Data
                                                           points stand out from these data; for example, the
Long-range trends in the incidence of various              alarming increased mortality rate for lung cancer
cancers in different populations provide clues to          in both males and females. Even though this
the causes of cancer. Because of the long latency          steep rise of mortality rate for females was slower
period between the first exposure to carcino-               to occur than in males, the death rate due to lung
genic agents and the appearance of clinically              cancer surpassed that of breast cancer in the late
detectable cancer, which may be up to 20 or 30             1980s. The mortality rate for lung cancer in males
years, current trends probably reflect carcino-             has declined to some extent since the 1990s, prob-
gen exposure that began decades earlier. Ge-               ably due to a decreased rate of smoking in young
netic predisposition plays a role here in that             males that started muchearlier. Unfortunately, the
individuals carrying genetic susceptibility genes          incidence of smoking among young adults ap-
may develop cancer more rapidly or at an earlier           pears to be on the rise again since the year 2000.
age. Another major factor that affects the overall            Of over 1.3 million new cases of cancer each
incidence of cancer is the change in the average           year in the United States, about 570,000 patients
age of the population. The average age at the              die every year. The overall mortality rates since
time of diagnosis (averaged for all tumor sites) is        the 1970s are disconcerting (Fig. 3–1); they are
67,1 and as a higher proportion of the popula-             basically flat. There is a small downward trend in
tion reaches age 60 and above the incidence of             mortality rates for men since the mid- to late
cancer will go up as a result of this factor alone.        1990s, most likely because of better treatment,
Moreover, with the long-term downward trends               although diagnosticians will argue that this is
in other causes of death, primarily infectious and         due to earlier diagnosis, primarily for prostate
cardiovascular diseases, more people live to an            cancer. While this conjecture may be partly cor-
age when the risk of developing cancer becomes             rect, the overall survival data do not suggest the
high. It is projected, for example, that about one         concept that prostate-specific antigen (PSA) lev-
in four males and one in five females born in               els have made a large difference in long-term
1985 in the United States will eventually die of           overall survival. This is a controversial area and
cancer.2 This is up from about 18% for males               will be discussed further in Chapter 7. The data
and 16% for females born in 1975. If current               for prostate cancer indicate a spike in cancer inci-
trends continue, about one in three Americans              dence for men from 1990 to 1995, due to the in-
now living will develop some form of cancer.3              troduction of large-scale PSA testing (Fig. 3–2).

THE EPIDEMIOLOGY OF HUMAN CANCER                                                                                                  63

                                                                                         Male Incidence


                                                                                  Male and Female Incidence
Rate per 100,000 Population

                                                                                       Female Incidence


                              300                                                        Male Mortality

                                                                                   Male and Female Mortality

                                                                                         Female Mortality


                                1975   1977   1979   1981   1983   1985    1987   1989     1991   1993      1995   1997   1999   2001
                                                                      Year of Diagnosis/Death

                                       Figure 3–1. Annual age-adjusted cancer incidence and death rates* for
                                       all sites, by sex, United States, 1975 to 2001. *Rates are age adjusted to the
                                       2000 U.S. standard position. (Source: Incidence data from Surveillance,
                                       Epidemiology, and End Results (SEER) Program, nine oldest registries, 1975
                                       to 2001, Division of Cancer Control and Population Sciences, National Can-
                                       cer Institute, 2004. Mortality data from U.S. Mortality Public Use Data Tapes,
                                       1960 to 2001, National Center for Health Statistics, Centers for Disease
                                       Control and Prevention, 2004. From Jemal et al.,3 with permission.)

If a significant number of these men were go-                                      The overall number of yearly cancer cases
ing to die of prostate cancer, the mortality rate                              by state in the United States is shown in Figure
should have fallen more dramatically than it                                   3–3.4 These figures are not normalized by pop-
has in recent years. That is not to say that some                              ulation, so they reflect the absolute numbers only.
men’s lives have not been saved by early diag-                                 It should be noted that basal cell and squamous
nosis of prostate cancer, but on a population                                  cell carcinoma of the skin are excluded from
basis, it is by no means clear that the current                                these data and those discussed above, because
PSA test has been a huge success.                                              these cancers are almost always curable, even
64                                                                                                                                                                            CANCER BIOLOGY

                              260                                             Male                                         260                                        Female

                              240                                                                                          240
                              220                                                                                          220

                              200                                                                                          200

                              180                                                                                          180
Rate per 100,000 Population

                              160                                                                                          160

                              140                                                                                          140                                               Breast

                              120                                                                                          120

                                                                                      Lung and Bronchus
                              100                                                                                          100

                               80                                             Colon and Rectum                              80

                                                                                                                                                      Colon and Rectum
                               60                                                                                           60

                                                                         Urinary Bladder                                            Lung and Bronchus
                               40                                                                                           40
                                                            Non-Hodgkin Lymphoma                                                                                            Uterine Corpus
                               20                                                                                           20                    Ovary
                                                                              Melanoma of the Skin                                                                          Non-Hodgkin Lymphoma
                                0                                                                                            0

                                                            Year of Diagnosis                                                                              Year of Diagnosis

                                                   Figure 3–2. Annual age-adjusted cancer incidence rates* among males and
                                                   females for selected cancer types, United States, 1975 to 2001. *Rates are age
                                                   adjusted to the 2000 U.S. standard population. (Source: Surveillance, Epi-
                                                   demiology, and End Results (SEER) Program, nine oldest registries, 1975 to
                                                   2001, Division of Cancer Control and Population Sciences, National Cancer
                                                   Institute, 2004. From Jemal et al.,3 with permission.)

though the number of cases is enormous. The in-                                                                          ences among developing and developed countries
cidence rate for these nonmelanoma skin can-                                                                             in the incidence rates of certain cancers, lung
cers is over one million new diagnoses per year.3                                                                        cancer is the most common cancer among men
                                                                                                                         in both regions of the world and breast cancer is
                                                                                                                         the most common cancer in women (Fig. 3–4).5
Cancer Is a Global Problem
                                                                                                                            There are, however, regional differences in
Cancer is clearly a worldwide problem. The in-                                                                           the distribution of various cancers in different
cidence and mortality rates for various cancers                                                                          regions of the world that reflect differing etiologic
are similar, though not identical, among devel-                                                                          factors. For example, infectious etiology plays
oped countries. In the developing world, as coun-                                                                        a greater role in certain parts of the world, e.g.,
tries become more westernized and their pop-                                                                             the role of schistosomiasis infections in caus-
ulations achieve longer life expectancy, cancer                                                                          ing bladder cancer in parts of Africa and that of
rates are increasing. Although there are differ-                                                                         hepatitis B infections in liver cancer in China
THE EPIDEMIOLOGY OF HUMAN CANCER                                                                                                                65

                                                                                                                    VT      6000
                                                                                                                   3100                ME
                                    MT              ND                                                                                7300
         WA                        4600            3100              MN
        26,700                                                      21,900                                                                     MA

                                                    SD                                                                      NY                  RI
                                                                                   WI                                                          5800
                                                   3900                          25,800                                   85,900
       OR                 ID          WY                                                         MI
      17,300             5500                                                                  47,400                                          CT
                                     2300                                                                                                    16,600
                                                                       IA                                           PA                NJ
                                                     NE              15,300                                       70,800            42,300
                                                    8100                                       IN      OH
             NV                                                                         IL   31,200 60,300                                    DE
           10,300                                                                     59,900                                                 4100
                            UT                                                                             WV                        MD
                           6200            CO                                                             11,300 VA
                                          15,200                                                                                    24,400
                                                        KS               MO                         KY            32,800
                                                      12,600            29,500                    22,100
       CA                                                                                                                           DC
     125,000                                                                                                          NC           2700
                                                                                               TN                   39,600
                                                            OK                               30,500
                                                                           AR                                   SC
                           AZ         NM                   17,700        14,700
                         23,300      7400                                                                     20,600
                                                                                       MS        AL      GA
                                                                                      14,900   23,600   33,400

                AK                                                                                               96,100

                                           4900                                                                            PR

                      Figure 3–3. Estimated number of new cancer cases in the United States, by
                      state, for 2003. Rates are age adjusted to the 2000 U.S. standard population.
                      Excluded are basal and squamous cell skin cancers and in situ carcinomas
                      except urinary bladder. (From Cancer Facts and Figures—2003,4 with per-
                      mission from American Cancer Society, Inc., 2006.)

and other parts of the Southeast Asia. Cancers                      the incidence of cigarette smoking, with a lag of
of the stomach and uterine cervix are higher in                     about 20 years, in both men and women (Fig.
parts of the developing world, whereas colorectal                   3–5).6a Since the mid-1900s in men and 1950s for
and prostate cancers are higher in the developed                    women, the rate of lung cancer has risen dramat-
world.5 The global death rate due to cancer is                      ically. The highest rates include the United
estimated to be more than 6 million people an-                      States, United Kingdom, Japan, and Australia
nually, with about 10 million new diagnoses every                   and the lowest rates are in Africa and Southern
year.5 Over 22 million individuals have been                        Asia.6 Female incidence rates are highest in the
diagnosed with cancer worldwide. Thus, there is                     United States, Canada, Denmark, and the United
a global and growing cancer problem.                                Kingdom but lower in countries such as France,
                                                                    Japan, and Spain, where the prevalence of smok-
                                                                    ing among women has been low until recently.
DATA FOR SOME PREVALENT                                                The etiology of lung cancer is predominantly
HUMAN CANCERS                                                       related to cigarette smoking. The data for this are
                                                                    overwhelming.6 Other proposed causes include
                                                                    environmental pollution, occupational exposure
Lung Cancer
                                                                    (e.g., asbestos, coal mining), passive smoke in-
Lung cancer is the most common cancer world-                        halation, and radon exposure. These are all minor
wide and the leading type of cancer mortality in                    players and, in fact, the role of passive smoke and
men. The incidence of lung cancer closely tracks                    radon in the home as causes are insignificant, in
                                    Developed                                                            Developing

                 Lung               482                                                                                         481
                                          424                                                                             423
             Stomach                                     196                                                             405
                                                               129                                               316
         Colon/rectum                           353                                                 196
                                                           160                               118
              Prostate           513                                                               165
                                                               130                          91
                 Liver                                                74                                            366
                                                                      71                                          344
           Esophagus                                                   58                                  256
                                                                       50                            210
              Bladder                                     175                                99
                                                                       52              56
           Oral cavity                                                 65                    111
                                                                          22           58
            Leukemia                                                  70                    101
                                                                        46               79
Non-Hodgkin lymphoma                                                 83                    92
                                                                         39            59
               Larynx                                                  58                 82
                                                                          28          50
             Pancreas                                                 71               54
                                                                      71              48
          Kidney, etc.                                               87                 42                         Incidence
                                                                         40           22
        Other pharynx                                                    41             65                         Mortality
                                                                          20          48
 Brain, nervous system                                                   41             67
                                                                          27          48

                          700 600   500 400           300 200 100   0   100                        200     300         400 500        600 700

                                    Developed                                                            Developing

                Breast     636                                                                                                    514
                                                         190                                             221
         Colon/rectum                             312                                              160
                                                           154                              96
           Cervix uteri                                              83                                                   409
                                                                          40                              234
             Stomach                                             115                                     214
                                                                    84                             170
                 Lung                                    195                                         191
                                                           161                                     168
           Ovary, etc.                                               97                      108
                                                                       62              63
          Corpus uteri                                         136                      62
                                                                         29       21
                 Liver                                                   36                   147
                                                                        38                  143
           Esophagus                                                       16               130
                                                                           15             110
            Leukemia                                                   55               75
                                                                        38            59
Non-Hodgkin lymphoma                                                     68            58
                                                                          34         38
             Pancreas                                                 65             43
                                                                      68            40
           Oral cavity                                                   26            72                          Incidence
                                                                             8      38
               Thyroid                                                   44            60                          Mortality
                                                                             7    17
          Kidney, etc.                                                   33        25
                                                                          25      14

                          700 600   500 400           300 200 100   0   100                        200     300         400 500        600 700

              Figure 3–4. Estimated numbers of new cancer cases (incidence) and deaths
              (mortality) in 2002. Data shown in thousands for developing and developed
              countries by cancer site and sex. (From Parkin et al.,5 with permission.)
THE EPIDEMIOLOGY OF HUMAN CANCER                                                                                  67

                                                              of diagnosis. They are in general more sensitive
                                                              to chemotherapy than other lung cancers, but
                                                              usually relapse and are fatal. Small cell lung
                                                              carcinomas (SCLC) also tend to produce a variety
                                                              of hormones and are associated with paraneo-
                                                              plastic syndromes (see Chapter 8).
                                                                 There are a number of genetic alterations ob-
                                                              served in progressive lung cancers (Table 3–1).
                                                              Mutations in p53 are common, and begin to be
                                                              observed in early tumors, particularly squamous
                                                              cell carcinomas. K-ras activation occurs more
                                                              commonly in adenocarcinomas than the other cell
Figure 3–5. Trends in smoking prevalence and lung             types. Loss of heterozygosity at chromosomal
cancer in British males and females. The data for this        locus 3p and of fragile histidine triad (FHIT)
chart are from England and Wales. In men, smoking             occur early in neoplastic transformation and are
(open circles) began to increase at the beginning of
                                                              associated with smoking. Silencing of the cyclin-
the twentieth century, but the corresponding trend in
deaths from lung cancer (filled circles) did not begin         dependent kinase inhibitor by gene methylation
until after 1920. In women, smoking (open squares)            or deletion has been observed in 30% or more
began later, and the increase in lung cancer deaths in        of adenocarcinomas and squamous cell carcino-
women (filled squares) has appeared only recently.             mas. The implications of these genetic changes
(From Loeb et al.,6a with permission.)
                                                              are described in Chapter 5. Other genes have
                                                              been associated with lung carcinogenesis, in-
spite of environmentalist hype (see Cancer Myths,             cluding erb b1 and b2, the myc family of genes,
below).                                                       c-myc, N-myc, and L-myc. Gene expression pro-
   Lung cancer is still classified by histological             filing has shown that a variety of genes are up- or
cell type into squamous cell carcinoma, adeno-                down-regulated in non–small cell lung cancers
carcinoma, and large- and small-cell carcinomas.              (see Chapter 7). Overall 5-year survival is poor
The first three are often lumped together as non–              for all types of lung cancer. It is only 15%, but it
small cell lung carcinomas. Squamous cell car-                is 49% if diagnosed when the disease is still lo-
cinoma is strongly associated with smoking and is             calized (about 15% of cases).
the most prevalent type. Adenacarcinomas are
less strongly associated with smoking. These tu-
                                                              Breast Cancer
mors are gland forming and mucous producing.
Small cell lung carcinomas are usually aggressive             Over 210,000 new cases of invasive breast can-
and invasive and are often metastic at the time               cer and over 40,000 deaths due to breast cancer

Table 3–1. Genetic Alterations in Lung Tumors
                                                                                  Frequency (% of tumors)

                                                                     Small Cell                         Squamous Cell
Gene                   Locus       Alteration                        Carcinoma      Adenocarcinoma      Carcinoma
p53                    17p13       Deletion, mutation (G:C > T:A),   70–90          30                  50
KRAS                   12p21       Mutation (GGT > TGT)              <1             15–60               8–9
CDKN2A /p16INK4        9p21        Deletion, mutation,               <1             27–59               33–40
LOH 3p                 3p          Deletion (loss of                 100                        50–85
FHIT                   3p14.2      Delection (loss of                76                         40–76
                                    transcriptional dysregulation
From World Cancer Report,6 with permission.
68                                                                                      CANCER BIOLOGY

occurred in 2003 in the United States, which is         years is associated with an increased risk of
a decrease of about 2,000 annual deaths from            breast cancer.
the peak year of 1995.3,3a Breast cancer is also           In contrast to the somewhat conflicting re-
the most common malignancy of women world-              sults on the role of dietary fat, there is a general
wide, with more than one million new cases              consensus that obesity is a significant risk factor
occurring annually. Although breast cancer is           for breast cancer. For example, a study of the
more common in developed Western societies,             relationships between body mass index (BMI)
with an incidence rate greater than 80 per              with serum estrogen levels and breast cancer in-
100,000 population per year, incidence rates are        cidence found that there was an increased risk of
increasing in the developing world. From                breast cancer with increasing BMI among post-
1975 to 1990, the largest increases (1%–5%) were        menopausal women.9 This result was largely as-
seen in Asia, Africa, and parts of Europe that          sociated with increased bioavailable serum con-
previously had low incidence rates.7                    centration of estrogen in the women with high
   Risk factors include duration of exposure to         BMIs. The mechanism of this is postulated to
female hormones (early menarche and late meno-          result from elevated production of estrogen
pause); reproductive factors (nulliparity, late age     by aromatase in adipose tissue and a decrease
of first pregnancy); dietary and low physical activ-     in the serum concentration of sex hormone–
ity factors (obesity; high-fat diet); ionizing radia-   binding globulin.10
tion during breast development; chronic use of             Early detection of breast cancer is key to sur-
hormone replacement therapy; and genetic in-            vival rates. The 5-year survival rate for localized
heritance (family history) of breast cancer such as     breast cancer is >95% but drops to 78% for re-
brca1, brca2, or p53 germline mutations.                gional spread, and 23% for metastic disease.4
   Although high dietary fat intake has been asso-      Screening mammography, though subject to de-
ciated with an increased risk of breast cancer in       bate regarding the women’s age to start this and
animal studies, international population compar-        its role in overall survival, is still the most widely
isons, and some case–control studies, this associ-      used screening tool. Improved methods such as
ation has not been found in some prospective            MRI may enhance the accuracy of diagnosis.
studies. Case–control studies have supported               Another issue related to mammography screen-
a positive risk correlation with high intake of         ing is the high incidence of ductual carcinomas in
saturated fat, but a pooled analysis of prospec-        situ (DCIS) that is detected, over 30% in some
tive studies found only a weakly positive asso-         screening centers. The catch is that some of these
ciation (reviewed in Reference 8).                      lesions will progress to invasive disease and some
   There have been some issues with the way             will not,11 and currently there is no good way to
these studies were carried out. For example, ear-       tell which ones will progress and which will not.
lier prospective studies on the association of          Gene expression microarrays may make this de-
fat intake and breast cancer included relatively        termination possible in the future.12
few premenopausal women who later developed                Germline mutations, including brca1 and
breast cancer. Since one would expect that high         brca2, account for only 15%–20% of familial-
fat intake during premenopausal years may be            related breast cancers and only 5% of all breast
the time at risk for later tumorigenesis, this is an    cancers.13 These genetic susceptibility genes have
important group to study. Cho et al.8 studied the       variable penetrance in various individuals, which
dietary fat intake and breast cancer risk among         most likely reflects the expression of hormone
90,655 women of ages 26–46 years enrolled in            metabolizing genes,14 DNA repair genes, im-
the Nurses Health study in 1991. Fat intake was         mune surveillance, H-ras, and androgen receptor
assessed with a food-frequency questionnaire            genes.13 The lifetime risk of developing breast
at baseline in 1991 and again in 1995. Breast           cancer for women with the brca1 mutation varies
cancers were confirmed by review of pathology            from 36% to 80% depending on the population
reports. The conclusion from this study was that        studied.13 The risk increases wih age: by age
intake of animal fat, mostly from red meat and          80, the relative risk is 80% for individuals with a
high-fat dairy products, during premenopausal           brca1 mutation. Thus, finding the additional genes
THE EPIDEMIOLOGY OF HUMAN CANCER                                                                                                                             69

that modulate breast cancer susceptibility is a key                                        the conditions of an affluent lifestyle, because
area for research.                                                                         the major risk factors are a diet abundant in fat,
   Some of the key genes involved in the path-                                             refined carbohydrates, and animal protein and
ogenesis and progression of breast cancer have                                             low in fiber, combined with physical inactivity.18
been identified. These include loss of het-                                                 However, the primary risk factor is age, with
erozygosity on chromosomal loci 13q, 9p, and                                               over 90% of cases diagnosed in people over age
16q that involve rb (the retinoblastoma gene),                                             50. This profile may be due to a lifetime of bi-
CDKN2 (encoding the p16 protein), and CDH1                                                 ological and chemical insults resulting from the
(encoding the E-cadherin protein).15 Other com-                                            above associated risk factors.
mon types of genetic alteration are the amplifi-                                               There is a clear overlay of genetic predisposi-
cation of erb B2, c-myc, cyclin D1, and insulin-                                           tion for colorectal cancer. Two of the genetically
like growth factor genes.16                                                                inherited syndromes are familial adenomatous
                                                                                           polyposis (FAP) and hereditary non-polyposis
                                                                                           colon cancer (HNPCC). The genes involved in
Colorectal Cancer
                                                                                           these syndromes have been identified (see below).
In the United States, 105,000 cases of colon                                                  The molecular genetic changes that occur dur-
cancer and 40,000 cases of rectal cancer were                                              ing malignant transformation and tumor pro-
expected in 2005, and an estimated 56,290                                                  gression have been well studied by Vogelstein and
deaths for both combined.3 This number ac-                                                 colleagues.19 Although the genetic alterations
counts for about 10% of all cancer deaths in the                                           that occur during colon cancer progression sug-
United States. Over 940,000 cases of colorectal                                            gest that the changes occur sequentially, this is
occur annually worldwide.17 Although cancers of                                            probably not the case. It is more likely that it is
the colon and rectum are relatively rare in de-                                            the accumulation of the changes that occur in
veloping countries, they are the second-most                                               the APC, K-ras, DCC, p53 genes, and other
frequent malignancies in the developed world.                                              genes that results in invasive colon carcinoma
This discrepancy appears to be largely due to                                              (Fig. 3–6). Moreover, Smith et al.20 have reported

 LKB                    PEUTZ-JEGHER SYNDROME                                                      LATE RER+   Hypermethylation     IGFIIR    RER+
                     JUVENILE POLYPOSIS SYNDROME                             Fam                   ADEMOMA                           BAX     CANCER
  DPC 4          HEREDITARY MIXED POLYPOSIS SYNDROME                             PMS2                                                E2F4
      ?                   COWDEN SYNDROME
 PTEN                                                                           PMS1                                                                  p15

                               Hypomethylation                              GTBP                                                                      Bub1

                  1p                                                      polymerase
                                                                                                                                                      cyclin D1
 NORMAL                             EARLY                         INTERMEDIATE           SMAD 4          LATE                     SPORADIC            7q
EPITHELIUM         APC             ADEMOMA           KRAS           ADEMOMA              SMAD 2        ADEMOMA       p53           CANCER             17q
                   CTNNB1                                                                 DCC
                                   2                                                                                                                  14q
                               B          YC
                                        CM                                                                                                            22q


                       eu                                                                                                                             8p
                            idy                                                                              ULCERATIVE COLITIS-
                                               LOW-GRADE                   HIGH-GRADE                      ASSOCIATED COLORECTAL                      tPa
                                                DYSPLASIA                   DYSPLASIA                            CARCINOMA                            CEA
                                    ?          EARLY MHAP/              INTERMEDIATE                        LATE                       SPORADIC
                                                SERRATED                  ADENOMA                         ADENOMA                       CANCER
             POLYP                                                                                                                                    E-cadherin
                      FLAT                              ? progression                       FLAT
                    ADENOMA                                                                CANCER

                               Figure 3–6. Putative genetic pathways in colorectal cancer. It is thought that
                               most tumors develop according to the original Vogelstein model (bold arrows).
                               MHAP, mixed hyperplastic adenomatous polyps. (From World Cancer Re-
                               port,17 with permission.)
70                                                                                      CANCER BIOLOGY

that in contrast to the predictions of the se-          include CDKN2A, M6P/ IGF2R, SMAD gene
quential model of mutation accumulation, only           family members, and cyclin D and A genes.
6.6% of colon cancers from a cohort of 106 co-
lorectal cancer patients had mutations in all
three genes: APC, K-ras, and p53, whereas               Pancreatic Cancer
38.7% of tumors had mutations in only one of            Even though pancreatic cancer is the tenth lead-
these genes. The most common combination of             ing cancer by incidence in men and ninth in
mutations was APC plus p53 (27.1%). Mutations           women, it is the fourth leading cause of cancer
in both K-ras and p53 were rare. These data             deaths in the United States. Overall survival is
suggest that colorectal cancers are a heteroge-         poor. For all stages combined, 1-year survival is
neous family of tumors that can arise from al-          21% and 5-year survival is 4%.4 It is a ‘‘silent’’
ternative pathways.                                     disease symptomatically and difficult to diagnose
   Treatment is primarily surgical and is often         early, but even for those patients diagnosed with
curative if disease is localized. Chemotherapy or       so-called early disease, 5-year survival is only 17%.
chemotherapy plus irradiation is used for deeply           Pancreatic cancer is primarily a disease of de-
invasive or metastatic disease. Usually metasta-        veloped countries. About 216,000 cases occur
sis occurs first in the liver, for which local hepatic   annually worldwide. Risk factors include smok-
radiation plus hepatic artery infusion improves         ing, obesity, physical inactivity, chronic pancre-
survival.21 However, 5-year survival for meta-          atitis, diabetes, liver cirrhosis, and high-fat diet.4
static colon cancer is dismal, 9%, in contrast to       Familial genetic risk appears to account for about
90% 5-year survival for patients with localized         10% of cases. K-ras and p53 gene alterations are
disease (Dukes A). Unfortunately, only about            the most common ones observed. As can be sur-
37% of patients are diagnosed at that stage.4 Re-       mised from the poor survival rates, no effective
cently, clinical studies have shown chemopre-           treatment currently exists.
ventive effects with aspirin and nonsteroidal
anti-inflammatory drugs (NSAIDS), particularly
COX-2 inhibitors.22                                     Cancers of the Female
                                                        Reproductive Tract
Liver Cancer                                            The most common cancers of the female repro-
Although not a major cause of cancer in Western         ductive tract are cervical, ovarian, and endome-
societies, primary hepatocellular carcinoma is a        trial arcinomas Annually worldwide, there are
huge problem in a number of areas of the world,         470,000 new cases of cervical cancer (with 230,000
especially Southeast Asia and Africa. Worldwide,        deaths; 80% in developing countries), 190,000
about 560,000 new cases occur annually.23 Liver         cases of ovarian cancer (mostly in postmenopausal
cancer is extremely difficult to treat and overall,      women in developed countries), and 188,000 new
5-year survival (all stages) is only about 7% in        cases of endometrial cancer (also mainly in post-
the United States and even lower in developing          menopausal women in developed countries).24
countries. The primary risk factor in parts of the
world where liver cancer is prevalent is hepatitis B
                                                             Cervical Cancer
infection. Other risk factors include infection with
hepatitis C virus, chronic liver cirrhosis, alcohol     This is almost two different diseases: one in the
abuse, aflatoxin exposure, and parasitic infections.     developing world and a different one in the de-
   The molecular genetic changes in liver cancers       veloped world, although the etiology is similar.
are diverse and probably reflect the various causa-      For example, in the United States and other
tive agents. For example, p53 mutations are an          developed nations, routine PAP smears and gy-
early event in high-incidence areas (e.g., China)       necologic examinations detect the majority of
but a late progression-related event in West-           cases early and the cure rate is close to 100%
ern countries.23 K-ras mutations and cyclin D1          for patients with pre-invasive lesions (cervical
amplification are only observed in a small subset        intraepithelial neoplasia, CIN). The cure rate is
of patients. Other observed genetic alterations         92% for localized carcinomas that are minimally
THE EPIDEMIOLOGY OF HUMAN CANCER                                                                          71

invasive.4 However, it is a different story in           and phase three is characterized by triploid for-
developing parts of the world, where 80% of the          mation linked to the 6qÀ/1qÀ pathway.
470,000 annual worldwide cases occur.
    The primary risk factor is sexually transmit-
                                                              Endrometrial Cancer
ted infection with certain species of human pap-
illoma virus and is linked to having early sex and       Cancers of the uterine corpus are the seventh-
multiple sex partners. Additional risk factors are       most common cancer of women worldwide, with
HIV infection and a weak immune system.                  189,000 new cases and 45,000 deaths annually
                                                         worldwide. Sixty percent of these cases are in
                                                         developed countries, with the United States and
     Ovarian Cancer
                                                         Canada having the greatest prevalence.24 The
About 190,000 new cases and 114,000 deaths               etiology is linked to reproductive history: highest
from ovarian cancer occur annually in the                in nulliparous women and those with late meno-
world.24 The highest case loads are in Scandi-           pause. The cause is linked to duration and type of
navia, Eastern Europe, the United States, and            estrogen exposure. Estrogen replacement ther-
Canada. Low incidence rates occur in Africa and          apy, tamoxifen treatment, early menarche, and
Asia. As with most cancers, the risk increases with      late menopause are examples of this. Addition of
age. Risk factors include obesity, a history of pel-     progesterone to estrogen in hormone replacement
vic inflammatory disease, polycystic ovary syn-           regimens lowers the risk of ovarian cancer, but
drome, and endometriosis. Hormone replacement            not of breast cancer, in which both estrogen and
therapy increases risk, whereas oral contracep-          progesterone are cell proliferation promoting.
tives and tubal ligation decrease risk. A family            Endometrial cancer is often diagnosed early
history of breast or ovarian cancer also increases       because abnormal uterine bleeding is a common
risk and involves mutations in brca1 and brca2           symptom. Treatment is by hysterectomy, followed
genes. Hereditary non-polyposis colon cancer has         by radiation and/or chemotherapy if the disease
also been associated with ovarian cancer.                has become invasive or metastatic. Five-year sur-
   Symptoms of ovarian cancer are vague and may          vival for local, regional, and distant stage disease
include abdominal bloating, vague digestive dis-         are 96%, 64%, and 26%, respectively. Common
turbances, and, rarely, abnormal vaginal bleeding.       genetic alterations are seen in K-ras, PTEN, p53,
For this reason, early diagnosis is very difficult        Her 2/neu, cell cycle regulatory genes (e.g. cyclin
and only about 25% of cases are detected at local-       D1, p16/INK4a), and E-cadherin.24
ized stage.4 Overall 5-year survival for all stages is
53%, and in the minority of cases with localized
                                                         Prostate Cancer
disease, 5-year survival is over 90%. The tumor
marker CA-125 has had some usefulness in diag-           An estimated 232,000 new cases of prostate
nosis, but it is not sufficiently sensitive or specific    cancer and 30,000 deaths are expected in the
to be used for routine screening.25 Treatment            United States in 2005.3 About 200,000 deaths
involves surgery, radiation, and chemotherapy.           occur worldwide annually, mostly affecting men
   Genetic alterations observed during disease           over age 65 in developed countries.27 African-
progression are overexpression of Her2/neu and           American men have the highest prostate cancer
c-myc. K-ras activation is often observed and p53        incidence rates in the world, for reasons that
mutations occur in 50% of cases.24 Ovarian can-          aren’t clear but may relate to genetic polymor-
cer appears to develop through multiple chro-            phisms in the synthesis and metabolism of an-
mosomal abnormalities.26 Ovarian carcinomas de-          drogenic hormones.5
velop through three or more phases of karyotypic           The only well-established risk factors are age,
derangements. In the first phase, karyotypic ab-          race, and heredity. Environmental factors are
normalities proceed in a step-wise fashion in-           involved, as indicated by incidence rates that
volving either hyperdiploid imbalances (with ad-         change as populations migrate. For example, first-
ditions of genetic material at chromosomes þ7/           generation Japanese American men (Nisei) born
þ8/þ12 or deletions at 6qÀ/1qÀ. A second                 in the United States experience an increase in
phase involves increased chromosomal instability         prostate cancer that more closely resembles the
72                                                                                       CANCER BIOLOGY

incidence rate of the Caucausian U.S. male pop-           found that phenotypically similar tumor foci had
ulation (this increase for risk is also true for breast   different genotypes,providing additional evidence
and colorectal cancers). Dietary risk factors are a       for the multifocality of tumor development in the
diet rich in red meat and high fat. Dietary antioxi-      prostate. This finding has important clinical im-
dants such as selenium and lycopene lower risk.28         plications, because these different tumor foci may
   Family history of prostate cancer is one of the        have different degress of invasiveness, androgen-
most consistent risk factors: up to 40% of pros-          dependence, and drug responsiveness.
tate cancer patients hava a familial association.28          Screening for prostate cancer by the PSA test
However, although some candidate genes have               has led to a large increase in the detection of
been identified, no smoking gun similar to brca1           early prostate lesions. The number of these that
and brca2 for breast cancer has been found.               would actually develop into invasive cancer that
   Several molecular genetic alterations associ-          would kill men is subject to intense debate.
ated with prostate carcinogenesis and progression         The huge increase in prostate cancer incidence
have been identified.28 These involve chromo-              seen in the United States between 1990 and
somes 8, 10, 13, 16 and 17. Loss of heterozygosity        1995, as detected by the PSA test, has not
on chromosome 8p appears to be an early event.            translated into a great change in overall mor-
Interestingly, hypermethylation of the promoter           tality (Fig. 3–2). Moreover, in the United King-
region of the glutathione S-transferase P1 (GST-          dom, the rise in incidence of prostate cancer,
P1) gene has been observed in about 90% of pros-          though shower and more delayed, also has not
tate carcinomas and 70% of high-grade prostatic           led to very great changes in overall prostate
intraepithelial neoplasias (PIN). Since GST-P1            cancer mortality.32 PSA, originally thought to
is involved in detoxification of carcinogens and           be a specific prostate epithelial marker, has been
promoter methylation has been implicated in               observed in other tissues, including liver, colon,
gene silencing (see Chapter 5), the hypothesis is         lung, kidney, breast, ovarian, and parotid tumors
that silencing of this gene reduces a protective          as well as in breast milk and amniotic fluid.32
effect in the prostate gland. Methylated DNA              Transiently increased serum PSA has also been
sequences can be detected in the urine and this           observed in patients infected with hepatitis A
may provide a tumor marker that is more sen-              virus.33 Thus, more sensitive and specific diag-
sitive and specific than PSA (see Chapter 5).              nostic tests for prostate cancer are needed. Also,
Measurement of age-related, oxygen radical–               methods to determine which prostate lesions
induced DNA damage, as measured by 8-hydroxyl             are likely to progress to invasive, metastatic dis-
adenine and guanine levels in normal vs. tumor            ease and which are likely to remain indolent for
prostatic tissue,29 suggests that age-accumulated         a man’s life span are badly needed.
damage to prostatic tissue is important for the              Treatment for prostate cancer includes surgery
neoplastic transformation in the gland. This may          or radiation therapy for localized lesions, followed
also explain why antioxidants such as lycopene            by androgen deprivation therapy, and finally by
and selenium have a protective effect.                    chemotherapy if the disease progresses. The high
   The origin of cancer in human tissues is               incidence of impotence and urinary incontinence
thought to be clonal, in that it arises from a single     of postsurgical resection is still a problem, though
clone of transformed cells that undergo expan-            now much less so with nerve-sparing surgical
sion, and through cellular evolution accompanied          approaches.34 Five-year survival for all stages is
by genetic instability, these cells eventually be-        close to 97%. However, 10-year and 15-year sur-
come a full-blown malignant neoplasm. Prostate            vival rates are 79% and 57%, respectively.4
cancer may be an exception to that. Through mi-
crodissection of tumors in different locations from
                                                          Urinary Bladder Cancer
the same patients and detemination of four DNA
microsatellite polymorphic markers, it was con-           Bladder cancer is the ninth-most common ma-
cluded that different tumor foci in a given patient       lignancy in the world.35 There are 330,000 new
have an independent origin.30 Similarly, Macin-           cases and 130,000 deaths each year. In the United
tosh et al.,31 employing precise microdissection          States alone there are over 63,000 new cases
of different tumor foci in individual patients,           annually and 13,000 deaths.3
THE EPIDEMIOLOGY OF HUMAN CANCER                                                                        73

   Smoking is the greatest risk factor and is es-       available and distinct genetic events such as chro-
timated to be a causative factor in 65% of males        mosomal translocations are clearly related to
and 30% females in some developed countries.            disease progression, it has been easier to study
Historically, some types of bladder cancer were         such events in lymphomas and leukemia than
associated with abuse of analgesic combinations         in solid tumors such as lung, breast, and colon.
containing phenacetin and occupational expo-            Chromosomal translocations are often the incit-
sure in the aniline dye industry (e.g., exposure to     ing events in lymphomas, in contrast to solid tu-
2-naphthylamine). In Egypt and some other               mors, where gene deletions are more common.
African nations, chronic bladder infections with        The translocation events often involve the im-
Shistosoma haematodium are a risk factor.               munogloblin (Ig) loci and a proliferative or anti-
   Bladder cancer is treated by surgical removal if     apoptotic gene such as BCL-2. About one-sixth of
disease is invasive. Superficial, localized cancers      all NHLs have translocations of the BCL-6 gene
can be treated by local instillation of immuno-         that encodes a transcriptional repressor of nor-
modulatory agents (e.g., bacilli Calmette-Guerin        mal B-lymphocyte differentiation.37 This favors
[BCG]) or chemotherapy. Chemotherapy and/or             cell proliferation and decreased cell senescence.
radiation therapy have been used as an adjuant or       Some NHLs have translocations that lead to over-
neo-adjuant (before surgery) to cystectomy.             expression of c-myc and D-type cyclins, which
   Five-year survival in the United States and          favor cell proliferation. A type of NHL called
other developed countries is over 90% for local-        mantle cell lymphoma (MCL) exhibits a genomic
ized disease, 48% for regional disease, and 6%          deletion of the cell cycle checkpoint gene p16
for metastatic disease.4 In developing countries,       (INK4A) or a genomic amplification of the gene
overall 5-year survival is only 30%–50%35 be-           bmi-1 that codes for a repressor of the p16
cause detection and treatment are usually more          (INK4A) locus. Both of these alterations lead to
delayed.                                                loss of cell cycle checkpoint control. Still other
                                                        lymphomas lose cell genome integrity by dele-
                                                        tion or mutation of the ATM or p53 genes.
                                                           These genetic alterations are summarized in
Lymphomas are generally classified as Hodg-              Figure 3–7. A number of these genetic lesions
kin’s and non-Hodgkin’s (NHL) lymphomas, al-            involve pathways that will be seen again in other
though there are a number of subtypes of NHL            cancers.
that differ in their cellular morphology, response         Gene expression microarrays are now being
to chemotherapy, and prognosis. The new cases           employed to molecularly categorize a number of
annually in the United States are in the range          human cancers. One of the first practical dem-
of 7300 for Hodgkin’s disease and 56,000 for            onstrations of this was for NHL. Alizadeh et al.38
non-Hodgkin’s.3 Deaths due to NHL are about             showed that diffuse large B-cell lymphomas
20,000 annually, whereas only about 1400 deaths         (DLBCL) can be categorized by prognosis using
occur from Hodgkin’s disease in the United              gene arrays. Although clinical parameters can
States. Globally, about 62,000 cases of Hodgkin’s       also predict survival, gene expression arrays are
disease occur annually, but over 280,000 cases of       independent and perhaps more reliable predic-
NHL occur annually, predominantly in more de-           tors of prognosis. Furthermore, gene expression
veloped countries.36                                    profiles of subgroups of DLBCL demonstrate
   The strongest known risk factors for NHL are         that they are pathogenetically distinct diseases.
chromosomal translocations, the inciting cause             Lymphomas are as a class generally respon-
for which isn’t usually clear. Viral infection, e.g.,   sive to chemotherapy. The advent of the Mus-
with Epstein-Barr virus, human herpes virus 8,          targen (nitrogen mustard), Oncovin (vincristine),
or human T-lymphotropic virus-1 (HTLV-1),               prednisone, procarbazine (MOPP) regimen by
and acquired immunodeficiencies due to AIDS              De Vita and colleagues39 and subsequent vari-
or immunosuppressive drugs, for example, have           ations on this theme have led to a high cure rate
been suggested as causative.                            for Hodgkin’s lymphoma. In general, NHLs
   The molecular pathogenesis of lymphomas has          are also responsive to combination chemother-
been well studied.37 Because tissue is readily          apy, although somewhat less so than Hodgkin’s
       c-myc translocation
           BL, DLBCL
         c-myc mutations                               Translation/
                                        c-myc           glycolysis
           BL, DLBCL                   function
       c-myc transcription                                genes
           ABC DLBCL
     cyclin D1 translocation
              MCL                                                         Cell         Cell Growth
     cyclin D2 transcription                                             cycle
           ABC DLBCL
     cyclin D3 translocation                                                           Proliferation
Ink4a deletion/bmi-1 amplification      p16INK4a       Senescence
              MCL                        p14ARF          Bypass

      BCL-6 translocation               BCL-6              p27KIP1
           B-NHL                       function

        Pax5 translocation              PAX5
          LPCL, B-NHL                  function                                            Block

      BCL-2 translocation
        FL, GCB DLBCL
       BCL-2 amplification
       BCL-2 transcription
      NFKB2 translocation
          DLBCL, CTCL                                      BCL-2
      BCL10 translocation
    API2-MALT1 translocation
              MZL                                            A1
         I B mutations                                                                  Apoptosis
               HD                        NF- B
       I B kinase activity              function                                        Inhibition
        ABC DLBCL, HD                                     BCL-XL
           EBV LMP1
           PTLD, HD
           HTVL-1 tax
        ALK translocation                 Stat3
              ALCL                      function
      p53 mutation/deletion                                  p53
               NHL                                         function
Ink4a deletion/bmi-1 amplification                                                      Genomic
              MCL                        p14ARF
     ATM mutation/deletion
         MCL, SLL/CLL

           Figure 3–7. Pathogenetic mechanisms in lymphomas. Arrows indicate pre-
           sumptive target genes and pathways affected by oncogenic events in various
           lymphoma types. An enhanced version of this figure with references is available
           as Supplemental Figure S1 at /cgi /content /full /2 /5/
           363/DC1. Abbreviations: ALCL, anaplastic large-cell lymphomia; ATL, adult
           T-cell lymphoma; BL, Burkitt’s lymphoma; B-NHL, B-cell non-Hodgkin’s
           lymphoma; CLL, chronic lymphocytic leukemia; CTCL, cutaneous T-cell
           lymphoma; DLBCL, diffuse large B-cell lymphoma; EBV, Epstein-Barr virus;
           FL, follicular lymphoma; GCB, germinal-center B cell; HD, Hodgkin’s dis-
           ease; LPCL, lymphoplasmacytoid lymphoma; MZL, marginal zone lymphoma;
           PTLD, post-transplant lymphoproliferative disorder; SLL, small lympho-
           cytic lymphoma. (From Staudt and Wilson,37 reprinted by permission from
THE EPIDEMIOLOGY OF HUMAN CANCER                                                                          75

lymphoma. Development of monoclonal anti-             their differentiation to mature, functional blood
bodies such as rituximab target lymphoma cell         cells. These cells accumulate in the bone marrow
surface receptors and are showing responses in        and spill over into the peripheral blood, crowding
chemoresistant NHLs.                                  out their functional counterparts and leading to
                                                      symptoms such as fatigue, weight loss, repeated
                                                      infections, and excessive bruising and nosebleeds
                                                      that bring patients to the physician. Through ex-
Leukemias are classified on the basis of their cell    amination of the peripheral blood leukocytes and
type and chronicity of the disease. Thus, there are   bone marrow a cell type of origin of the disease
acute and chronic myeloid leukemias (AML and          can be detected. Although histological examina-
CML), and acute and chronic lymphocytic leu-          tion is often sufficient to classify the leukemia
kemias (ALL and CLL). Lymphocytic leukemias           type, gene expression microarrays are now able
can be either of the B-cell or T-cell type. Rarer     tostratify patients on the basis of molecular genetic
forms include monocytic, basophilic, eosinophilic,    profile, drug sensitivity, and prognosis.42
and erythroid leukemias. Multiple myeloma is             A variety of chromosomal abnormalities occur
an immunoglobulin-producing subtype of B-cell         in various leukemias. The classic example is the
leukemia. Survival rates vary greatly depending       Philadelphia chromosome in CML, which was
on the cell type and chronicity. ALL is largely       originally detected by Nowell and Hungerford
a disease of young childhood and thanks to ef-        and later found by Rowley to be a translocation
fective combination chemotherapy has a 70% or         involving a piece of chromosome 22 translocated
higher cure rate. AML, however, is substantially      to chromosome 9 (see Chapter 2). This t(9;22)
less curable and 5-year survival is only 20%–30%.     translocation creates a fusion gene bcr/abl that
   Worldwide, about 250,000 new cases and             codes for a proliferation-promoting tyrosine ki-
195,000 deaths occur each year.40 The causes          nase. It is this activated kinase that is the target of
of leukemia are largely unknown, although some        Gleevec, which induces complete remissions and
cases occur as secondary to earlier chemother-        increased survival in CML patients.43 One reason
apy or exposure to radiation (e.g., atomic bomb       for this targeted therapy working so well in CML
survivors) or chemicals (e.g., benzene). Infection    is that the bcr/abl translocation is a key, if not the
with the virus HTLV-1 is associated with adult        key, event initiating and maintaining the neoplas-
T-cell leukemia in tropical countries and Japan.      tic character of CML cells. There are very few
   Peak incidence of ALL is in the first 4 years of    such examples in cancer biology, and this may be
life, and while leukemia is often thought of as a     the reason it is so difficult to achieve therapeutic
childhood disease, it is diagnosed 10 times more      success with targeted therapy as a monother-
often in adults. After infancy, there is a decline    apy for other cancers. A number of other gene-
in leukemia incidence, but from ages 25 to 85         activating translocations occur in other leukemias,
there is a steep increase in incidence (Fig. 3–8).    and these may also provide important targets for
ALL accounts for 25% of all childhood malig-          drug development (Table 3–2).
nancies. AML is the most common leukemia in              As for lymphomas, the leukemias as a class are
young adults up to about age 45, whereas CLL is       generally responsive to chemotherapeutic agents,
the most common form in older adults.                 but overall survival rates differ markedly. For
   With a better understanding of the molecular       example, the 5-year survival rate for ALL is about
basis of leukemias41 has come the development of      70%, whereas it is only 20%–30% for AML and
so-called targeted chemotherapy, which is based       CML. Patients with CLL tend to have 5-year
on the determination of altered signal transduc-      survival rates in the 70% range. Five-year survival
tion pathways required and sufficient to cause         for multiple myeloma is 30%.40
(and maintain) the malignant phenotype. STI-
571 (Gleevec) is the first example of such suc-
                                                      Skin Cancer
cessful therapy (see below under Chronic My-
leogenous Leukemia).                                  Over one million cases of nonmelanoma skin can-
   Leukemias result from clonal expansion of im-      cers occur in the United States annually. These are
mature hematopoietic cells that are blocked in        mostly basal cell or squamous cell cancers and
76                                                                                                                    CANCER BIOLOGY

                                                                All types combined
          Age-specific incidence rate/100,000



                                                            10 9

                                                            15 4

                                                            20 9

                                                            25 4

                                                            30 9

                                                            35 4

                                                            40 9

                                                            45 4

                                                            50 9

                                                            55 4

                                                            60 9

                                                            65 4

                                                            70 9

                                                            75 4

                                                            80 9
















                                                                                     Age (years)

                                                 Figure 3–8. Age-specific incidence rates in the United States of leukemia
                                                 overall and of different subtypes. ALL, acute lymphoblastic leukemia; AML,
                                                 acute myeloid leukemia; CLL, chronic lymphocytic leukemia; CML, chronic
                                                 myelogenous leukemia. Note the high incidence of ALL among children.
                                                 (From World Cancer Report,40 with permission.)

are highly curable by local excision (the Mohs                                         to proximity to the equator. Of the 133,000 new
procedure). Melanoma is a different story. Over                                        cases annually in the world, 80% occur in North
59,000 new cases of melanoma and over 7500                                             America, Europe, and Australia.45
deaths occur every year in the United States,                                             A large proportion of persons diagnosed with
primarily in fair-skinned people.3 Chronic accu-                                       melanoma are cured by surgical excision, because
mulation of ultraviolet (UV)-induced skin damage                                       the lesions are usually diagnosed early while the
is thought to be the primary cause of basal and                                        cancer is still localized. However, once melanoma
squamous cell carcinomas, whereas multiple, in-                                        has metastasized, it is extremely difficult to treat
termittent severe burns, particularly during the                                       and almost always fatal. Melanoma is one of
young years of life, are thought to be the carcino-                                    the most immunogenic of human cancers and
genic event for melanoma. There is also a strong                                       immunomodulatory approaches have shown re-
hereditary link for melanoma, in addition to skin                                      sponses.
pigmentation.44                                                                           Alterations of chromosomes 1, 6, 7, and 10 are
    Melanoma is mostly a disease of people with                                        prevalent in melanoma but seem to be late in
light skin pigmentation. It is 10 times higher                                         tumor progression.44 Deletion of the tumor
among Caucasians than African Americans, and                                           suppressor gene p16 INK4A are frequently
it varies among Caucasian populations in relation                                      observed. Epigenetic inactivation of the p16
THE EPIDEMIOLOGY OF HUMAN CANCER                                                                                77

            Table 3–2. Common Recurrent Chromosomal Abnormalities in Acute Leukemia
            Type and Subtype of Leukemia                                   Abnormality   Genes Involved
            M0                 Acute myeloblastic leukemia with            inv(3q26)     EVI1
                                 minimal differentiation                   t(3;3)
            M2                 Acute myeloblastic leukemia with            t(8;21)       AML1-ETO*
                                 maturation                                t(6;9)        DEK-CAN
            M3                 Acute promyelocytic leukemia                t(15;17)      PML-RARa*
                                                                           t(11;17)      PLZF-RARa*
            M4                 Acute myelomonocytic leukemia               11q23         MLL{
                                                                           inv(3q26)     EVI1
                                                                           t(6;9)        DEK-CAN
            M4Eo               Acute myelomonocytic leukemia               inv(16)       CFBb-MYH11*
                               with abnormal eosinophils                   t(16;16)
            M5                 Acute monocytic leukemia                    11q23         MLL{
                                                                           t(8;16)       MOZ-CBP
            Secondary                                                      5q*           Unknown{
                                                                           7q*           Unknown{
                                                                           11q23         MLL{
                                                                           t(12;21)      TEL-AML1*
                                                                           t(1;19)       E2A-PBX1*
                                                                           t(4;11)       MLL{
                                                                           t(9;22)       BCR-ABL{
                                                                           t(8;14)       MYC{
            *Confers good prognosis.
             Confers poor prognosis.
            (From Stewart and Schuh,41 reprinted by permission from Elsevier.)

INK4A gene by promoter methylation has also                     distinct biology, prognosis, and therapeutic ap-
been observed. Cyclin D1 amplification has                       proach. In the United States, there are about
been detected, but curiously, this has only been                17,000 new intracranial tumors diagnosed each
observed in melanomas in relatively sun-                        year and about 13,000 deaths. Brain metastasis
protected areas of the body, suggesting a dif-                  from other primary sites is much more common,
ferent mechanism of melanoma initiation in                      with about 100,000 patients dying from this com-
these areas compared to sun-exposed areas of                    plication per year in the United States.46
the body. Interestingly, about 5% if melanoma                      Tumors of the CNS account for less than 2% of
cases clinically present as metastatic lesions                  all global malignancies (about 175,000 cases), and
with no known primary site, thus the immune                     the incidence does not vary widely among regions
system may have dealt with the primary skin                     of the world.47 The cause of CNS tumors is largely
location but had no effect on the metastases.44                 unknown. Certain hereditary syndromes produce
                                                                a susceptibility for CNS malignancies. These in-
                                                                clude neurofibromatosis, von Hippel-Lindau dis-
Cancers of the Central Nervous System
                                                                ease, tuberous sclerosis, and Li Fraumeni syn-
‘‘Brain tumors’’ are in reality a very diffuse col-             drome. The location, age of clinical manifestation,
lection of tumors of the central nervous system                 5-year survival, and associated genetic alterations
(CNS) (Table 3–3). Each of these tumors has a                   are shown in Table 3–4. Glioblastomas are the
78                                                                                                     CANCER BIOLOGY

         Table 3–3. Histologic Classification of Tumors of the Central Nervous System*
         Astrocytic tumors                                       Pineal parenchymal tumors
         Astrocytoma                                             Pineocytoma
         Anaplastic astrocytoma                                  Pineoblastoma
         Glioblastoma multiforme
         Pilocytic astrocytoma                                   Embryonal tumors
         Pleomorphic xanthoastrocytoma                           Medulloblastoma
         Subependymal giant-cell astrocytoma                     Primitive neuroectodermal tumor
         Oligodendroglial tumors                                 MENINGEAL TUMORS
         Oligodendroglioma                                       Meningioma
         Anaplastic oligodendroglioma                            Hemangiopericytoma
         Mixed gliomas                                           Melanocytic tumor
         Anaplastic oligoastrocytoma
                                                                 PRIMARY CENTRAL NERVOUS SYSTEM
         Ependymal tumors                                        LYMPHOMAS
                                                                 Germ Cell Tumors
         Anaplastic ependymoma
         Myxopapillary ependymoma                                Germinoma
         Subependymoma                                           Embryonal carcinoma
                                                                 Yolk-sac tumor (endodermal-sinus tumor)
         Choroid-plexus tumors                                   Choriocarcinoma
         Choroid-plexus papilloma                                Teratoma
         Choroid-plexus carcinoma                                Mixed germ-cell tumors
         Neuronal and mixed neuronal–glial tumors                TUMORS OF THE SELLAR REGION
         Gangliocytoma                                           Pituitary adenoma
         Dysembryoplastic neuroepithelial tumor                  Pituitary carcinoma
         Ganglioglioma                                           Craniopharyngioma
         Anaplastic ganglioglioma
         Central neurocytoma                                     METASTATIC TUMORS

         *This table has been abridged and modified from the World Health Organization classification.
         (From De Angelis,46 reprinted with permission from the Massachusetts Medical Society.)

most common brain tumors in adults, and they                     parts of the world. When populations move from
are really bad actors, because they are essentially              one country to another, the rates for many can-
incurable by surgery and are highly resistant                    cers tend toward that of the local population ra-
to radiation therapy and chemotherapy. Neuro-                    ther than that of their country of origin. A classic
blastomas are the most common neural-related                     example is the incidence rates among Japanese
tumors in children, and the outlook for this dis-                individuals living in Osaka, Japan, in contrast to
ease is much better. A number of these tumors                    those who have moved to Hawaii (Fig. 3–9).49
regress with age and the cure rate is over 90%                   Within a generation, the incidence for prostate,
for infants with the disease. A number of genetic                colon, and breast cancer begin to approach those
pathways have been implicated in the progression                 of the United States population, whereas the in-
of CNS tumors,48 holding out the hope that better                cidence of stomach cancer, more prevalent in
therapies will be developed in the future.                       Japan, decreases. Another interesting point from
                                                                 the data in Figure 3–9 is that from 1970–71 to
                                                                 1988–92, some of the ‘‘more Western cancers’’
ROLE OF VARIOUS FACTORS IN                                       became more prevalent in Japan, presumably be-
THE DEVELOPMENT OF CANCERS                                       cause of a more Westernized diet and lifestyle.
                                                                    Some specific causes of cancer are known, the
Most cancer types vary in incidence and mor-                     most prominent of which is cigarette smoking
tality among different populations in different                  (Table 3–5). However, the causes for the large
THE EPIDEMIOLOGY OF HUMAN CANCER                                                                                      79

Table 3–4. Summary of Epidemiological Data on Intracranial Tumors
                                                 Age at Clinical Manifestation
                                                         (% of Cases)       5-Year
Tumor                     Typical                                           Survival Rate   Genetic
(WHO Grade)               Location         0–20 years 20–45 years >45 years (% of Patients) Alterations
Pilocytic astrocytoma     Cerebellum,        74          20              6       >85            NF1 (neurofibromatosis
   (grade I)                optic nerve                                                          cases)
Low-grade             Cerebral               10          61            29        >50            p53 mutation
  diffuse astrocytoma   hemispheres
  (grade II)
Glioblastoma              Cerebral               3       25            72         >3            EGFR amplification,
  (grade IV)                hemispheres                                                          PTEN mutation, p16
                                                                                                 deletion, LOH
                                                                                                 chromosome 10
Oliogodendroglioma        Cerebral               8       46            46        >50            LOH 1p, 19q
  (grade II/ III)           hemispheres
Ependymoma                Ventricles,        37          38            25        <30            NF1 (spinal tumors)
  (grade IV)                spinal cord
Medulloblastoma           Cerebellum         74          23              3       >50            Isochromosome 17,
 (grade IV)                                                                                       mutations of p53,
                                                                                                  PTCH, b-catenin
Neuroblastoma             Abdomen          >95                                   >90 (<1 year   LOH 1p, 11q,
 (grade IV)                                                                       old) 20–50     MYCN amplification,
                                                                                  (>1 year)      trisomy 17q
LOH, loss of heterozygosity.
(From World Cancer Report,47 with permission.)

global variation in the most common cancers such                  medical, and social. The data are obtained from
as breast, prostate, and colon-rectum remain un-                  the International Agency for Research on Cancer
clear. In any case, the inescapable conclusion                    (IARC), based on studies of a working group that
from these data is that environmental and life-                   periodically accumulates and evaluates the evi-
style factors play the predominant role in cancer                 dence for human carcinogenesis. Although the
causation. That is not to say that genetic sus-                   information presented in Table 3–6 suggests that
ceptibility factors, many of which remain to be                   most human cancer is caused by occupational or
elucidated, are not important (see below). Yet                    medical exposures, this is not the case. Rather,
the implication here is that a high percentage of                 the cause of the bulk of human cancers is un-
cancers are preventable, or at least ‘‘delayable.’’ If            known, and only after certain discrete exposures
it is true as some say, that if we lived long enough              are well documented can cancer causation be
we’d all get cancer, then delayable may be a                      laid at the feet of a particular agent. Sometimes
better term than preventable. Since there are                     rare cancers such as the hepatic angiosarcomas
three billion nucleoside bases in every cell’s DNA                associated with occupational exposure to vinyl
and there are about 1014 cells in the human body,                 chloride call attention to certain agents. Most
the chances of one or a few of these bases being                  epidemiologists ascribe only about 2% to 5% of
misreplicated during cell division is enormous.                   human cancers to occupational exposures. The
Even if DNA editing and repair mechanisms were                    best estimates of the proportion of cancer deaths
to take care of most of these replication errors,                 due to various factors are shown in Table 3–5.
the chances that over a lifetime at least some of                 Surprisingly, some authors ascribe only about 1%
these would lead to malignant transformation in                   to 5% of cancer deaths to pollution.51 There
one or a few cells seems very likely indeed.                      is considerable debate about the role of environ-
    Table 3–6 illustrates the agents or circum-                   mental exposure to air and water pollutants in
stances for which there is good evidence of car-                  cancer causation; this is discussed in more detail
cinogenicity in humans. These agents are gener-                   later. Obviously, the ranges cited for each of the
ally divided into three categories: occupational,                 potential causes are very broad in some cases,
80                                                                                                                     CANCER BIOLOGY

                                                                       Osaka 1970–71           Hawaiian Japanese 1988–92
                                                                       Osaka 1988–92           Hawaiian Caucasian 1968–72
          Cumulate rate by age 75 (%)                                                          Hawaiian Caucasian 1988–92



                                                 Prostate           Colon(M)            Stomach(M)                Breast(F)
                                                                             Cancer type

                                        Figure 3–9. Cancer rates in migrants become similar to those in the local
                                        population. Cancer rates from 1988 to 1992 among Japanese migrants to
                                        Hawaii, and around 1970 and 1990 in Japan (Osaka) and in Hawaiian Cau-
                                        casians. Local rates for prostate, colon, and breast cancer increased over time
                                        (partly because of increased completeness of diagnosis and registration,
                                        particularly for prostate cancer in Hawaiian Causcasians) and stomach cancer
                                        decreased; but the effects of migration were larger. (From Peto,49 reprinted
                                        by permission from Macmillan Publishers Ltd.)

reflecting the degree of uncertainty about the                                    this form of tobacco use is generally considered
whole business of ascribing definitive causes at                                  much less dangerous because the smoke is
the current state of our knowledge.                                              usually not inhaled. A number of studies have
                                                                                 also suggested a correlation between ‘‘passive

Cigarette Smoking
                                                                                 Table 3–5. Estimated Percentage of Total Cancer
As noted above, epidemiologists have attributed                                  Deaths Attributable to Established Causes of Cancer
as many as 30% of all cancer deaths to tobacco
use, primarily cigarette smoking. In 2005, for                                   Risk Factor                                        Percentage
example, approximately 163,000 deaths due to                                     Tobacco                                            30
lung cancer were expected to occur in the United                                 Adult diet and obesity                             30
                                                                                 Sedentary lifestyle                                 5
States (90,000 men and 73,000 women), making                                     Occupational factors                                5
up about 31% of all deaths from cancer in men                                    Family history of cancer                            5
and 27% in women.3 Although the most direct                                      Viruses and other biologic agents                   5
                                                                                 Perinatal factors and growth                        5
correlation is between cigarette smoking and                                     Reproductive factors                                3
lung cancer, tobacco use has also been implicated                                Alcohol                                             3
in cancers of the mouth, pharynx, larynx, esopha-                                Socioeconomic status                                3
                                                                                 Environmental pollution                             2
gus, urinary bladder, pancreas, kidney, and, more                                Ionizing and ultraviolet radiation                  2
recently, stomach and liver49 and perhaps colo-                                  Prescription drugs and medical procedures           1
rectal cancer.52 Smoking of pipes or cigars has                                  Salt, other food additives, and contaminants        1
been implicated in the occurrence of cancers of                                  *Estimates according to Harvard Center for Cancer Prevention.
the mouth, pharynx, larynx, and esophagus, but                                   (From America’s War on Carcinogens50)
THE EPIDEMIOLOGY OF HUMAN CANCER                                                                                               81

Table 3–6. Established Human Carcinogenic Agents and Circumstances
                                                          Type of Exposure*
Agent or Circumstance                          Occupational             Medical      Social    Site of Cancer
Aflatoxin                                                                             þ         Liver
Alcoholic drinks                                                                     þ         Mouth, pharynx, larynx,
                                                                                                 esophagus, liver
Alkylating agents
  Cyclophosphamide                                                      þ                      Bladder
  Melphalan                                                             þ                      Marrow
Aromatic amines
  4-Aminodiphenyl                              þ                                               Bladder
  Benzidine                                    þ                                               Bladder
  2-Naphthylamine                              þ                                               Bladder
Arsenic{                                       þ                        þ                      Skin, lung
Asbestos                                       þ                                               Lung, pleura, peritoneum
Benzene                                        þ                                               Marrow
Bis(chloromethyl) ether                        þ                                               Lung
Busulfan                                                                þ                      Marrow
Cadmium{                                       þ                                               Prostate
Chewing (betel, tobacco, lime)                                                       þ         Mouth
Chromium{                                      þ                                               Lung
Chlornaphazine                                                          þ                      Bladder
Furniture manufacturer (hardwood)              þ                                               Nasal sinuses
Immunosuppressive drugs                                                 þ                      Recticuloendothelial system
Ionizing radiations{                           þ                        þ                      Marrow and probably all other sites
Isopropyl alcohol manufacture                  þ                                               Nasal sinuses
Leather goods manufacture                      þ                                               Nasal sinuses
Mustard gas                                    þ                                               Larynx lung
Nickel{                                        þ                                               Nasal sinuses, lung
  Unopposed                                                             þ                      Endometrium
  Transplacental (DES)                                                  þ                      Vagina
Overnutrition (causing obesity)                                                      þ         Endometrium, gallbladder
Phenacetin                                                              þ                      Kidney (pelvis)
Polycyclic hydrocarbons                        þ                        þ                      Skin, scrotum, lung
Reproductive history:
  Late age at first pregnancy                                                         þ         Breast
  Zero or low parity                                                                 þ         Ovary
  Schistosoma haematobium                                                            þ         Bladder
  Chlonorchis sinensis                                                               þ         Liver (cholangioma)
Sexual promiscuity                                                                   þ         Cervix uteri
  Anabolic (oxymetholone)                                               þ                      Liver
  Contraceptives                                                        þ                      Liver (hamartoma)
Tobacco smoking                                                                      þ         Mouth, pharynx, larynx, lung,
                                                                                                 esophagus, bladder
Ultraviolet light                              þ                                     þ         Skin, lip
Vinyl chloride                                 þ                                               Liver (angiosarcoma)
Virus (hepatits B)                                                                   þ         Liver (hepatoma)
*A plus sign indicates that evidence of carcinogenicity was obtained.
Certain compounds or oxidation states only.
For example, from X-rays, thorium, Thorotrast, some underground mining, or other occupations.
(From Doll and Peto51)

smoking’’—that is, exposure to smoke and others’                            over a number of years. The vast majority of these
cigarettes in the home or workplace—and lung                                studies indicate cigarette smoking as a major cause
cancer, but the data for this are questionable (see                         of lung cancer. Epidemiologic studies, autopsy
The Great Cancer Myths, below). An enormous                                 reports, and experimental animal data reviewed
amount of research on the relationship between                              in the original U.S. Surgeon General’s Report of
tobacco smoking and cancer has been carried out                             1964 and in subsequent U.S. Department of
82                                                                                        CANCER BIOLOGY

Health, Education, and Welfare reports strongly           radioactive polonium210Po, hydrazine, urethan,
support a casual relationship. The data can be            formaldehyde, nitrogen oxides, and nitrosodi-
summarized as follows.53                                  ethylamine.56 Moreover, mutagenic activity is
                                                          5- to 10-fold higher in the urine of smokers than
     1. A strong relationship between cigarette
                                                          that of nonsmokers.57
        smoking and lung cancer mortality in men
                                                             An increased incidence of chromosomal ab-
        has been demonstrated in numerous pro-
                                                          normalities has been observed in smokers’ pe-
        spective and retrospective studies, with risks
                                                          ripheral blood lymphocytes compared to lym-
        for all smokers as a group ranging from
                                                          phocytes of nonsmokers.58,59 A meta-analysis of
        11 to 22 times those of nonsmokers.54
                                                          DNA adducts present in peripheral white blood
     2. A dose–response relationship between
                                                          cells or tissue of smokers who have cancer com-
        cigarette consumption and risk of devel-
                                                          pared to smokers that don’t showed that current
        opment of lung cancer for both men and
                                                          smokers with high levels of adducts have an in-
        women has been demonstrated in numer-
                                                          creased risk of lung and bladder cancers.60 How-
        ous studies, with risk being much higher
                                                          ever, only a fraction of smokers develop lung
        for men and women who are heavy smok-
                                                          cancer, thus suggesting individual variability in
        ers than for nonsmokers. Light smokers
                                                          susceptibility. A study by Wei et al.61 found that
        have an intermediate risk.54,55
                                                          a low DNA repair capacity correlated with in-
     3. Mortality from lung cancer directly attrib-
                                                          creased risk of lung cancer in a population of
        utable to cigarette smoking is increased
        in the presence of urbanization and such
                                                             There is a worldwide epidemic in lung cancer
        occupational hazards as uranium mining
                                                          that reflects the increasing amount of tobacco
        and exposure to asbestos.
                                                          use in the world. Over 900,000 new cases of lung
     4. Cessation of smoking results in lowered
                                                          cancer are reported yearly by the World Health
        risk or mortality from lung cancer in com-
                                                          Organization (WHO).6 From 1960 to the 1980s,
        parison with continuation of smoking.54,55
                                                          mortality rates from lung cancer in women in-
     5. Results from autopsy studies show that
                                                          creased more than 100% in Japan, Norway, Po-
        changes in the bronchial mucosa that are
                                                          land, Sweden, and the United Kingdom; more
        thought to precede development of bron-
                                                          than 200% in Australia, Denmark, and New
        chogenic carcinoma are more common in
                                                          Zealand; and over 300% in Canada and the Uni-
        smokers than in nonsmokers, and there
                                                          ted States.62 A WHO survey in the late 1980s also
        is a dose–response relationship for these
                                                          showed that one in three males about age 15 in
                                                          developed countries smoke cigarettes and one-
     6. Chronic inhalation of cigarette smoke or
                                                          half of such individuals smoke in developing
        the intratracheal instillation of various frac-
                                                          countries. The rates for females are lower but
        tions of tobacco smoke produce lung can-
                                                          rising rapidly. It is estimated that the incidence of
        cer in such experimental animals as dogs
                                                          lung cancer in developing countries will be the
        and hamsters.
                                                          same as in developed nations in 40 years.62 And
     7. Cell culture studies show that various
                                                          this is just for lung cancer. When one considers
        constituents found in tobacco and ciga-
                                                          the other tobacco-related cancers, the number of
        rette smoke condensate produce malignant
                                                          preventable new cancer cases could be more than
        transformation of cells.
                                                          double that estimated for lung cancer alone.
     8. Numerous complete carcinogens and co-
                                                          Clearly, unless smoking habits and other lifestyle
        carcinogens (tumor promoters) have been
                                                          and pollution factors change, the future looks
        isolated from cigarette smoke conden-
                                                          bleak for attempts to prevent cancer.
                                                             An encouraging trend, however, has been
  Numerous mutagens and carcinogens have                  noted in the United States, United Kingdom,
been identified in the particulate or vapor pha-           Australia, and a few other countries, which may
ses of tobacco smoke; these include benzo(a)-             encourage other nations to reduce tobacco con-
pyrene, dibenza(a)anthracene, nickel, cadmium,            sumption. Lung cancer rates in men under age
THE EPIDEMIOLOGY OF HUMAN CANCER                                                                          83

54 started to decline in the 1980s in the United         even moderate consumption (less than three
States, and there has been a leveling off or de-         drinks per week) was associated with increased
cline of mortality rates reflecting that. In the          risk. However, considering all the data, a clear
oldest age groups, however, a steady increase in         relationship between alcohol consumption and
lung cancer mortality rates continues, because           breast cancer at such low doses is not clear. An
of a cohort of individuals who started smoking in        increased risk of breast cancer has been reported
high numbers at an early age. The numbers for            in women who consume one or more drinks per
women are less optimistic since they started             day, but this increased risk is alleviated by high
smoking later in the twentieth century as a group        folate intake.
than men, and the 20-year lag in lung cancer is
still catching up with them. Thus, lung cancer
mortality rates in women are expected to climb
for a number of years to come in much of the             Dietary factors are postulated to account for
world. Although the decline in lung cancer in-           about 30% of cancers in developed countries and
cidence and mortality rates in some developed            about 20% in developing countries.68 This would
countries is encouraging, there is a disturbing          make diet the second leading risk factor after
upward trend in smoking rates among adoles-              tobacco. However, the exact dietary compo-
cents that doesn’t bode well for the future.63           nents that either increase or decrease cancer risk
                                                         are not clear, despite numerous case–control and
                                                         prospective studies on dietary factors. The best-
                                                         known link between diet and cancer is the cor-
Alcohol is thought to interact with smoking in the       relation of obesity with the increased incidence
causation of certain cancers, particularly oral and      of various cancers such as esophagus, colorec-
esophageal cancers. Alcohol appears to be syn-           tum, breast, endometrium, and kidney.68 Never-
ergistic with tobacco in causing cancers of the          theless, the large differences in cancer rates
mouth, pharynx, larynx, and esophagus, but not           among various countries and regions of the world
that of the lung.64,65 In liver cancer, there is good    suggest that diet and lifestyle are key to explain-
evidence that alcohol consumption sufficient              ing these differences. Some dietary risk factors
to cause cirrhosis of the liver increases the inci-      and dietary protective factors are listed in Table
dence of liver cancers, perhaps secondary to the         3–7. A number of probabilities and possibilities
chronic damage to the liver caused by alcohol            are listed in the table because the data are for the
abuse. Pure alcohol is not by itself carcinogenic        most part speculative or inconclusive. Some of
in animals and may exert its carcinogenic effect         the data for various dietary components are dis-
secondarily to tissue damage, as in the case of          cussed below.
hepatic cirrhosis, or by facilitating uptake of car-        Dietary fat intake has long been thought to be
cinogens by exposed tissues, as may be the case          a factor in the causation of various cancers in-
for oral and esophageal cancer.51 Other potential        cluding breast, prostate, and colon cancer. This
mechanisms include (1) a carcinogenic effect of          link has been supported by comparison of av-
other chemicals such as N-nitrosamines in alco-          erage fat intake among high-incidence and low-
holic beverages; (2) a solvent action that facilitates   incidence countries of the world, particularly for
absorption of carcinogens found in tobacco               breast cancer. In a large Nurses’ Health Study
smoke; and (3) a carcinogenic effect due to acetal-      that examined the relationship between dietary
dehyde, a major metabolite of ethanol. Support           fat and breast cancer risk over a 14-year period
for the latter concept includes data from Japanese       in 3000 women, there appeared to be a weak
populations in which a genetic polymorphism              overall increased risk due to total dietary fat.69
that results in high circulating acetaldehyde levels     However, the results of six other large pro-
is associated with increased cancer risk.66              spective studies that examined the relation-
   A small positive association between alcohol          ship between fat intake and breast cancer all
and breast cancer risk has been seen in some but         showed a weak or no association. A pooled anal-
not all epidemiological studies.67 In some studies,      ysis of all these studies indicated no significant
84                                                                                                    CANCER BIOLOGY

Table 3–7. Dietary Risk Factors, Dietary Protective Factors, and Other Major Risk Factors for Common
                          Dietary and Diet-Related                           Dietary Protective       Other Major
Cancer                    Risk Factors                                       Factors                  Risk Factors
Oral cavity,              Alcohol                                            Probably fruit and       Smoking
  pharynx and             Very hot drinks                                      vegetables
  esophagus               Obesity (adenocarcinoma of the oesophagus)
                          Chinese-style salted fish (nasopharyngeal cancer)
Stomach                   Probably high intake of salt-preserved             Probably fruit and       Infection by
                            foods and salt                                     vegetables               Helicobacter pylori
Colorectum                Obesity                                            Probably fruit and       Sedentary lifestyle
                          Possible red and processed meat                      vegetables and other
                                                                               plant foods
                                                                               rich in fiber
Liver                     High alcohol intake                                None established         Hepatitis viruses
                          Foods contaminated with aflatoxins
Pancreas                  None established                                   None established         Smoking
Larynx                    Alcohol                                            None established         Smoking
Lung                      None established                                   Possibly fruit &         Smoking
Breast                    Obesity after menopause                            None established         Reproductive and
                          Alcohol                                                                       hormonal factors
Endometrium               Obesity                                            None established         Low parity
Cervix                    None established                                   None established         Human papillomavirus
Prostate                  None established                                   None established         None established
Kidney                    Obesity                                            None established         None established
From Key et al.,        with permission.

relationship between fat intake and breast can-                     and vegetables have the greatest risk of colo-
cer. Similar conclusions have been reached for                      rectal cancer71 and that high consumption of
colon cancer risk.70                                                cruciferous vegetables is associated with reduced
   Another widely held belief is that high dietary                  risk of prostate cancer.72 In addition, serum lev-
fiber, particularly from grains, reduces colon can-                  els of selenium have been shown to correlate
cer risk. Again, a number of studies do not sup-                    with decreased risk of esophageal and gastric
port this hypothesis.69 Furthermore, no evidence                    cancers.73 Selenium74 and vitamin E75 have been
has been found that dietary fiber is related to the                  reported to decrease risk for prostate cancer.
occurrence of colon adenomas in either men or                          One relationship of cancer to diet that is
women. An inverse relationship, however, was                        clearer than a lot of other data is the role of folic
found between high dietary fiber and the risk of                     acid. A number of studies have indicated that
diverticular disease of the colon as well as coro-                  the intake of folic acid reduces cancer risk (re-
nary heart disease.                                                 viewed in Reference 69). For example, a 15-year
   High consumption of fruits and vegetables has                    follow-up of the Nurses’ Health Study showed a
also been thought to have beneficial effects in                      reduction in colon cancer risk with folate sup-
reducing cancer risk, according to case–control                     plementation. The mechanism for this effect isn’t
studies. In recent cohort studies the association                   clear, but two theories are (1) there is a folate-
was much weaker. In a Nurses’ Health Study                          mediated production of methyl donors (via me-
and Health Professionals follow-up study, no                        thionine) for DNA methylation, which regulates
overall association between dietary intake of fruits                gene expression (see Chapter 5); and (2) there is
and vegetables and colon cancer was observed.69                     a facilitated conversion of uracil to thymine for
Some studies, however, have found that indi-                        DNA synthesis and repair. High folate intake was
viduals who consume very low amounts of fruits                      found to protect against some of the carcinogenic
THE EPIDEMIOLOGY OF HUMAN CANCER                                                                      85

effects of alcohol consumption; protection was        agent in breast cancer. Studies in blood levels and
seen for colon cancer in men and breast cancer        urinary excretionpatternsofhormonesinpatients
in women.69                                           with breast cancer or women at risk to develop
   Because there are no clear dietary factors that    breast cancer have yielded conflicting results.
explain the global diversity of cancer incidence      However, in a study of women whose mothers
and mortality, one might then ask what lifestyle      had bilateral breast cancer and who thus had a
factors are involved. Environmental pollution         high familial risk of also developing the disease,
and occupational hazards aside (because they          it was concluded that this increased risk was
only contribute a relatively small percentage to      associated with elevated plasma levels of pro-
overall risk), there are some correlations that       lactin, progesterone, and estrogen.79
may explain some of this diversity. These are            In men, late descent of the testes is associated
obesity and physical activity, or to put it another   with an increased susceptibility to testicular can-
way, energy intake versus output. Studies in an-      cer, for reasons that are not clear, although this
imals suggest that energy (caloric) restriction has   may reflect some faulty differentiation response
a powerful influence on tumor formation. For           in the testicular tissue. This cancer usually oc-
example, a 30% restriction in caloric intake can      curs in younger men (average age at diagnosis is
reduce mammary tumors in rats by 80% (re-             32 years).
viewed in Reference 69). Although an exact mea-          Reproductive patterns are related to cancers
surement of energy balance in humans is hard to       of the uterine endometrium, ovary, and breast,
come by, there are significant data correlating        all three of which are less common in women
obesity and lack of physical activity as cofactors    who have had children, particularly if early in
in colon cancer in men and women and in breast        their reproductive lives, than in women who have
cancer.69                                             not had children.
                                                         Cancer of the uterine cervix is associated with
                                                      early and frequent sexual contact with a variety
Sexual Development, Reproductive
                                                      of partners. There is strong evidence for an as-
Patterns, and Sexual Behavior
                                                      sociation of a transmissible virus, human papil-
The duration of hormonal exposure appears to          loma virus, with cervical cancer (see Chapter 2).
play a role in the susceptibility to breast cancer
in women. The carcinogenic effects of hormones
                                                      Industrial Chemicals and
were first demonstrated in animals. In 1932,
                                                      Occupational Cancers
Lacassagne reported the induction of mammary
carcinomas in mice injected repeatedly with an        The chemicals and industrial processes that have
ovarian extract containing estrogen. Later, he        a known or suspected etiologic role in the de-
also showed that the synthetic estrogen diethyl-      velopment of cancer are listed in Table 3–6. As
stilbestrol produced mammary tumors in suscep-        noted above, about 2%–5% of all cancer deaths
tible strains of mice.76 Furthermore, ovariecto-      are attributed to occupational hazards. Of those
mized mice and rats have a decreased frequency        agents listed as carcinogenic for humans, a num-
of breast cancer, whereas rodents subjected to        ber were identified because of their close asso-
increased levels of estrogen, progesterone, and       ciation between an abnormal clustering of certain
prolactin have an increased frequency of breast       cancers and exposure to an industrial chemical or
cancer, although timing of exposure to individ-       process. For example, epidemiologic studies of
ual hormones appears to be crucial.77,78 Simi-        workers occupationally exposed to industrial lev-
larly in humans, a role of hormones in the devel-     els of 4-aminobiphenyl have a higher incidence
opment of breast cancer has been deduced from         of bladder cancer.80 Occupational exposure to
the known risk factors associated with the dis-       asbestos fibers results in a higher incidence of
ease. These factors include early age of menar-       lung cancer, mesotheliomas, gastrointestinal tract
che, delayed age of first pregnancy, and delayed       cancers, and laryngeal cancers.81 As mentioned
menopause, suggesting longer duration of ex-          earlier, cigarette smoking and occupational ex-
posure to hormonal stimulation as an etiologic        posure to asbestos act synergistically to increase
86                                                                                   CANCER BIOLOGY

the incidence of lung cancer. Several epidemio-         of PAH and DNA repair mechanisms, smoking
logic studies have shown increased frequency            habits, amount of air pollution with PAH in the
of leukemia in workers exposed to benzene.81–83         place of residence, and amount of PAH in the
Two studies of workers exposed to bis(chlorome-         drinking water and diet.
thyl) ether have indicated an increased risk of
lung cancer, primarily small-cell carcinoma.81
There is also an increased risk of lung cancer
among workers in chromium industries. Occu-             Herbicides are a heterogeneous class of che-
pational exposure to 2-naphthylamine has long           micals widely used in agriculture, forestry, and
been known to be associated with urinary bladder        gardening to kill undesirable weeds and foliage.
cancer.81 Since the time of Percival Pott and his       Although agricultural workers and workers in the
study of chimney sweeps (see Chapter 2), coal           plants that manufacture them are exposed to
soot has been known as a cause of skin cancer.          the highest concentrations, the entire popula-
Since that time, occupational exposure to soot,         tion is probably exposed to some level of her-
coal tar, pitch, coal fumes, and some crude shale       bicide contamination, albeit a low level. This
and cutting oils has been shown to be associated        could come about from residual contamination
with cancers of the skin, lung, bladder, and gas-       of foodstuffs, runoff into ground water used for
trointestinal tract. The carcinogenicity of these       drinking supplies, or airborne contamination in
latter agents is probably related to their content of   areas of heavy spraying.
polycyclic aromatic hydrocarbons (PAH).                    Herbicides that have been used commercially
   The highest levels of human exposure to PAH          include the phenoxy compounds 2,4-dichloro-
occur in industrial processes involving the use of      phenoxyacetic acid (2,4-D) and 2,4,5-trichloro-
coal tar and pitch and in the production of coke        phenoxyacetic acid (2,4,5-T; also known as Agent
from coke ovens.84 Epidemiological data indi-           Orange), triazines, amides, benzoics, carbamates,
cate that ambient coal tar and pitch in iron and        trifluralin, and uracils. 2,4,5-T, and to some ex-
steel foundries contains carcinogenic substances        tent 2,4-D, preparations were contaminated
and may lead to an increased incidence of lung          with dioxins and furans, particularly prior to
cancer, particularly in smokers who work in             1975 when government manufacturing restric-
such environments. A study by van Schooten              tions limited the amount of allowable contami-
et al.84 showed that coke-oven workers were             nation. One contaminant, 2,3,7,8-tetrachlorodi-
exposed to substantial concentrations of PAH in         benzo-p-dioxin (TCDD), is a potent mutagen
the air, including benzo[a]pyrene and pyrene.           and a powerful carcinogen in animal studies.
Forty-seven percent had detectable levels of            Prior to 1975, TCDD concentrations of 1 part
PAH-DNA adducts in their white blood cells,             per million (ppm) were observed in commercial
compared to 30% of control subjects who worked          phenoxy herbicide preparations, whereas cur-
in another part of the plant. In both groups,           rent levels are below 0.1 ppm.85
smokers had significantly higher levels of PAH              A study of phenoxy herbicide applicators has
adducts than nonsmokers. Since the carcino-             shown detectable blood levels of TCDD.85 Be-
genic mechanisms of PAH involves metabolic              cause TCDD is stored in fat tissue, its half-life
activation and alteration of DNA function (see          in the body may be as long as 7 years. Calendar
Chapter 2), these data suggest that exposure            period of exposure and intensity of use of 2,4,5-
to various carcinogens and/or susceptibility to         T were determinants of serum levels of TCDD.
their DNA-damaging effects could be moni-               TCDD serum levels were also associated with
tored by measuring DNA-adduct formation in              intensity of exposure to 2,4-D, but this was
peripheral white blood cells or perhaps in urine,       confined to individuals exposed before 1975.
if sensitive and specific enough assays could be         Based on the assumed half-life of 7 years, some
developed. It should be noted that such mea-            workers would have had serum TCDD levels of
surements would be subject to individual vari-          up to 329 parts per trillion (ppt); a maximum
ations relating to variation in daily exposure lev-     of 26 ppt has been reported in the general pop-
els, genetic differences in metabolic activation        ulation. In a Vietnam veteran heavily exposed to
THE EPIDEMIOLOGY OF HUMAN CANCER                                                                      87

Agent Orange, serum levels of 1530 ppt were
                                                      Air and Water Pollutants
estimated to have been reached during the time
of his peak exposure. Adipose tissue concentra-       Air, water, and soil pollution is estimated to
tions of 540 ppt have been shown to cause thy-        account for only 1%–4% of all cancers. A small
roid cancers in animals.85                            percentage of lung cancer (less than 5%) may be
   There is considerable controversy over the         due to chronic inhalation of outdoor air pollut-
long-range health effects of exposure to 2,4,5-T,     ants such as industrial or engine exhaust che-
but it is clear that some individuals heavily ex-     micals. Indoor air pollutants such as secondhand
posed to 2,4,5-T prior to 1975 would most likely      smoke and radon are thought to be contributors,
have achieved body concentrations of TCDD that        but this risk is most likely exaggerated (see be-
have been shown to be carcinogenic in some            low). In China and some other Asian countries,
animals.85 As yet, however, there is not conclusive   chronic inhalation of cooking oil smoke may be
evidence for TCDD carcinogenicity in humans.86        a causative agent of lung cancer.90 The contam-
The manufacture of 2,4,5-T has been discontin-        ination of the atmosphere by chlorofluorocar-
ued in most Western countries and its use banned      bons (whose production is now banned in de-
in the United States since 1983.                      veloped countries) in refrigerant and propellants
   The association of herbicide use with cancer in    has been implicated in destruction of the ozone
humans has been reviewed.87 Review of studies         layer and a resultant increase in skin cancer due
from several countries and states in the United       to a lower filtering of UV irradiation from the
States shows significant evidence supporting a         sun. Occupational exposure to inhaled asbestos,
relationship between non-Hodgkin’s lymphoma           such as occurred in Liberty Ship building in
in farmers and exposure to phenoxy herbicides.        World War II, has been clearly linked to me-
Several studies have also reported a relationship     sothelioma.
to increased risk of soft tissue sarcomas with ex-       Regarding water pollution, high exposure to
posure to phenoxy herbicides. Although also im-       arsenic in drinking water in certain countries
plicated in some studies to be related to increased   (e.g., Bangladesh) and areas of the United States
risk of cancers of the colon, lung, nasal passages,   (Alaska) and South America (Argentina, Chile)
prostate, and ovary as well as leukemia and mul-      appears to be related to an increased risk of
tiple myeloma, there have been too few definitive      bladder and skin cancers.90 A number of other
studies to demonstrate an exposure–risk rela-         groups of water pollutants have been investi-
tionship. An increased risk of non-Hodgkin’s          gated as possible sources of cancer risk, but the
lymphoma has been associated with chronic ex-         data are not conclusive, even though a popular
posure to a number of pesticides including 2,4-D,     myth is that the contents of our drinking water
mecoprop, dicamba, and malathione.88                  are causing cancer.
   Another dilemma is that experimental animal           Evidence that potential carcinogens in the air
studies don’t convincingly demonstrate the car-       or water might cause cancer is based on several
cinogenicity of 2,4-D and 2,4,5-T.89 Thus it is       assumptions as well as on epidemiologic data.
difficult to develop a consistent public policy on     One of the assumptions is that there is a lin-
the useofsuchsubstances,particularlywhen most         ear, nonthreshold dose–response relationship
regulatory decisions are made on carcinogenicity      between the given dose of carcinogen and the
testing in animals. Nevertheless, it seems only       number of cases of cancer. This assumption is
prudent to carefully monitor exposure to herbi-       based primarily on dose–response studies in ex-
cides and other occupationally related chemicals      perimental animals. Such a dose–response re-
and to promote minimal exposure safety prac-          lationship carries the implication that there is no
tices, particularly among farmers and other work-     such thing as a safe level of exposure to a car-
ers who may experience high exposure. In the          cinogen (discussed in Chapter 2). Taking the
United States, the National Cancer Institute, in      nonthreshold approach to evaluation of expo-
collaboration with the Environmental Protec-          sure to environmental agents is, of course, the
tion Agency, is undertaking a long-term cohort        most conservative policy; it tends to predict
study of pesticide-exposed farmers.                   the largest response (i.e., the largest number of
88                                                                                  CANCER BIOLOGY

cancers) for any given level of low-dose expo-        or dumping of chemical wastes on or near bodies
sure. Since the possible consequences of expo-        of water can contribute significant levels of haz-
sure to carcinogens in the general environment        ardous substances to public water supplies.
are so enormous, a number of investigators think         The Environmental Protection Agency and
that it is appropriate to use this approach. Al-      other groups have undertaken studies of several
though this approach seems reasonable to en-          large metropolitan areas to evaluate the level of
vironmentalists, currently only limited evidence      contamination of public drinking water supplies
supports it. Evidence from air pollution studies,     and to assess the carcinogenic risk associated
for example, indicates that estimates of cancer       with this contamination. In a survey of 80 cities,
risk by extrapolation of dose–response relation-      a number of potentially dangerous trihalometh-
ships may be an oversimplification of the prob-        anes, including chloroform, bromodichloro-
lem. Large metropolitan areas have a substan-         methane, dibromochloromethane, and bromo-
tially higher level of atmospheric carcinogens,       form, were detected.97 Chloroform, a known
such as benzo[a]pyrene, resulting from com-           carcinogen in animals, was found in the drinking
bustion of fossil fuels, than rural areas, yet some   water of 80 cities. Carbon tetrachloride, also a
studies91 show that nonsmokers in urban areas         known carcinogen, was found in the drinking
do not have a significantly higher risk of lung        water of 10 cities. In one survey98, 325 organic
cancer than that of rural nonsmokers. However,        chemicals were identified in the drinking water of
urban smokers do have a significantly higher           various cities. Only about 10% of these have been
incidence of lung cancer than comparably heavy        adequately tested for carcinogenicity. Among
smokers in rural areas. These observations and        the known or suspected carcinogens identified
others, such as the potentiation of lung cancer in    in drinking water are benzene, bis(chloro-
uranium miners92 and asbestos workers93,94 who        methyl) ether, carbon tetrachloride, chloro-
smoke, support the idea that a combination of         form, dieldrin, polychlorinated biphenyls, 1,1,2-
urban air pollution and smoking is the most           trichloroethylene, and vinyl chloride. Thus, it is
carcinogenic.                                         evident that the general public is exposed to a
   Numerous potential carcinogens have been           wide variety of environmental chemical carcino-
found in air and water, particularly in areas near    gens. Since there is a 20- to 30-year latent period
or downstream from large industrial complexes.        between exposure to certain carcinogenic agents
For example, nitrosamines, a class of chemicals       and the development of clinically detectable
thatare among the mostpotent carcinogensknown         cancer, it will probably take several decades to
from experimental animal studies, are present in      fully evaluate the impact of our contaminated
the environment, albeit usually at very low con-      environment. It should be pointed out, however,
centrations.                                          that the expected correlations between exposure
   In addition to industrial sources, domestic sew-   to a given carcinogen in the drinking water and
age treatment plant effluents may contain carci-       the type of cancer expected to result from such
nogenic substances that may find their way into        exposure have not been established. For exam-
drinking water supplies. More than 50 chlori-         ple, even though chloroform, a hepatocarcino-
nated hydrocarbons have been identified in do-         gen, is the predominant organic contaminant in
mestic sewage effluents.95 This same study esti-       the drinking water of certain communities in
mated that over 1000 tons of chlorinated organic      Louisiana that take their water from the Mis-
compounds are discharged by sewage treatment          sissippi, there is no increased mortality from he-
plants into the nation’s waterways annually.          patic cancers in those communities.99
Chlorinated hydrocarbons result from the chlo-           Nevertheless, a considerable debate over the
rination of water heavily polluted with organic       role of environmental pollutants in human cancer
chemicals.96 Some of these chlorinated com-           continues. On the basis of studies of cancer in-
pounds are known to be carcinogenic in animals.       cidence in various regions in Africa, Higginson
   Although discharges from industrial and mu-        and Oettle100 provided some definitive data on
nicipal waste treatment plants may be continu-        the impact of environmental factors in the cau-
ous sources of pollution, spills resulting from       sation of human cancer. The work of Higginson
industrial operations, transportation accidents,      and colleagues has generally been credited with
THE EPIDEMIOLOGY OF HUMAN CANCER                                                                       89

establishing the fact that about two-thirds of all    there is a much lower incidence of these malig-
human cancers have an environmental cause and         nant diseases. At present, however, it is unjusti-
thus, theoretically at least, are preventable. This   fied to link these differences in incidence directly
has led many people to believe that the environ-      to recent food additives or chemical pollutants.
mental agents responsible for cancer are chemi-       Lifestyle differences appear to play a large role in
cals that we inhale or ingest. However, as Hig-       the causation of these and other cancers.101 For
ginson himself has reiterated,101,102 what he         example, the varying incidence of cancer of the
meant by ‘‘the environment’’ is the total milieu in   breast, ovary, and uterus can be related at least
which people live, including cultural habits, diet,   partly to differences in average age at onset of
exposure to various infectious agents, average age    menarche, sexual behavior, and reproductive
of menarche, number of children a woman bears,        patterns among different population groups. Ta-
age of menopause—in short, the cultural as well       ken together, all the data accumulated to date
as the chemical environment.                          suggest that cancer distribution patterns repres-
   Although we have seen that clear correlations      ent a variety of differences in lifestyle, with
between excess occupational exposure to carci-        exposure to chemical pollutants in the ambient
nogenic chemicals and some cancers can be             environment of industrialized societies contrib-
made, the contribution of these occupationally        uting to some but an as-yet unclear percent of
related cancers to the total incidence of cancer      the total number of cancer deaths.
in industrialized nations is small. Furthermore,
although urban–rural differences in cancer in-
cidence have been reported in several countries,      Radiation
these differences tend to disappear when ho-
mogeneous populations with a similar lifestyle,            Ultraviolet
for example, the Mormons, are studied.103 And,        It has been known for a long time that exposure
although in England and Wales, certain occu-          to UV or ionizing irradiation can cause cancer in
pations have been associated with a different         humans. The association between skin cancer
risk of cancer from that of the general popula-       and exposure to sunlight was observed more than
tion, nearly 90% of such variation is eliminated      100 years ago, and in 1907 William Dubreuilh, a
if individual groups of similar social class and      French dermatologist, reported epidemiologic
habits are compared.104 Other inconsistencies         evidence implicating sunlight as a cause of skin
also occur. For example, bladder cancer is            cancer, supporting the earlier observation.105 In
linked to certain chemical and allied industries      1928, George Findlay, a British pathologist, ex-
in United States, but no clear industrial associ-     perimentally verified this by inducing skin can-
ation has been found in Japan. Prostate cancer is     cer in mice exposed to ultraviolet radiation.106
higher in blacks than in whites living in the same       Ultraviolet radiation is a low-energy emission
counties in the United States, and both black         and does not penetrate deeply. Hence the skin
and white males in the United States have a           absorbs most of the radiation and is the primary
higher incidence of prostate cancer than men in       carcinogenic target. Because nonmelanomatous
the industrialized United Kingdom and Japan.          skin cancer is the most easily detectable and cur-
However, the differences in incidence between         able human cancer, the fact that it is also the
the United States and the United Kingdom may          most clearly identifiable is often overlooked. The
be mainly due to PSA screening because PSA            fear of skin cancer, however, is apparently not
screening is not widely used in the United            sufficient to prevent people from overexposing
Kingdom. Interestingly, the mortality rates are       themselves to this carcinogenic agent.
similar between the two countries.32                     The evidence for the association of skin cancer
   Thus, the overall distribution patterns of can-    and UV radiation is compelling and can be sum-
cer observed in North America and Western             marized as follows:107
Europe, with high frequencies of lung, colon,
breast, and uterine cancer, suggest some com-            1. Skin cancer occurs primarily on exposed
mon factors in the environment of these regions             areas, that is, the head, neck, arms, hands,
in comparison with regions in Africa in which               and legs (in women).
90                                                                                  CANCER BIOLOGY

     2. Skin cancer is relatively rare in dark-       creased incidence of leukemia among radiolo-
        skinned races in whom skin pigment filters     gists was recognized.109
        out UV radiation, whereas it is common in         In more recent times, follow-up studies of
        fair-skinned people.                          atomic bomb survivors and people exposed to
     3. The incidence of skin cancer and the          the aftereffects of the Chernobyl disaster have
        amount of exposure to sunlight are related.   shown that exposure to radioactive fallout is
     4. Skin cancer frequency and the intensity of    linked to an increased susceptibility for thyroid
        solar radiation are related. Going toward     cancer, breast cancer, and leukemia. Both the
        the equator, the prevalence of skin cancer    age at exposure and the dose of radiation re-
        in Caucasians increases in proportion to      ceived are key factors in determining risk of
        the intensity of UV radiation.                cancer.
     5. Skin cancer can be induced in laboratory          The types of neoplasms produced in individ-
        animals by repeated exposure to UV ra-        uals exposed to ionizing radiation depend on a
        diation.                                      number of factors, including dose of radiation,
     6. The inability to repair DNA damaged by        age at time of exposure, and sex of the individual.
        UV radiation is associated with skin can-     Within 25 to 30 years after whole-body or trunk
        cer. Thus, individuals with xeroderma         irradiation, there is an increased incidence of
        pigmentosum, an inherited disease with a      leukemia and cancers of the breast, thyroid, lung,
        DNA-repair defect, almost always develop      stomach, salivary glands, other gastrointestinal
        skin cancer.                                  organs, and lymphoid tissues. Other malignant
                                                      neoplasms have been observed in tissues that
   Both malignant melanoma and nonmelanoma
                                                      were locally exposed to high doses of radiation. A
skin cancers are associated with exposure to UV
                                                      number of unfortunate but striking examples are
radiation, although the dose–response curve is
                                                      available. During the 1920s, in a factory in Or-
less steep for melanoma.107 The most common
                                                      ange, New Jersey, watch dials were painted with
types of skin cancer are basal cell carcinoma,
                                                      radium and mesothorium to make them lumi-
which may be locally invasive but is almost never
                                                      nescent. To get a fine tip on the brushes used to
metastatic; squamous cell carcinoma, which is
                                                      paint the dials, the workers wetted the brush tips
more aggressive than basal cell carcinoma, in-
                                                      on their tongues, leaving a deposit of radioac-
vades locally, and may rarely metastasize; and
                                                      tive material on the tongue. Approximately 800
melanoma, which is less common than the other
                                                      young women were exposed to the radioactive
forms of skin cancer but is often highly malignant
                                                      materials in this manner. Radium is a radioiso-
and rapidly metastatic, with an average 5-year
                                                      tope that becomes deposited in bone, and sev-
survival rate of over 90% if detected and treated
                                                      eral years later, a high incidence of osteogenic
early but only 14% if metastatic.
                                                      sarcoma became evident in these workers.110
                                                      Another radioisotope that becomes deposited in
                                                      internal organs is thorium, which was used in
      Ionizing Radiation
                                                      the preparation of a radiocontrast solution called
The carcinogenic effects of ionizing radiation        Thorotrast, once used for diagnostic purposes.
were discovered from studies of pioneer radia-        The overall incidence of malignant diseases in
tion workers who were occupationally exposed,         patients who received this material has been
individuals who were exposed to diagnostic or         found to be twice the expected incidence, with
therapeutic radiation, and atomic bomb survi-         liver tumors and leukemias being about sixfold
vors. Malignant epitheliomas of the skin were         higher than expected.111 Another example of
observed in the earliest experimenters with X-        local irradiation producing cancer a number of
rays and radium within a few years after their        years later is the observation of thyroid cancers
discovery in 1895 and 1898, respectively. By          in individuals who had been irradiated over the
1914, a total of 104 case reports of radiation        neck during childhood for either a so-called
cancers had been noted and analyzed.108 In            enlarged thymus gland or hypertrophied ton-
1944, the role of ionizing radiation in the in-       sils and adenoids. An 83-fold increased risk
THE EPIDEMIOLOGY OF HUMAN CANCER                                                                       91

for thyroid cancer has been noted in these                An increased incidence of lung cancer in
cases.112                                              deep-well miners was observed in uranium and
    The period between irradiation and the ap-         other ore miners in eastern Germany and western
pearance of cancer depends to some extent on the       Czechoslovakia over 60 years ago. Exposure to
age at irradiation.113 Juvenile tumors and leuke-      radon among these miners was very high, ap-
mias associated with prenatal irradiation become       proaching 3000 picocuries (pCi) per liter of air.
evident in the first 2 to 3 years after birth, with a   In the early 1950s, an increase in lung cancer
peak incidence at 5 years of age. The latency          incidence was noted among uranium miners in
period following postnatal irradiation, however, is    Colorado. Later, an increased rate of lung cancer
5 to 10 years for leukemia and more than 20 years      was also noted among miners working in iron,
for most solid tumors. The increased incidence of      zinc, tin, and fluorspar mines. In these mines,
leukemia and solid tumors also appears to be           radon levels were also high. Although these min-
higher after prenatal than after postnatal irradia-    ers were also exposed to other potentially carci-
tion. In general, the data suggest that the relative   nogenic dusts, the common feature was exposure
risk for cancers other than leukemia decreases         to radon. The excess number of lung cancer
with increasing age at the time of irradiation.        deaths in these miners (compared to nonminers)
    In the case of breast cancer, for example,         ranges from 0.3% to 13% and varies depending
women who received a radiation dose before age         on the ambient air concentration.115 This risk
40 have increased risk of developing breast can-       goes up in more than an additive manner for in-
cer.114 After age 40, radiation has a small effect     dividuals who are also smokers.
on breast cancer risk. Women who were below               Monitoring of homes began in a rather hap-
age 10 at the time of exposure have an increased       hazard fashion in the 1980s. Nevertheless, some
risk that does not become apparent until these         regions with high indoor levels were found, in-
women reach the age at which breast cancer             cluding the Reading Prong geological region
usually occurs.114 This increased risk persists for    extending from Pennsylvania to New York. Esti-
at least 35 years and may remain throughout life.      mates of risk for lung cancer from radon exposure
Current evidence indicates that a very small num-      in residences is based on extrapolations from
ber, probably less than 1%, of breast cancer           miner risk data. These estimates may or may not
cases result from diagnostic radiography.114           be realistic, as there is some evidence both ways.
                                                       In a case–control study of 400 women with lung
                                                       cancer, performed by the Department of Health
                                                       in New Jersey, an increased risk was found for
Radon is a radioactive gas that is ubiquitous in       exposure levels of 2 pCi/L of air, but the results
earth’s atmosphere. It is formed from the radioac-     were not statistically significant.116 A study done
tive decay of radium-236. Radium is found in           in China, in which median household radon
substantial but varying amounts in soil and rocks      levels ranged from 2.3 to 4 pCi/L of room air, no
and ends up in some building materials. Various        positive associations between radon levels and
parts of the country have varying amounts, as do       lung cancer was found,117 a finding suggesting
certain localities within a small geographic area.     that projections of lung cancer risk from sur-
There is extensive epidemiologic evidence that         veys of miners exposed to high-radon levels are
exposure to high levels of radon produces bron-        overestimates.
chogenic carcinoma (reviewed in Reference 115),           Estimates of increased risk due to radon resi-
most of which comes from studies of workers            dential exposure very widely. For an average life-
involved in deep mining of uranium and other           time exposure to 1 pCi/L, estimates vary from
ores. Because this epidemiologic evidence is           5000 to 20,000 excess lung cancer deaths per
quite compelling and because radon is so wide-         year in the United States.115 These estimates
spread, the potential that large numbers of peo-       uphold the conservative tradition of radiation pro-
ple might develop lung cancer from such expo-          tection. To put this in some perspective, a life-
sure has produced a radon scare in the United          time exposure to 4 pCi/L is estimated to cause a
States, not unlike the asbestos scare.                 1% increase in lung cancer, whereas the risk of
92                                                                                                              CANCER BIOLOGY

smoking cigarettes increases the risk of lung                          risk than that for age-matched controls. Expo-
cancer at least 10-fold over that of nonsmokers.                       sure to therapeutic radiation and anticancer
Most homes in the United States have indoor                            drugs such as alkylating agents nitrogen mus-
radon levels less than 2 pCi/L and these are                           tard, cyclophosphamide, procarbazine, and ni-
levels usually found in basements.118                                  trosoureas are known risk factors for second
                                                                       cancers. About 25% of those patients who de-
                                                                       velop second cancers are known to have some
                                                                       genetic susceptibility such as Li-Fraumeni syn-
Drugs that have been associated with human                             drome, retinoblastoma, neurofibromatosis, or a
cancers are listed in Table 3–8. Some of these                         sibling with cancer120 (see below). The most com-
have been used as therapeutic or diagnostic                            mon malignant familial condition predisposing
agents in medical practice. Among these, the                           to second neoplasm is retinoblastoma, in which
anticancer drugs in particular have been impli-                        osteosarcomas and soft tissue sarcoma are most
cated. A number of these (e.g., cyclophospha-                          common.
mide, melphalan, and busulfan) are alkylating                             The risk of second cancers in children sur-
agents, known to interact with DNA in a manner                         viving ALL, however, is relatively low. In a study
similar to that of known chemical mutagens and                         of 9720 children treated for ALL from 1972 to
carcinogens (see Chapter 2).                                           1988, 43 second cancers were seen.121 Second
   Second cancers arising later in life from the                       cancers may occur within 5 years from first di-
effects of treatment of childhood cancers are a                        agnosis. These are usually acute myelogenous
particular concern. It is estimated that from 3%                       leukemia (AML), CML, or non-Hodgkin’s lym-
to 12% of children treated for cancer will de-                         phoma. Secondary solid tumors, the most com-
velop a new cancer within 20 years from the                            mon of which are brain tumors, may be seen 5 to
time of first diagnosis.120 This is a 10-fold higher                    15 years later. Brain tumors are most often seen

     Table 3–8. Medicinal Drugs Classified as Carcinogenic to Humans
     Drug or Drug Combination
     IARC Group I                                                                          Cancer Site and Cancer Type
     Analgesic mixtures containing phenacetin                                              Kidney, bladder
     Azathioprine                                                                          Lymphoma, skin, liver and bile ducts,
                                                                                             soft connective tissue
     N,N-bis(2-chloroethyl)-2-naphthylamine (Chlornaphazine)                               Bladder
     1,4-Butanediol dimethane-sulfonate (Myleran; Busulfan)                                Leukemia
     Chlorambucil                                                                          Leukemia
     1-(2-Chloroethyl)-3-(4methyl-cyclohexyl)-1-nitrosourea (Methyl-CCNU)                  Leukemia
     Ciclosporin                                                                           Lymphoma, Kaposi’s sarcoma
     Cyclophosphamide                                                                      Leukemia, bladder
     Diethylstilbestrol                                                                    Cervix, vagina
     Etoposide in combination with cisplatin and bleomycin                                 Leukemia
     Fowler’s solution (inorganic arsenic)                                                 Skin
     Melphalan                                                                             Leukemia
     8-Methoxypsoralen (Methoxsalen) plus ultraviolet radiation                            Skin
     MOPP and other combined (anticancer)                                                  Leukemia
       chemotherapy including alkylating agents
     Estrogen therapy, postmenopausal                                                      Breast, uterus
     Estrogen, non-steroidal                                                               Cervix, vagina
     Estrogens, steroidal                                                                  Uterus, breast
     Oral contraceptives, combined*                                                        Liver
     Oral contraceptives, sequential                                                       Uterus
     Tamoxifen{                                                                            Uterus
     Thiotepa                                                                              Leukemia
     Treosulfan                                                                            Leukemia
     *There is also conclusive evidence that these agents have a protective effect against cancers of the ovary and endometrium.
     There is conclusive evidence that tamoxifen has a protective effect against second breast tumors in patients with breast cancer.
     (From World Cancer Report,119 with permission.)
THE EPIDEMIOLOGY OF HUMAN CANCER                                                                         93

in patients who received cranial irradiation.          and starts to decrease in the perimenopausal pe-
Most other secondary solid cancers in this pa-         riod. Prolactin levels have been reported to de-
tient population also appear to result from ir-        crease in women after full-term pregnancy125 and
radiation. Of those not treated with irradiation,      this may provide some protective effect. Some
most will have received alkylating agents and          studies indicate that the rate of cell proliferation
a number appear to have some genetic predis-           is greater in nulliparous women than in parous
position. A recent study has shown a 1.4-fold          women (reviewed in Reference 125), and this
increased risk of breast cancer following radio-       may reflect the lower hormonal levels in the latter
therapy and alkylating agent chemotherapy in           group. Moreover, mitotic activity of breast epi-
young women (age 30 years or younger).122              thelium varies during the menstrual cycle and
Somewhat surprisingly, patients treated with           peaks during the luteal phase, suggesting that
alkylating agents alone had a lower risk of breast     progesterone also has a role in regulating the
cancer compared to case–controls, presumably           mitotic rate in breast tissue. Presumably, these
because of ovarian damage that inhibits estro-         hormones could act as promoters for cells initi-
gen production.                                        ated by some carcinogens, and the amount of du-
    A higher incidence of lymphomas has been           ration of exposure to these hormones could then
seen in patients who have received organ trans-        increase the risk in a woman who has a propensity
plants for which they were treated with immu-          to develop breast cancer. That other risk factors
nosuppressive drugs, some of which are also used       are also involved is evident from comparative data
in cancer chemotherapy. The most widely used           from United States and Japanese women. In these
immunosuppressive drug, cyclosporine, which            two groups, data on the first birth, nulliparity, and
has revolutionized the organ transplant field, has      age at menopause show that the lower rate in
also been noted to cause lymphoproliferative           Japanese women is not accounted for by these
disease, including lymphomas. This was partic-         factors.125 The remaining difference may be re-
ularly a problem in the earlier clinical trials when   lated to dietary fat and total body weight, both of
higher doses were being employed in combina-           which are, on the average, higher in United States
tion with high doses of corticosteroids and anti-      women. The breast cancer risk associated with
thymocyte globulin.123 More recently, the inci-        body weight is thought to operate through the
dence of lymphoproliferative disease has been          increased levels of conversion of adrenal andro-
reduced to about 1% in transplant patients             gens to estrogen and lower levels of sex-hormone
treated with lower doses of cyclosporine.              binding globulin in obese women.126
    Other drugs are also suspected of causing             Estrogens have been used extensively in the
cancer in humans. For example, 10 cases of liver       treatment of postmenopausal symptoms and for
cell tumors have been reported in patients with        the prevention of osteoporosis. There is a clear
blood disorders treated for long periods with the      association between use of ‘‘unopposed’’ estrogen
androgenic steroid oxymetholone.124 Several            therapy (i.e., without progestins) and increased
studies indicate that chronic abuse of analgesics      risk of endometrial cancer. Combination of lower
containing phenacetin leads to papillary necro-        doses of estrogen and a progestin protects against
sis of the kidney. It has been suggested that this     the estrogen-alone effect on the endometrium.
is related to the subsequent development of tran-      The problem is that estrogen–progestin combined
sitional cell carcinoma of the renal pelvis in a       hormone replacement therapy (CHRT) enhances
number of these cases.124                              breast cancer risk.127,128 In a case–control study of
                                                       over 2500 postmenopausal women, CHRT was
                                                       associated with a 10% higher risk of breast can-
                                                       cer for each 5 years of use.127
As noted earlier, the risk factors associated with        This risk association has been confirmed by ‘‘a
breast cancer include age at menarche, age at the      Million Woman’’ study in the United King-
time of the first full-term pregnancy, and age at       dom that reported a 1.74-fold increase in breast
menopause. These factors suggest a role for estro-     cancer in current HRT users who had taken
gens and progesterone in breast cancer. Produc-        hormones for 1–4 years and a 2-fold increase
tion of these hormones increases near menarche         risk in 5- to 9-year current users.129 Curiously,
94                                                                                 CANCER BIOLOGY

no increased risk was observed in women who          here that an association between infection with
had used HRT in the past, even if use was for        Epstein-Barr virus and Burkitt’s lymphoma,
greater than 10 years. Overall, current users        hepatitis B and C viruses and liver cancer, hu-
were at a 1.66-fold increased risk of developing     man T-cell lymphotropic virus (HTLV) and leu-
breast cancer and 1.22-fold increased risk of        kemia, and human papilloma virus and cervical
dying from it. To put this in some context one       cancer are examples of this linkage. A number of
should consider the fact that the risk of devel-     neoplasms have also been associated with HIV
oping breast cancer for a 50-year-old woman          infection in patients with AIDS. These include
who has not had breast cancer is 11% in her          Kaposi’s sarcoma and non-Hodgkin’s lymphoma
lifetime.130 Thus, the increased risk for a post-    primarily, but central nervous system tumors
menopausal woman taking CHRT is 10% of               and Hodgkin’s disease are also seen in patients
11%, or 1.1%, for each additional 5 years of her     with AIDS.132
life if she continues to take CHRT. If she lives        Infection with certain parasites also seems to
to be 80, the increased risk is 6.6% (age 50 to      be able to initiate a cascade of events culmi-
80 ¼ six 5-year periods; 6Â1.1% ¼ 6.6%).             nating in malignant neoplastic disease in certain
   Many epidemiologic studies have shown that        populations. The high incidence of bladder can-
oral contraceptives do not significantly affect the   cer in patients whose urinary bladders are in-
risk of breast cancer. However, a small increase     fected with the shistosome parasite indigenous
has been reported in some studies in certain         to Egypt and other parts of Africa as well as the
groupsofwomen.Theseincludewomenwhohave               occurrence of a type of liver cancer (cholangio-
used contraceptives for several years before age     sarcoma) in patients with clonorchiasis, a para-
25 and/or before the first full-term pregnancy;       sitic infection of the liver, common in parts of
women who continue to use oral contraceptives        China, are examples of this association. Infection
at age 45 and older; women with a history of         with the bacterium Helicobacter pylori (H. pylo-
benign breast disorders; multiparous, premeno-       ri) is associated with gastric cancer, but other
pausal women with early menarche; and women          agents in the diet are likely to be cofactors.
with a family history of breast cancer (reviewed
in Reference 114).
   Several studies have reported that use of         AGING AND CANCER
diethylstilbestrol (DES) during pregnancy is
associated with an overall risk of about 1.5-fold    Cancer is a disease of aging. The average age at
for developing breast cancer.114 The well-           diagnosis is 67 and the median age of patients
documented appearance of vaginal adenocarci-         with cancer in the United States is 70 years.133
nomas in women whose mothers had been                The incidence of cancer rises exponentially with
treated with DES in early pregnancy with the         age from ages 40 to 80 (Fig. 3–10). This age-
intent of preventing abortion is another example     related increase in cancer probably relates to the
of hormonally induced neoplasm.131                   combined effects of accumulated genetic alter-
                                                     ations (mutations, translocations, etc.), increased
                                                     epigenetic gene silencing, telomere dysfunction,
                                                     and altered tissue stroma as tissues age. There is
Cancer is not an infectious disease in the usual     evidence for each of these factors playing a role
sense of the term. Doctors, nurses, and spouses      (reviewed in Reference 134).
who come into close contact with cancer pa-             Increased somatic mutations have been ob-
tients do not have a higher risk of developing       served in aged cells and tissues from humans
cancer than the rest of the population.51 How-       and mice. This process most likely occurs as a
ever, there is now known to be a clear associa-      result of accumulated DNA damage due to ex-
tion between infection with certain types of         ogenous and endogenous agents such as oxygen
viruses and neoplastic disease. Infection with       free radical–forming agents (see Chapter 2).
certain viruses probably acts in concert with        Such mutations can also result from error-prone
other carcinogenic agents or processes. This is      repair during DNA replication. Whether de-
discussed in detail in Chapter 2. Suffice it to say   creasing DNA repair capacity with aging is the
THE EPIDEMIOLOGY OF HUMAN CANCER                                                                                                  95

           Percent with invasive cancers




                                                        0–39                      40–59                    60–79
                                                                            Age range (years)

                                           Figure 3–10. Cancer incidence as a function of age. Incidence of invasive
                                           cancer plotted against age ranges reveals exponential increase from age 40
                                           to 80 years.1 Note that beyond age 80, incidence of cancers plateaus. (From
                                           De Pinho,134 reprinted by permission from Macmillan Publishers Ltd.)

culprit is not clear. However, in cells from aged                                 transcription (see Chapter 5). DNA methyla-
mice and humans, an increased level of chro-                                      tion and histone deacetylation are involved in si-
mosomal abnormalities has been observed. In                                       lencing genes, and if tumor suppressor genes
addition, an age-related decline in repair of                                     are targets for these reactions one can pre-
UV-induced DNA damage has been found in                                           dict the consequences. There is evidence for an
cultured primary skin fibroblasts and lympho-                                      age-related progressive increase in CpG island
blastoid cell lines, when comparing normal do-                                    DNA methylation, a finding leading to the con-
nors up to 10 years of age with normal donors in                                  cept that this methylation, which often involves
their 80s or 90s (reviewed in Reference 134).                                     promoter sequences in tumor suppressor genes,
There is also an age-associated decrease in cel-                                  is responsible for the age-related alterations
lular levels of proteins involved in DNA repair,                                  leading to cancer.
such as ERCC3 for excision repair, replication                                       Telomere dysfunction has also been impli-
protein A, and p53. However, other investiga-                                     cated in the increased incidence of cancer among
tors have not observed significant differences in                                  the elderly. Telomeres are shorter in some human
DNA repair machinery of human keratinocytes                                       cancer cells than in normal cells in the tissue of
in response to UV damage, so age-dependent                                        origin, which suggests that telomere shorten-
changes in DNA repair capability is apparently                                    ing occurs during part of the carcinogenic pro-
not a sine qua non of the aging process. Germ-                                    cess.134 Telomerase, the enzyme that maintains
line defects in mismatch repair genes such as                                     telomere length (see Chapter 5), is then often
MSH6, by contrast, are associated with late-                                      reactivated. This reactivation is part of what pro-
onset colon cancer (see Genetic Factors, below).                                  vides the ‘‘immortality’’ of cancer cells. With the
Thus a variety of mechanisms may be involved                                      accumulation of somatic mutations over time,
in the deterioration of genome maintenance in                                     loss of cell cycle checkpoint controls, and acti-
the elderly. Epigenetic mechanisms may also be                                    vated telomerase, transformed cells are well on
involved in the age-related increase in cancer                                    their way to progression to a dysregulated, ge-
incidence. These mechanisms involve DNA                                           netically unstable population of cells.
methylation and histone deacetylation reactions                                      It is now well known that the mesenchymal
that regulate chromatin structure and gene                                        stroma, on which epithelial cells grow, divide,
96                                                                                     CANCER BIOLOGY

and differentiate, plays a key role in maintaining        Damage due to oxidation appears to underlie
a normal tissue (see Chapter 4). It is also clear      much of the age-associated effects involved in
that the ‘‘malignant stroma’’ has a lot to do with     an increased risk of cancer. This phenomenon
fostering the carcinogenic process. For example,       may be due to formation of reactive oxygen spe-
only prostate cancer–associated fibroblasts were        ciesthatcauseDNA strand breaksorbasechanges
able to sustain the growth of malignant pros-          in DNA and also to protein oxidation. The oxy-
tatic cells in vivo, whereas fibroblasts from           gen free-radical hypothesis of aging postulates
nonmalignant prostate tissue were not.135 Some         that the progressive decline in functional activity
of the changes in cancer stroma are similar to         and chromosomal integrity with age results from
some age-related changes in senescent der-             the accumulation of oxidative damage by reac-
mal fibroblasts—i.e., their increased production        tive oxygen species produced by cells’ normal
of cytokines, proteases, and other extracellular       metabolic activity. There is support for this hy-
matrix–degrading enzymes.136                           pothesis in that steady-state levels of carbony-
   An interesting model for the aging process is       lated proteins, produced by protein oxidation,
the roundworm C. elegans. This organism lives          increase with age.139 These oxidized proteins
for only a few weeks, thus changes in gene ex-         are then targeted for intracellular proteolysis. It
pression and protein levels are telescoped over a      is also possible that the reverse is true, that tran-
short time, allowing for correlation with the          scriptional (e.g., frameshift or base change mu-
aging process. It has been found that the insu-        tations) or translational errors produce misfolded
lin and insulin-like growth factor (IGF-1) path-       proteins that are then oxidized, and it is the
ways regulate the aging process by up-regulating       oxidation reaction that targets proteins for deg-
longevity-favoring genes, such as antioxidant          radation. Indeed, the age-related cellular lesion
enzymes, stress-response proteins involved in          in the latter case would be the accumulation
protein folding, and antimicrobial genes, and          of misfolded proteins, which then become oxi-
down-regulating specific life-shortening genes in       dized.139 Thus, it may be aged cells’ inability to
C. elegans.137 These same pathways appear to           fold proteins correctly, due to transcriptional or
affect life span in fruit flies and mice as well, and   translational errors (as a result of chromosomal
probably in humans.                                    changes over time), rather than an age-related
   DNA microarray technology has been used to          decrease in oxidative defense systems that is the
measure mRNA levels in dividing fibroblasts             central defect. In any case, antioxidants have
isolated from young, middle-aged, and elderly          been reported to decrease the aging process and
normal humans and patients with progeria, an           extend life span in some organisms. For exam-
inherited genetic disorder characterized by pre-       ple, exposure of C. elegans to small molecule
mature aging.136 Genes whose lower expression          synthetic superoxide dismutase and catalase
correlated with aging include genes involved in        mimetics increased life span by 44%.140 Super-
cell cycle checkpoints, chromosomal segregation,       oxide dismutase and catalase are cellular en-
chromatin structure, and proteasome function.          zymes that protect many cell types, including
Ly et al.136 propose ‘‘that the underlying mech-       mammalian cells, from oxidative damage by
anism of the aging process involves increasing         neutralizing reactive oxygen species, and the abil-
errors in the mitotic machinery of dividing            ity to pharmacologically enhance these activities
cells . . . and that this dysfunction leads to chro-   would seem to be a good target for anti-aging
mosomal pathologies that result in misregula-          drugs.
tion of genes involved in the aging process.’’ Of
all the factors that have been associated with
increasing longevity in animals, caloric restric-      GENETIC FACTORS IN CANCER
tion is the best documented. One reason why
caloric restriction appears to be a key to longer      A number of inherited traits are related to cau-
survival is that it enhances gene expression           sation of cancer. A few cancers have a definite
linked to suppression of DNA damage caused by          inheritance, whereas others may arise in indi-
mitotic recombination.138                              viduals with a genetic defect that makes them
THE EPIDEMIOLOGY OF HUMAN CANCER                                                                       97

more susceptible to potentially carcinogenic         the retinoblastoma gene virtually ensures that
agents. There are really two different aspects to    the carrier will develop such a tumor, oncogenesis
the genetics and cancer issue. First, the initia-    at the cellular level must be a rare event because
tion and promotion–progression events that oc-       only three to four tumors, on average, develop in
cur in the body over time are due to changes in      a retinal cell population of several million (this
the structure and function of the genome in          assumes that the cancer arises from the progeny
adult cells (or in the case of pediatric cancers,    of one or a small number of precursor cells,
children’s cells). These are called somatic muta-    which appears to be the case for most types
tions and usually involve activation of oncogenes    of cancer; see Chapter 2). Thus, the genetically
and inactivation of tumor suppressor genes.          dominant inherited mutation is not in itself
These changes accumulate over time and may           sufficient to ensure that a retinal cell bearing the
be progression related, as described in the ‘‘Vo-    gene will become a cancer cell.
gelgram’’ for colorectal cancer, discussed in           Approximately 50 forms of hereditary cancers
Chapter 5. The second type of genetic basis          have been reported, some of which are listed in
for cancer is inherited defects. These are called    Table 3–9. For example, for years breast cancer
germline mutations and they increase cancer sus-     has been considered to have a familial associa-
pectibility, usually by some interaction with the    tion. Similar associations have been noted for
environment.                                         ovarian cancer. Genetic studies have also associ-
                                                     ated occurrence of breast cancer with a variety of
                                                     other tumors in the same families,143 including
Inherited Cancers
                                                     associations between breast cancer and ovarian
A list of inherited cancer syndromes is shown in     cancer, gastrointestinal tract cancer, soft tissue
Table 3–9. These neoplasms represent a small         sarcoma, brain tumors, or leukemia. In a study of
fraction, perhaps 1% to 2%, of total cancers.142 A   12 pedigrees that had a clustering of breast and
high percentage of certain tumors, however, are      ovarian cancer among female relatives, the data
genetically determined. For example, dominant        suggested that a genetic factor was transmitted
genetic inheritance accounts for about 40% of        from affected mothers to half of their daughters
retinoblastomas and 20% to 40% of Wilms’ tu-         and, in some families, father-to-daughter trans-
mors (embryonal renal tumors) and neuroblasto-       mission appeared to occur. These observations
mas. Familial multiple polyposis of the large        suggest the possibility of X-chromosome linkage
bowel is another example of a disease that is        in the transmission of breast and ovarian cancer
transmitted as a Mendelian-dominant trait, with      in these families. In a number of these family-
about an 80% penetrance rate. Cancer of the          associated cancers, specific chromosomal ab-
large bowel will eventually occur in nearly 100%     normalities or genetic mutations have been ob-
of untreated patients with familial multiple poly-   served. These will be discussed in detail in
posis. There is also a predisposition to develop a   Chapter 5. The following are few examples.
variety of other neoplasms, particularly subcu-         Table 3–10 lists some of the high-risk cancer
taneous tumors and osteomas, in these latter         susceptibility genes, their chromosomal location,
patients.                                            and the associated cancers. Some of these genes
   The probability that an individual carrying the   are involved in genome integrity (brca1 and
retinoblastoma gene will develop a tumor is          brca2), some are cell cycle regulator genes ( p16
about 95% and an average of three to four tu-        and CDK4), and some are DNA mismatch repair
mors occur in such a gene carrier. A child born      genes (e.g., hMLH1, hMSH2, and hMSH6). These
without the gene has only 1 chance in 30,000 of      genetic mutations usually demonstrate incom-
developing retinoblastoma. This amounts to an        plete pentrance. For example, a woman carrying
about 100,000-fold increased risk in the gene        a mutated brca1 gene has about a 70%-80%
carrier group (assuming an average of three tu-      lifetime risk for developing breast cancer. It is
mors per gene carrier with an incidence of 95        clear that other cofactors are involved in this risk.
per 100, compared to 1 tumor per 30,000 in the       The so-called high-risk susceptibility genes may
general population). Although the presence of        also be involved in sporadic cancers for which no
98                                                                                                                CANCER BIOLOGY

Table 3–9. Inherited Cancer Syndromes Caused by a Single Genetic Defect*
Syndrome                                   Gene            Location               Cancer Site and Cancer Type
Familial retinoblastoma                    RB1             13q14                  Retinoblastoma, osteosarcoma
Multiple endocrine neoplasia II            RET             10q11                  Medullary thyroid carcinoma, pheochromocytoma
Multiple endocrine neoplasia I             MEN1            11q13                  Adrenal, pancreatic islet cells
Neurofibromatosis type I                    NF1             17q11                  Neurofibromas, optic gliomas, pheochromocytoma
Neurofibromatosis type II                   NF2             22q2                   Bilateral acoustic neuromas, meningiomas,
                                                                                    cerebral astrocytomas
Bloom syndrome                             BLM             15q26                  Leukemia, lymphoma
Familial adenomatous polyposis             APC             5q21                   Colorectal, thyroid
Von Hippel-Lindau                          VHL             3p25                   Renal cell carcinoma, pheochromocytoma
Familial Wilm’s tumor                      WT1             11q                    Wilms tumor (kidney)
Xeroderma pigmentosum                      XP(A–D)         9q,3p,19q,15p          Basal cell carcinoma, squamous cell
                                                                                    carcinoma, melanoma (skin)
Fanconi anemia                             FAC             16q, 9q, 3p            Acute leukemia
Li-Fraumeni syndrome                       p53             17p13                  Breast and andrenocortical carcinomas,
                                                                                    bone and soft-tissue sarcomas,
                                                                                    brain tumors, leukemia
Cowden syndrome                            PTEN            10q22                  Breast, thyroid
Gorlin syndrome                            PTCH            9q31                   Basal cell carcinoma
X-linked proliferative disorder            XLP             Xq25                   Lymphoma
Peutz-Jeghers syndrome                     LKB1            19p                    Breast, colon
Ataxi telangiectasia                       ATM             11q22                  Leukemia, lymphoma
*The lifetime risk of cancer is high. There are usually recognizable phenotypic features that make the syndromes easy to identify clinically.
(From World Cancer Report,141 with permission)

clear gene association has been found. The in-                          seen in common types of cancer and, as noted, in
herited cancer syndrome genes usually have a                            sporadic cancers. However, the distinction be-
very high degree of penetrance and are relatively                       tween the rarer inherited cancer syndromes and
rare (e.g., the incidence of p53 gene mutations                         those mutations found in the more common
involved in the Li-Fraumeni syndrome is 1 in                            cancers is somewhat arbitrary. For example, rb1,
10,000 individuals).141 The inherited susceptibil-                      apc, p53, and PTEN mutations are involved in
ity gene mutations are more common and are                              both inherited and sporadic cancers.

Table 3–10. High-risk Susceptibility Genes and Their                    Gene–Environment Interactions
Chromosomal Location*
                                                                        Another way in which inherited susceptibility to
Gene              Location        Associated Tumors                     cancer may be expressed is the way in which an
BRCA1             17q             Breast, ovary, colon, prostate        individual can handle carcinogenic insults from
BRCA2             13q             Breast, ovary, pancreas,              the environment. For example, some individuals
p16 INK4A         9p              Melanoma, pancreas
                                                                        have a reduced capacity to metabolize carcino-
CDK4              6q              Melanoma, other tumors                gens such as arylamines because of a slow
                                    (rarely)                            acetylator phenotype, related to polymorphisms
hMLH1             3p              Colorectal, endometrial,
                                    ovarian cancer
                                                                        in the N-acetyltransferase-2 gene. Others may
hMSH2             2p              Colorectal, endometrial,              have a decreased ability to detoxify a number of
                                    ovarian cancer                      carcinogenic agents due to polymorphisms in
hMSH6             2p              Colorectal, endometrial,
                                    ovarian cancer
                                                                        the glutathione S-transferase gene GSTM-1 or
PMS1              2q              Colorectal cancer,                    cytochrome P-450 genes CYP2A6 or CYP2D6.
                                    other tumors (rarely)               Genes that regulate metabolism of drugs and
PMS2              7p              Colorectal cancer,
                                    other tumors (rarely)
                                                                        other xenobiotics are often discussed under the
HPC2              17p             Prostate (rarely)                     heading of pharmacogenetics. A number of ge-
*Inherited mutations in these genes are associated with some com-
                                                                        netic polymorphisms that are related to human
mon cancers.                                                            pharmacogenetic disorders are listed in Table
(From World Cancer Report,141 with permission)                          3–11. Evidence from a study of monozyotic
THE EPIDEMIOLOGY OF HUMAN CANCER                                                                            99

Table 3–11. Classification of Some Human Pharmacogenetic Disorders
Succinyicholine apnea                                    Trisomy 21
Acetylation polymorphism                                 Dysautonomia
  Isoniazid-induced neurotoxicity                        Leprechaunism
  Drug-induced lupus erythematosus                     Defective absorption
  Phenytoin-isoniazid interaction                        Juvenile pernicious anemia
  Isoniazid-induced hepatitis                            Folate absorption-conversion
  Arylamine-induced bladder cancer                     Increased metabolism
Increased susceptibility to drug-induced hemolysis       Succinylcholine resistance
  Glucose-6-phosphate dehydrogenase deficiency            Atypical liver alcohol dehydrogenase
  Other defects in glutathione formation or use          Atypical aldehyde dehydrogenase
Hereditary methemoglobinemia                           CHANGE IN DRUG RESPONSE DUE
Hypoxanthine-guanine phosphoribosymtransferase-        TO ENZYME INDUCTION
  (HPRT)-deficiency                                     The porphyrias
P450 mono-oxygenase polymorphisms                      The Ah locus
  Debrisoquine 4-hydroxylase deficiency
  Vitamin D-dependent rickets type I                   ABNORMAL DRUG DISTRIBUTION
  C21-Hydroxylase polymorphism                           Thyroxine (hyperthyroidism or hypothyroidism)
Enzymes of methyl conjugation                            Iron (hemochromatosis)
Hyperbilirubinemia                                       Copper (Wilson’s disease)
  Crigler-Naijar syndrome type II
  Gilbert’s disease                                    DISORDERS OF UNKNOWN ETIOLOGY
Fish-odor syndrome                                     Corticosteroid-induced glaucoma
                                                       Malignant hyperthermia associated with general
INCREASED RESISTANCE TO DRUGS                            anesthesia
Inability to taste phenyithiourea                      Halothane-induced hepatitis
Coumarin resistance                                    Chloramphenicol-induced aplastic anemia
Possibility of (or proven) defective receptor          Phenytoin-induced gingival hyperplasia
  Steroid hormone resistance                           Thromboembolic complications caused by anovulatory
  Cystic fibrosis                                         agents
From Nebert and Weber144

(MZ) and dizygotec (DZ) twins has shown that           and 3% to geophysical factors (mostly exposure
susceptibility to carcinogens and mutagens is          to sunlight). Thus, about 84% of all cancers
highly heritalde. Mutagen sensitivity was mea-         should be avoidable, if these estimates are cor-
sured by exposing peripheral blood lymphocytes         rect. The ‘‘ideal’’ man, then, should not smoke
to mutagens in vitro and the correlation coeffi-        or drink; should eat a diet low in fat, rich in fiber
cients were all significantly higher in MZ than in      and yellow vegetables; should protect himself
DZ twins.144a                                          from hazardous chemicals in the workplace and
                                                       home; should minimize intake of drugs and
                                                       avoid unneeded X-rays; and should protect him-
AVOIDABILITY OF CANCER                                 self from sunlight. The ‘‘ideal’’ woman (from the
                                                       point of view of avoiding cancer) should do all
If, as a number of cancer epidemiologists con-         this and, in addition, have at least one child early
tend, lifestyle accounts for about 80% of all malig-   in her reproductive life and avoid multiple sex
nant cancers, then presumably the same pro-            partners. Assuming that we cannot totally avoid
portion of cancers should be avoidable. To be          the pollution in our environment, which epide-
more specific, about 30% of all cancers are             miologists tell us accounts for no more than 2%
thought to be related to smoking, 3% to alcohol        of cancers, exposure to certain infectious agents
consumption, 30% to diet, 7% to sexual and             (5% of cancers), and certain other unknown fac-
reproductive patterns, and another 5% to occu-         tors including genetic determinants (about 4%),
pational hazards and industrial products.50,51         we should be able to decrease our cancer mor-
Moreover, about 1% are estimated to be related to      tality rate by about 84% by simple, direct actions
drugs and medical procedures (primarily X-rays)        as individuals. This conclusion is almost certainly
100                                                                                                       CANCER BIOLOGY

overly sanguine but appears to be worth the                       to mutagens occurs from chemicals in our diet,
experiment.                                                       water, and air, from products that we use as cos-
   Evidence to support the idea that most can-                    metics and drugs, and from cigarette smoking.
cers are avoidable comes from different sorts of                  As noted before, several mutagenic substances
data: (1) differences in the incidence of cancer                  have been identified in cigarette smoke. Muta-
in various areas of the world (Table 3–12); (2)                   gens are also found among the natural products
differences in the incidence rates for various                    contained in foods such as products elaborated
types of cancer between residents of a country                    by molds (e.g., aflatoxin) or by edible plants that
and those of the same ethnic group who have                       synthesize a variety of toxins, presumably to
emigrated to another country (Fig. 3–9); (3)                      ward off insects, as well as among synthetic
variations over time in the incidence of cancers                  chemicals such as pesticides, industrial pollut-
within a given society or community; and (4) the                  ants, and weed killers.146 In short, we live in a
identification of specific causes of cancer and                     sea of mutagens and carcinogens. Identification
preventive measures resulting from this (e.g.,                    of potentially mutagenic substances in our en-
aniline dyes in bladder cancer, vinyl chloride in                 vironment is a major public health and political
hepatic angiosarcomas, asbestos in mesothelio-                    issue. Of the myriad of potential mutagens and
mas, etc.).                                                       carcinogens in our diet, only a few have been
                                                                  studied in detail. In addition, more than 65,000
                                                                  synthetic chemicals are produced in the United
Risk Assessment
                                                                  States, and about 1000 new chemicals are in-
As was discussed more fully in Chapter 2, most                    troduced each year.147 Only a small number of
chemical carcinogens are also mutagens. Be-                       these were examined for mutagenic and carci-
cause mutagenicity is much easier to measure                      nogenic potential before being marketed. Ob-
experimentally than carcinogenicity, many of the                  viously, this is a major epidemiologic problem
tests used to assess the carcinogenic risk potential              and one that has a major economic impact on
of substances in our environment are based on                     private industry as well as on the consumer and
mutagenicity assays. Exposure of human beings                     taxpayer. What is needed are accurate, rapid,

         Table 3–12. Worldwide Variation in Incidence of Common Cancers, with Range of
         Variation Expressed for Ages 35–64
                                                                                                              Range of
         Type of Cancer               High-Incidence Area                  Low-Incidence Area
         Skin                         Australia (Queensland)               India (Bombay)                     >200
         Buccal cavity                India                                Denmark                             >25
         Nasopharynx                  Singapore*                           England                              40
         Bronchus                     England                              Nigeria                              35
         Esophagus                    Iran                                 Nigeria                             300
         Stomach                      Japan                                Uganda                               25
         Liver                        Mozambique                           Norway                               70
         Colon                        U.S.{ (Connecticut)                  Nigeria                              10
         Rectum                       Denmark                              Nigeria                              20
         Pancreas                     New Zealand{                         Uganda                                5
         Breast                       U.S.{ (Connecticut)                  Uganda                                5
         Uterine cervix               Colombia                             Israel                               15
         Uterine corpus               U.S.{ (Connecticut)                  Japan                                10
         Ovary                        Denmark                              Japan                                 8
         Bladder                      U.S.{ (Connecticut)                  Japan                                 4
         Prostate                     U.S.**                               Japan                                30
         Penis                        Uganda                               Israel                              300
          The U.S. data are taken from the Connecticut Tumor Registry because it is the oldest continued cancer registry
         based on a defined population in this country.
         **African Americans.
         (From Doll145)
THE EPIDEMIOLOGY OF HUMAN CANCER                                                                    101

and economically feasible tests to predict the        ase (HGPRT) locus is used as a marker; the end
mutagenic and carcinogenic potential of the           point of the assay is loss of sensitivity to pu-
numerous chemicals in our environment. In             rine antimetabolites that must be activated by
practice, however, it has not been possible to        HGPRT to be effective, thus leading to the se-
develop a ‘‘perfect’’ short-term test. A number       lection of HGPRTÀ clones. Cultured fibroblast
of false positives and false negatives result from    cell lines such as Chinese hamster V79 or ovary
using these tests.                                    (CHO) cells are frequently used in this way.
   More than 100 short-term tests for mutage-            Agents that damage DNA can often be de-
nicity and carcinogenicity have been developed.       tected by examining an index of genotoxicity,
Some of the most widely used are bacterial muta-      such as unscheduled DNA synthesis, sister chro-
genesis (the Ames test), mutagenesis in cell cul-     matid exchange, or chromosome breakage in cul-
ture systems, direct measurement of damage            tured cells exposed to the agents in question.
to DNA or chromosomes in exposed cells, and              Carcinogenic potential has also been esti-
malignant transformation of cell cultures.148         mated by the ability of chemicals to ‘‘transform’’
One of the most popular of the short-term tests       smooth, well-organized monolayers of normal
is the Ames test, developed by Bruce Ames and         diploid fibroblasts into cells that grow piled up
colleagues.147 The basis of this assay is the         on one another (transformed foci) or into a cell
ability of a chemical agent to induce a genetic       type that can grow suspended in soft agar (nor-
reversion of a series of Salmonella typhimurium       mal fibroblasts do not usually grow on soft agar).
tester strains, which contain either a base sub-      Sometimes the putative malignant cells are then
stitution or a frameshift mutation, from histidine    injected into immunosuppressed or immunode-
requiring (hisÀ) to histidine nonrequiring (hisþ).    ficient (‘‘nude’’) mice to further demonstrate
These strains have been specially developed           that they are malignant. All of these estimates of
for this assay by selecting clones that have a de-    carcinogenic potential are fraught with danger
creased cell surface barrier to uptake of che-        in that a significant number of false negatives or
micals and a decreased excision repair system.        false positives can occur.
Other advantages of this system are the small            No single short-term test is foolproof; how-
genome of the bacteria (4 Â 106 base pairs), the      ever, if definitive evidence of genotoxicity has
large number of cells that can be exposed per         been obtained in more than one test, a chemical
culture dish (about 109), and the positive selec-     is highly suspect. An agent found to be mutagenic,
tion of the mutated organisms (i.e., only the         DNA damaging, and a chromosome breaker is
mutated organisms will grow under the test con-       almost certain to also be carcinogenic.148 Final
ditions). This system has great sensitivity: only     proof of mutagenicity and carcinogenicity in-
about 1 in 1000 to 1 in 10,000 of the mutated         volves the chronic exposure of whole animals to
bacteria need to be detected to give a positive       the test chemical. Although the short-term in
test, and nanogram amounts of a potent mutagen        vitro tests have several advantages, a number
can be detected as a positive. Both base substi-      of important components, such as absorption,
tution and frameshift mutagens can be detected,       pharmacokinetics, tissue distribution, metabo-
and, using the appropriate tester strains, the type   lism, age or sex effects, and species specificity,
of mutagen can be deduced because frameshift          cannot be duplicated in vitro. Tests in whole
mutagens usually revert only frameshift muta-         animals take a long time and, unfortunately, are
tions of the tester strains and not base substi-      very expensive.
tution mutations, and vice versa. Because many           One key question remains: how does one es-
mutagens must be metabolized to be active, a liver    timate the danger of low-dose exposures? More
homogenate fraction containing microsomes is          importantly, how does one estimate the risk
usually added to the incubation to provide the        of low-dose exposure over a lifetime? These are
drug metabolizing enzymes.                            extremely difficult questions to answer, but in
   The potential of various chemical agents to        practical terms, as long as an individual’s DNA-
mutagenize mammalian cells has also been used         repair enzymes are working (see Chapter 2), there
as a short-term test. Frequently, mutation at the     probably is some low level of exposure below
hypoxanthine-guanine phosphoribosyltransfer-          which DNA lesions can be removed efficiently
102                                                                                    CANCER BIOLOGY

without permanent damage. The low level of              never smoked, 14 of 1000 will die of heart dis-
mutation in human genes seems to argue in               ease, 5 of lung cancer, and 7 of breast cancer by
support of this conclusion.149 If one could, in fact,   the time they reach 70 years of age. For 60-year-
measure the amount of DNA-adduct formation              old men who are smokers, 84 of 1000 will die
(i.e., the amount of DNA bases bound to car-            of heart disease and 98 of lung cancer, but only
cinogen) after exposure to various doses of car-        4 of 1000 will die of prostate cancer.
cinogen, one could probably get a much better              A few years ago there was debate about whe-
estimate of the risks involved in exposure to           ther the incidence of childhood cancers is going
various amounts of carcinogenic agents.150 A shift      up, with the Environmental Protection Agency
in the dose–response curve with low doses of            and the National Cancer Institute taking opposite
carcinogen could occur for several reasons, all         points of view.152 Data obtained between 1975
of which make linear extrapolations of dose–            and 1995 showed a slight increase in cancers
response data from animal studies tenuous. Some         of children, which appeared to be due to an in-
of these reasons relate to differences in metabo-       crease in brain cancer. However, the rates of leu-
lism, distribution, and overall pharmacokinetics        kemia and lymphoma, which together account
among species.                                          for about 35% of all childhood cancers, did not
                                                        change. Since there are only about 1800 new
                                                        cases of brain cancer in the United States per
THE GREAT CANCER MYTHS                                  year, a small number of patients being diagnosed
                                                        in any given year could skew the numbers. This is
Cancer is a dreadful disease. In most polls, it is      not to belittle the devastating effects of childhood
the most feared disease of all. Coupled with this       cancer, but it must be kept in mind that cancer
are the almost daily media reports of another           in children is a rare disease, about one-third of
carcinogen or cancer risk being found in our en-        which is due to leukemia, and for which the
vironment that produce a setting for the some-          overall 5-year survival rate for all childhood
times hysterical fear that cancer lurks around          cancers combined is 70% to 94%.
every corner. Epidemiological pronouncements               While there is a tendency to blame environ-
that one out of eight women will die of breast          mental causes for cancer in children, this prob-
cancer or one of every four men will get pros-          ably plays a small role. Hereditary gene muta-
tate cancer, while perhaps having some statisti-        tions probably play a larger role. Most experts
cal validity if everyone would reach age 80 and         agree that a mother’s smoking during pregnancy,
die of nothing else, belies the real risk of getting    electromagnetic fields from power lines, or other
and dying of cancer. A study published by Wo-           environmental toxicants play little role.152
loshin et al.151 puts this rate in a more rational         There are always debates about what is or is
context.                                                not a human carcinogen. Many of them have
   These authors have developed charts for men          been identified by occupation, a rare medical
and women that show the chance of dying from            exposure, atomic bomb fallout, or viral or other
various causes based on age and smoking history         infections. Determination of whether a chemi-
(Figs. 3–11 and 3–12). Instead of giving risks in       cal is a human carcinogen by high-dose expo-
terms of population percentages, these data show        sure in rodents, frequently at doses that no
risk in terms of individual risks. For example,         human being would ever be exposed to, is no-
their data indicate that a 60-year-old woman,           toriously inaccurate and has led to many false-
even one who smokes, has a 4.5% chance of               positive claims. Another point of view is ex-
dying of a heart attack in the next decade, a 6.5%      pressed by Bruce Ames, who has said that he is
chance of dying of lung cancer, and a 0.7%              a ‘‘contrarian in the hysteria over tiny traces of
chance of dying of breast cancer. Or to look at it      chemicals that may or may not cause cancer.
another way, for every 1000 60-year-old women           If you have thousands of hypothetical risks
who are smokers, 45 will die of heart attacks,          that you are supposed to pay attention to, that
65 of lung cancer, and 7 of breast cancer in the        completely drives out the major risks you should
next 10 years. For 60-year-old women who have           be aware of,’’153 which I would add include
THE EPIDEMIOLOGY OF HUMAN CANCER                                                                    103

               Figure 3–11. Risk chart for women who have never smoked. The chart in-
               dicates the number of women per 1000 who will die from various diseases
               and for any reason during the next 10 years, beginning at the indicated age.
               (*The numbers of each row do not add up to the chance of dying from any
               reason because there are many other causes of death in addition to the ones
               listed here.) Shaded area indicates age group and disease combinations with
               fewer than 1 death per 1000. (From Woloshin et al.,151 with permission.)

cigarette smoking, obesity, lack of antioxidants      adage in pharmacology: ‘‘a tiny amount of some-
in the diet, sun overexposure, and inadequate         thing doesn’t necessarily cause anything and
access to health care in large parts of the United    enough of something can cause anything.’’
States and the world.                                    Described below are some commonly held
   It has often been said that we live in a sea of    myths about agents that cause cancer.
carcinogens. Indeed, every time you’re stuck in
traffic or behind an exhaust-belching truck, you
                                                      Passive Smoking
are inhaling a lung-full of potential carcinogens.
If you live in a city and and drink chlorinated       Inhalation of smoke in an enclosed space, es-
water, you are exposing yourself to a host of         pecially for prolonged periods of time, is not
potential carcinogens. Most of the modern con-        healthy. It may trigger an asthmatic attack in a
veniences that we take for granted contain            child, for example. There are studies that show if
carcinogenic substances. The chair you sit in         nonsmokers are in a 10 Â 10 square foot room
probably has polyurethane, another carcinogen,        with smokers, within a short time of exposure,
in the cushions. Another point is that modern         effects on the cardiovascular system, e.g., heart
technology such as high-sensitivity mass spec-        rate and blood pressure, can be observed. This
trometers can detect parts per billion of che-        is not surprising. Nicotine, after all, is a drug
mical substances. Thus, one must ask, is the          that can cause cardiovascular and central ner-
detection of any level of a carcinogen danger-        vous system effects. The data for passive smok-
ous? Perhaps it is worth keeping in mind an old       ing causing lung cancer, however, are skimpy at
104                                                                                 CANCER BIOLOGY

              Figure 3–12. Risk chart for men who never have smoked. The chart indicates
              the number of men per 1000 who will die from various diseases and for any
              reason during the next 10 years, beginning at the indicated age. (*The
              numbers in each row do not add up to the chance of dying from any reason
              because there are many other causes in death in addition to the ones listed
              here.) Shaded area indicates age group and disease combinations with fewer
              than 1 death per 1000. (From Woloshin et al..151 with permission.)

best. In a multicenter case–control study of ex-     coronary heart disease orlung cancer, even before
posure to environmental tobacco smoke (ETS)          or after taking into consideration seven potential
and lung cancer in Europe, no association be-        confounding factors and before or after exclud-
tween childhood exposure to ETS and lung can-        ing participants with pre-existing disease. This
cer risk was found.154 There was weak evi-           was true for follow-up periods 1960–65, 1966–
dence of a dose–response relationship between        72, 1973–85, and 1973–98. The authors con-
risk of lung cancer in spouses and workplace         cluded that ‘‘the association between exposure
ETS, but no detectable risk after cessation of       to environmental tobacco smoke and coronary
exposure. Similarly, no association between ex-      heart disease and lung cancer may be consid-
posure to ETS and female breast cancer mor-          erably weaker than generally believed.’’
tality was found in two large cohort studies, one
involving over 146,000 women155 and one in-
                                                     Radon in the Home
volving over 116,000 women.156
   One of the most definitive studies involved        As noted above in the sections on the role of
118,094 adults in California enrolled in late        various factors in cancer development, radon is a
1959 and followed until 1998.157 Of these, 35,561    well-established occupational carcinogen. There
were never smokers who had a spouse in the           are significant data indicating an increased risk
study with known smoking habits. No significant       of lung cancer in deep-well miners, particularly
associations were found for current or former        among miners who smoke, and there is a dose–
exposure to environmental tobacco smoke and          response relationship to this risk. Since radon in
THE EPIDEMIOLOGY OF HUMAN CANCER                                                                    105

the soil can seep into homes in areas where           the electromagnetic spectrum. RF radiation at
there is a high natural soil context of radon gas,    sufficiently high levels can produce heat by in-
there has been concern that this exposure             ducing small electric currents. A typical cell
presents a potential risk of lung cancer. These       phone operates with a power output that could
risk assessments used the exposure dose–              only cause, at a maximum, a rise 0.18 C.165 This
response relationship from studies of uranium         amount could not be expected to have any sig-
miners and miners of other ores and extended          nificant biological effect. In addition, RF does
this relationship by a linear nonthreshold model      not possess sufficient energy to remove elec-
down to zero exposure.158 Based on assumptions        trons from atoms or molecules and thus does not
from this model and extrapolations from occu-         produce ionizing radiation,165 which is the kind
pational data, it has been stated that radon is the   that could damage DNA.
second leading cause of lung cancer after ciga-          Because of the weak thermal and ionizing
rette smoking,159 although these authors admit        potential of RF from cell phones, it seems highly
that ‘‘the effect of smoking on lung cancer risk      unlikely that this RF would cause cancer. A
appears to be an order of magnitude greater than      smattering of reports, however, from rodent stud-
the effect of radon.’’ The difference between the     ies have suggested some associated risk between
linear nonthreshold extrapolation model and the       RF in the potential range of a cell phone’s MHz
threshold model for cancer risk estimation was        and tumors, a range that could damage DNA
discussed in Chapter 2.                               (reviewed in Reference 165). A number of hu-
   A significant amount of data does not support       man epidemiological and occupational exposure
this claim of the association of radon exposure in    studies do not support any association of cell
the home and lung cancer. A large case–control        phone risk and cancer (reviewed in Reference
study of 1055 case subjects and 1544 controls in      165). A study of 250,000 cell phone users in the
Finland did not indicate an increased risk of lung    United States did not show any increased cancer
cancer from indoor radon exposure.160 A meta-         risk, and a case–control study from Sweden in-
analysis of lung cancer risk from residential radon   dicated no increase in brain tumors. In a study of
in eight epidemiological studies reported that in     195,775 workers engaged in manufacturing and
four of the eight studies the data for an associa-    testing of cell phones, no association between RF
tion were positive or weakly positive and the re-     exposure and brain, other nervous system can-
maining four showed no increased risk.161 Other       cers, or leukemia was found. A nationwide cohort
major studies also showed results that were           study involving 420,000 cell phone users in Den-
equivocal.161 In a small case–control study (138      mark found no association between cell phone
cases, 291 controls) in an Italian alpine valley      use and tumors of the brain or salivary glands,
with high radon levels, an increased risk was ob-     leukemia, or other cancers.
served, but the association was confined to male
smokers.162 A similar study done in Sweden in-
                                                      Electromagnetic Fields
dicated some risk due to radon exposure, but
again it was higher among smokers.163 A re-           There have been some studies suggesting a link
cent study done in the United Kingdom indi-           between magnetic fields generated by electrical
cated that radon is not a risk factor for childhood   power lines and childhood leukemia. For exam-
cancers.164                                           ple, an excess incidence of leukemia in Swed-
                                                      ish children was associated with the estimated
                                                      electric current flow, based on historical records
Cell Phones
                                                      of local power companies;166 however, that risk
The use of cell phones has increased rapidly in       of childhood leukemia did not correlate with
the past few years. They are found in most parts      residential measurements of magnetic field made
of the world, even in remote areas of developing      shortly after the time of diagnosis (reviewed in
countries. Cell phones emit radiofrequency            Reference 167). In fact, there have been a num-
(RF) signals in a range between 800 and 2000          ber of shortcomings in earlier epidemiologi-
MHz, which puts it in the microwave range of          cal studies on this topic, such as discrepancies
106                                                                                   CANCER BIOLOGY

between results based on proxy measurements            another cohort study,171 women who consumed
and those based on direct magnetic field mea-           15 or more (!) grams of ethanol per day had a
surements, the absence of supportive laboratory        26% increased risk. To put this in context, if the
data, and lack of a substantive biological expla-      risk of a woman over age 50 in contracting breast
nation for causing cancer.167                          cancer is 11%,130 an increased risk of 25% is
   A very careful study done by National Cancer        0.11Â0.25 ¼ 2.7%, or 2 to 3 of every 100 women.
Institute investigators and their collaborators        If the risk is 9%,169 then the increased risk is 1%,
directly measured magnetic fields in cases’ and         or 1 of 100 women. Given the notorious under-
controls’ bedrooms, three or four other rooms,         reporting of personal alcohol use, the odds
and the front doors of their houses.167 In addition,   are that at least in some of these studies the
they measured magnetic fields in homes where            amount of alcohol actually consumed was under-
case subjects’ and controls’ families lived during     estimated.
their mothers’ pregnancies. The results of this           A hypothesis for an association of alcohol use
study showed that the risk of acute lymphoblastic      and breast cancer is that alcohol increases cir-
leukemia (the most common malignancy of child-         culating levels of estrogen. Here again the data
hood) was not increased among children who             are conflicting. In one study of premenopausal
lived in homes with the highest exposure to mag-       women, who consumed 30 grams per day of
netic fields, and there was also no significant asso-    ethanol, increased blood levels of estrogen were
ciated risk with magnetic-field levels of the homes     reported, whereas in another study, alcohol in-
where the mothers resided when pregnant.               take was not associated with plasma estrogen
                                                       levels, but was associated with increased levels of
                                                       androstenedione (reviewed in Reference 170).
Chronic alcohol abuse is associated with in-
                                                       Organochlorine Compounds,
creased risk of certain cancers such as liver and
                                                       Polycyclic Aromatic Hydrocarbons,
oral pharyngeal cancers, in which there is usu-
                                                       and Breast Cancer
ally some associated tissue toxicity as a prodro-
mal factor. There have been a number of studies        Environmental exposure to organochlorine
of the potential risk of alcohol consumption and       compounds such as polychlorinated biphenyls
breast cancer. One such study suggested that a         (PCBs), 2,20 -bis (p-chlorophenyl)-1, 1, 1-tri-
woman who had as few as three alcoholic drinks         chloroethane (DDT) and its metabolite DDE,
a week had an increased risk of breast cancer.         and organochloro pesticides has been suggested
This is one of those conclusions that doesn’t pass     as a risk factor for breast cancer. The basis for
the ‘‘common sense’’ test, which should be used        this claim is that some of these are carcino-
for all reports, particularly those in the media,      genic in animals, have estrogenic activity, and
about what causes cancer or what cures it.             are inducers of cytochrome P-450 enzymes that
   In a meta-analysis of 28 case–control and 10        metabolize drugs, hormones, and various xeno-
cohort studies, comparing drinkers and non-            biotics. Some epidemiological studies have sug-
drinkers, the risk of breast cancer increased          gested an association between this class of com-
24% with consumption of two drinks per day.168         pounds and breast cancer risk, but these studies
Another study, pooling data from six prospective       have been contradictory and inconclusive (re-
studies, reported a 9% increase in breast cancer       viewed in Reference 172).
incidence with each 10 grams of alcohol con-              Because of a purported clustering of breast
sumed per day.169 These data were mostly ob-           cancer on Long Island, New York, and a fair
tained from postmenopausal women. In another           amount of political pressure, a $30 million Long
case–control study, recent alcohol consumption         Island breast cancer study project was launched
of 13 grams per day (about equivalent to 3 drinks      to examine the relationship between exposure to
of 100 proof whiskey) was associated with a 21%        environmental agents and breast cancer inci-
increased risk of breast cancer, but in the age        dence. The study was carried out under the
group less than 30 years of age there was no in-       auspices of the National Cancer Institute and
creased risk (reviewed in Reference 170). In           the National Institute of Environmental Health
THE EPIDEMIOLOGY OF HUMAN CANCER                                                                       107

Sciences. (In fact, the incidence of breast cancer     was not a dose-dependent relationship between
on Long Island is not significantly different from,     exposure and adduct formation, suggesting that
and in many instances less than, other locales in      there is a threshold effect. This latter point is
New York State).173 The published results of this      interesting because there are known polymor-
study indicated that there was no association          phisms in the enzymes that activate and detoxify
between increased rates of breast cancer and           PAHs, thus a pharmacogenetic analysis could
exposure to PCBs, DDT, or pesticides (reviewed         reveal who may be at higher risk.
in Reference 174). Studies done in other parts of         Finally, it is worth noting that a careful anal-
the United States and of the world also show no        ysis of all risk factors for breast cancer must be
significant correlation between serum or plasma         done before one can conclude that a ‘‘cluster’’ of
levels of PCBs or other organochlorine com-            breast cancer cases is related to some local en-
pounds and breast cancer.172,175–177 The only ca-      vironmental factor. For example, a study done
veat to this is a report from a small cohort study     in the San Francisco Bay area, involving both
that women living within one mile of hazardous         Caucasian and African-American women, found
waste sites containing organochlorines had a           that the elevated breast cancer incidence in the
higher incidence of breast cancer (reviewed in         Bay area could be completely accounted for by
reference 177a). However, when the data were           regional differences in known risk factors, e.g.,
pooled in a combined analysis, there was no as-        parity, age at first pregnancy, months of breast
sociation between breast cancer risk and blood         feeding, and ages at menarche and menopause.180
levels of PCBs or DDE.177a
   An argument has been made that since PCBs
and DDTs have been banned since the 1970s,
these may not have been the correct chemicals          Recently, a rumor that underarm antiperspirants
to look at or exposure of female babies in utero       or deodorants caused breast cancer was widely
may be the key factor here. While there may be         circulated on the Internet and picked up in the
some truth to these assumptions since organo-          media. This is another example of widely dis-
chlorines do persist in the environment and can        seminated urban myth. In a case control study
remain in the body for more than a decade.             of 813 women with breast cancer diagnosed be-
However, a case–control study based on cohorts         tween 1992 and 1995, compared with 793 wo-
of women who donated blood in 1974, 1989,              men without breast cancer, there was no link
or both and who were matched on age, race,             between breast cancer and regular use of anti-
menopausal status, and month and year of blood         perspirants or deodorants, even when applied
donation showed that even after 20 years of            after underarm shaving.181 Both the American
follow-up after exposure to relatively high con-       Cancer Society and the National Cancer Insti-
centrations of DDE or PCBs there was no asso-          tute posted notices on their Web sites to assure
ciation with an increased risk of breast cancer.178    the public that there is no scientific basis for this
   One might argue that a better way to assess         rumor.
risk is to look at damage to the target in the body
to which environmental agents might bind. This
                                                       Water Chlorination
was done in a study that looked at polycyclic
aromatic hydrocarbon–DNA adduct levels in              Some of the compounds used to disinfect water
blood mononuclear cells of women who live on           in urban drinking water systems are carcinogenic
Long Island. Samples from 576 breast cancer            at high doses in rodents. A study by Komulainen
cases and 427 age-matched controls were as-            et al.182 reported that administration of 3-chloro-
sayed for PAH–DNA adducts by ELISA.179 The             4-(dichloromethyl)-5-hydroxy-2 (5H)-furanone
levelsofPAH–DNAadductswereslightly,though              (MX) to rats in their drinking water produced
not significantly, higher among cases than con-         thyroid tumors and bile-duct neoplasms (cho-
trols. Also, there was no consistent association be-   langiomas). Based on linear extrapolation of
tween adduct levels and passive cigarette smoke        the data on dose-exposure for induction of cho-
exposure or consumption of grilled or smoked           langiomas, the upper-bound cancer risk per unit
foods. The authors also concluded that there           dose for lifetime exposure to MX was estimated
108                                                                                 CANCER BIOLOGY

to be 100% per milligram MX per kilogram body         high levels can cause lung cancer and meso-
weight per day. The cancer-causing dose esti-         thelioma of the pleura and peritoneium. Meso-
mates for humans are based on assumptions that        theliomas in asbestos workers can be induced by
rats and humans absorb, metabolize, and excrete       exposure to asbestos alone, whereas lung can-
MX in the same manner, that rats and human tis-       cers are more likely to be caused by exposure to
sueshavethesamecarcinogenictargetorresponse,          asbestos in smokers. However, since mesothe-
and that the extrapolation model for estimating       liomas are quite rare, lung cancer cases are much
low-exposure dose-response is an accurate reflec-      more common among asbestos workers.
tion of human risk. None of these assumptions            Since the 1980s, regulations on occupational
is likely to be true. Furthermore, MX levels in       exposure have greatly reduced exposure to as-
United States and Finland water supplies have         bestos. Based on the known carcinogenic effect
been reported to range from 3 to 67 parts per         of occupational exposure to asbestos and its
trillion (ppt). Based on the highest estimate of      wide use as fire retardant material in homes and
67 ppt, daily human exposures would be several        public buildings, a ground swell of public and
orders of magnitude lower than those employed         political pressure has been mounted to remove
by Komulainen et al.183 Using the extrapolation       every scrap of asbestos from schools, homes,
model, this could potentially produce two cancers     and public buildings, creating a whole new in-
per one million people.183 Similar conclusions can    dustry. The validity of such overwrought con-
be reached for other water disinfectants such as      cerns has been questioned for several rea-
trihalomethanes. Thus, the overall cancer risk due    sons.185 First of all, estimates of risk were based
to disinfectants in drinking water is very small      on extrapolations from occupational exposure to
and the risk–benefit ratio is very large. Disastrous   environmental exposure that are 1/100,000 of
consequences could result if water chlorination       those to which asbestos workers were exposed in
were stopped. For example, a cholera epidemic         the past. Second, dose–response estimates used
involving 300,000 people occurred in Peru as a        to extrapolate the data varied by a factor of 1000
consequence of inadequate disinfection of drink-      among various studies. Third, estimates of risk
ing water supplies.183                                of asbestos-induced cancer have not been vali-
                                                      dated in nonoccupational exposed populations.
                                                      Finally, in a study of nonoccupationally exposed
Abortion or Miscarriage
                                                      women in two chrysotile-asbestos mining re-
and Breast Cancer
                                                      gions, no increased risk of lung cancer was
Some reports have suggested that incomplete           observed.185 The authors concluded that the
pregnancies, terminated either by induced abor-       Environmental Protection Agency’s extrapola-
tion or miscarriage, increases the risk of breast     tion model overestimates the risk of asbestos-
cancer (reviewed in Reference 184). A number of       induced lung cancer by at least a factor of 10-
other studies have not shown an increased risk of     fold.
breast cancer in women who have undergone
induced abortions.184 A well-controlled study
of the effects of induced abortion and miscar-
riage on breast cancer incidence, involving age-,     As discussed in Chapter 2, there have been a
parity-, and race-matched cases and controls,         number of reports that sodium saccharin, either
showed that neither induced abortion nor mis-         in the diet or by direct installation, causes uri-
carriage increased breast cancer risk.184 This        nary bladder cancer in rats (reviewed in Refer-
claim appears to be more of an issue of politics      ence 186). On the basis of these observations,
and religious beliefs than science.                   the United States Food and Drug Administra-
                                                      tion proposed to ban saccharin from human use.
                                                      This proposal was overturned by a moratorium
                                                      passed by the United States Congress, but sac-
It has been known for several decades, since the      charin use was banned in Canada. It turns out
follow-up of Liberty Ship builders in World War       that the bladder carcinogen effect is seen in
II, that occupational exposure to asbestos at         certain species of male rats (but not in mice,
THE EPIDEMIOLOGY OF HUMAN CANCER                                                                      109

hamsters, or guinea pigs) in which calcium             reaction between amino acids (such as aspara-
phosphate–sodium saccharin–containing pre-             gine, which is present in potatoes and cereals)
cipitates form in the urine after high-dose ad-        and reducing sugars. However, the amount of
ministration of the artificial sweetener (higher        acrylamide produced can vary depending on the
than humans would likely ever ingest). Dietary         food, temperature and duration of heating, wa-
administration of high doses of other sodium           ter, and starch content. Of course, as Bruce Ames
salts such as sodium ascorbate, sodium citrate,        points out, if we eliminated all the foods that
or even sodium chloride caused the same effect.        contain potential human carcinogens (including
Even more conclusive is a study in monkeys             peanut butter and a number of fruits and veg-
using doses of sodium saccharin 5 to 10 times          etables), there wouldn’t be much left to eat. This
the allowable daily intake for humans that showed      was probably summed up best by Walter Will-
no carcinogenic effect on the primate urinary          ett, Professor of Epidemiology and Nutrition
bladder.186                                            at the Harvard School of Public Health, who is
                                                       quoted as saying, ‘‘We don’t even know if it
                                                       [acrylamide] is a carcinogen in humans based on
Acrylamide in Foods
                                                       the amounts we eat. Out of the 100 things you
A report from a Swedish group in April 2002            should worry about, I’d put this at 200.’’187
announced that there were high levels of acryl-
amide in certain cooked or fried foods such as
French fries and potato chips. Since acrylamide
is listed by the World Health Organization             Alar was the trade name for a formulation con-
(WHO) as a probable human carcinogen, this             taining daminoxide, a hormone-like substance
raised a state of alarm. A WHO official stated:         that can slow the growth of certain varieties of
‘‘Given that we know acrylamides are cancer-           apples that tend to rot as soon as they ripen. Alar
causing in animals and probably in humans, it is       was approved by the FDA in 1968 and was used
intolerable that they are in foods at the levels       by apple growers to improve the efficiency of
found, and we have to find a remedy.’’187 A U.S.        the harvest. Its safety testing included a 2-year
Food and Drug Administration (FDA) repre-              animal carcinogenicity study in rats that were
sentative concurred, stating that ‘‘it is clear that   negative (reviewed in Reference 188). During the
acrylamide is a problem. It doesn’t need to be in      1970s, however, Alar and its byproduct showed
food.’’ Some food safety advocacy groups started       increases in certain tumors in mice fed enormous
planning lawsuits against fast-food companies.         doses of either substance (larger than the maxi-
   There is no consensus on the risk posed by          mum tolerated dose [MTD]). Additional studies
this ‘‘new threat’’ to human health. Indeed, skep-     based on feeding mice doses exceeding the MTD
tics point out that acrylamide is also found at        showed similar results. This caused a media up-
some level in breads, meat, and vegetables that        roar. In one well-publicized instance, a famous
have been eaten by people for thousands of             actress interviewed on TV stated that Alar was
years. Even given the huge increase in fast-food       going to give children cancer. A 60 Minutes T V
consumption in the last two to three decades,          segment in 1989 described Alar as ‘‘the most
there have not been significant increases in            potent cancer-causing agent in the food supply.’’
cancers that could be associated with acrylam-            In fact, there has never been any evidence that
ide.187 In addition, there are no data proving that    Alar causes childhood or any other type of can-
acrylamide is a human carcinogen. The amount           cer.188 Nevertheless, sales of apples decreased,
given to rats to produce tumors would require          parents poured apple juice down the drain, and
that a person consume 35,000 potato chips or           some schools removed all apple products from
182 pounds of French fries per day.                    their lunch menus. Even though the FDA, En-
   Further studies are under way to determine          vironmental Protection Agency, and U.S. Depart-
how acrylamide is generated in foods during            ment of Agricultur tried to assure the public,
cooking. One hypothesis is that in foods heated        the Alar scare continued, and the manufacturer
above 1208 C, acrylamide can be generated by a         halted sale of all Alar products for use on food
110                                                                                     CANCER BIOLOGY

crops. Since 1989, a number of agencies, includ-       about this issue, given the reports of SV40 DNA
ing a British government advisory group, a United      in human tumors, monitoring of people who
Nations panel, and the American Medical Asso-          received potentially SV40-contaminated polio-
ciation, have reviewed the scientific data and          virus vaccinations should probably continue.
concluded that Alar did not pose a risk to human
SV40 Virus in Early Polio Vaccines                       1. B. F. Hankey, L. A. Gloeckler-Ries, B. A. Miller,
                                                            and C. L. Kosary: Overview. In Cancer Statis-
Polio vaccines have been blamed for everything              tics, 1973–1989, NIH, Publ. No. 92–2789, I.1–
from initiating the AIDS epidemic to being a                17, 1992.
Western plot to subvert the developing world.            2. H. S. Seidman, M. H. Mushinski, S. K. Gelb,
                                                            and E. Silverberg: Probabilities of eventually de-
Poliovirus vaccines that were used during the               veloping or dying of cancer—United States,
late 1950s and early 1960s were contaminated                1985. CA Cancer J Clin 35:36, 1985.
with simian virus 40 (SV40), a monkey virus that         3. A. Jemal, T. Murray, E. Ward, A. Samuels, R. C.
came from the monkey cells in which early                   Tiwari, A. Ghafoor, E. J. Feuer, and M. T.
batches of the vaccine were grown. A survey                 Thun: Cancer statistics—2005. CA Cancer J
                                                            Clin 55:10, 2005.
done in 1961 indicated that about 90% of U.S.           3a. A. Jemal, R. Siegel, E. Ward, T. Murray, J. Xu,
citizens younger than 20 years of age (those born           et al. Cancer statistics-2006. CA-A Cancer
between 1941 and 1961) had received at least                Journal for Clinicians 56:106, 2006.
one immunization with poliovirus vaccine that            4. Cancer Facts and Figures—2003. Atlanta, GA:
may have contained SV40 virus.189 It is difficult            American Cancer Society, 2003.
                                                         5. D. M. Parkin, F. Bray, J. Ferlay, and P. Pisani:
to determine the average exposure because the               Global cancer statistics, 2002. CA Cancer J Clin
titers of live SV40 in different vaccine lots varied        55:74, 2005.
from undetectable to high.                               6. Lung cancer. In B. W. Stewart and P. Kleihues,
   Most epidemiological studies of populations              eds.: World Cancer Report, World Health Orga-
who were immunized with polio vaccine poten-                nization. Lyon: IARC Press, 2003, pp. 182–187.
                                                        6a. L. A. Loeb, V. L. Ernster, K. E. Warner, J.
tially containing live SV40 during early childhood,         Abbotts, and J. Laszlo: Smoking and lung can-
which is presumed to be the highest at-risk age             cer: An overview. Cancer Res 44:5940, 1984.
group for a lifetime risk of developing an exoge-        7. Breast cancer. In B. W. Stewart and P. Kleihues,
nous agent-induced cancer, have failed to show              eds.: World Cancer Report, World Health Or-
any association of polio vaccination with increased         ganization. Lyon: IARC Press, 2003, pp.
risk of cancer more than 30 years following              8. E. Cho, D. Spiegelman, D. J. Hunter, W. Y.
vaccination (reviewed in Reference 189). A                  Chen, M. J. Stampfer, G. A. Colditz, and W. C.
population-based study in Denmark, comparing                Willett: Premenopausal fat intake and risk of
cancer incidence data from 1943 to 1997 with ex-            breast cancer. J Natl Cancer Inst 95:1079, 2003.
posure to SV40-contaminated poliovirus vaccine,          9. Endogenous hormones and breast cancer col-
                                                            laborative group: Body mass index, serum sex
found that there was no associated increased in-            hormones, and breast cancer risk in postmeno-
cidence of mesothelioma, ependymoma, choroid                pausal women. J Natl Cancer Inst 95:1218, 2003.
plexus tumor, non-Hodgkin’s lymphoma, or leu-           10. H. L. Judd, I. M. Shamonki, A. M. Frumar, and
kemia.190 However, SV40 DNA sequences have                  L. D. Lagasse: Origin of serum estradiol in post-
been reported to be present in certain human                menopausal women. Obstet Gynecol 59:680,
cancers such as childhood brain tumors, osteo-          11. V. L. Ernster, R. Ballard- Barbask, W. E. Bar-
sarcomas, and pleural mesotheliomas.                        low, Y. Zheng, D. L. Weaver, et al.: Detection
   Pleural mesotheliomas are the cancers most               of ductal carcinoma in situ in women undergoing
often reported to contain SV40 DNA. Yet age-                screening mammography. J Natl Cancer Inst
specific trends in the U.S. pleural mesothelioma             94:1546, 2002.
                                                        12. M. J. van de Vijver, Y. D. He, L. J. van’t Veer,
incidence rates are not consistent with an effect           H. Dar, et al.: A gene-expression signature as a
caused by exposure to SV40-contaminated po-                 predictor of survival in breast cancer. N Engl J
liovirus vaccine.189 To be totally comfortable              Med 347:1999, 2002.
THE EPIDEMIOLOGY OF HUMAN CANCER                                                                            111

13. K. N. Nathanson, R. Wooster, and B. L. Weber:        28. W. Isaacs, A. De Marzo, and W. G. Nelson:
    Breast cancer genetics: What we know and                 Focus on prostate cancer. Cancer Cell 2:113,
    what we need. Nat Med 7:552, 2001.                       2002.
14. H. S. Feigelson, R. McKean-Cowdin, G. A.             29. D. C. Malins, P. M. Johnson, T. M. Wheeler,
    Coetzer, D. O. Stram, L. K. Kolonel, and B. E.           E. A. Barker, N. L. Polissar, and M. A. Vinson: Age-
    Henderson: Building a multigenic model of breast         related radical-induced DNA damage is linked to
    cancer susceptibility: CYP17 and HSD17B1 are             prostate cancer. Cancer Res 61:6025, 2001.
    two important candidates. Cancer Res 61:785,         30. L. Cheng, S-Y. Song, T. G. Pretlow, F. W.
    2001.                                                    Abdul-Karin, H-J. Kung, et al.: Evidence of in-
15. J. Baselga and L. Norton: Focus on breast                dependent origin of multiple tumors from
    cancer. Cancer Cell 1:319, 2002.                         patients with prostate cancer. J Natl Cancer
16. R. B. Dickson and M. E. Lippman. Cancer of               Inst 90:233, 1998.
    the breast: Molecular biology of breast cancer.      31. C. A. Macintosh, M. Stower, N. Reid, and N. J.
    In V. T. J. DeVita, S. Heklman, and S. A.                Maitland: Precise microdissection of human
    Rosenberg, eds.: Principles and Practice of On-          prostate cancers reveals genotypic heterogene-
    cology. Philadelphia: Lippincott Williams and            ity. Cancer Res 58:23, 1998.
    Wilkins, 2001, pp. 1633–1645.                        32. A. Shibata and A. S. Whittemore: Correspon-
17. Colon cancer. In B. W. Stewart and P. Kleihues,          dence RE: Prostate cancer incidence and
    eds.: World Cancer Report, World Health                  mortality in the United States and the United
    Organization. Lyon: IARC Press, 2003, pp.                Kingdom. J Natl Cancer Inst 93:1109, 2001.
    198–202.                                             33. X. Bosch and O. Bernadich: Increased serum
18. J. D. Potter: Colorectal cancer: Molecules and           prostate-specific antigen in a man and a
    populations. J Natl Cancer Inst 91:916, 1999.            woman with hepatitis A. N Engl J Med 337:1849,
19. E. R. Fearon and B. A. Vogelstein: A genetic             1997.
    model for colorectal tumorigenesis. Cell 61:759,     34. J. L. Stanford, Z. Feng, A. S. Hamilton, F. D.
    1990.                                                    Gilliland, R. A. Stephenson, et al.: Urinary and
20. G. Smith, F. A. Carey, J. Beattie, M. J. V.              sexual function after radical prostatectomy for
    Wilkie, T. J. Lightfoot, et al.: Mutations in APC,       clinically localized prostate cancer—The pros-
    Kirsten-ras, and p53-alternative genetic path-           tate cancer outcomes study. JAMA 283:354,
    ways to colorectal cancer. Proc Natl Acad Sci            2000.
    USA 99:9433–9438, 2002.                              35. Bladder cancer. In B. W. Stewart and P. Kleihues,
21. N. Kemeny, Y. Huang, A. M. Cohen, W. Shi,                eds.: World Cancer Report, World Health
    J. A. Conti, et al.: Hepatic arterial infusion of        Organization. Lyon: IARC Press, 2003,
    chemotherapy after resection of hepatic metas-           pp. 228–231.
    tases from colorectal cancer. N Engl J Med,          36. Lymphoma. In B. W. Stewart and P. Kleihues,
    341:2039, 1999.                                          eds.: World Cancer Report, World Health Or-
22. S. D. Markowitz, D. M. Dawson, J. Willis, and            ganization. Lyon: IARC Press, 2003, pp. 237–
    J. K. V. Willson: Focus on colon cancer. Cancer          241.
    Cell 1:233, 2002.                                    37. L. M. Staudt and W. H. Wilson: Focus on
23. Liver cancer. In B. W. Stewart and P. Kleihues,          lymphomas. Cancer Cell 2:363, 2002.
    eds.: World Cancer Report, World Health              38. A. A. Alizadeh, M. B. Eisen, R. W. Davis, C.
    Organization. Lyon: IARC Press, 2003, pp.                Ma, I. S. Lossos, et al.: Distinct types of diffuse
    203–206.                                                 large B-cell lymphoma identified by gene ex-
24. Cancers of the female reproductive tract. In             pression profiling. Nature 403:503, 2000.
    B. W. Stewart and P. Kleihues, eds.: World           39. V. T. De Vita, Jr., A. A. Serpick, and P. P.
    Cancer Report, World Health Organization.                Carbone: Combination chemotherapy in treat-
    Lyon: IARC Press, 2003, pp. 215–222.                     ment of advanced Hodgkin’s disease. Ann
25. C. Crump, M. W. McIntosh, N. Urban, G.                   Intern Med 73:881, 1970.
    Anderson, and B. Y. Karlan: Ovarian cancer           40. Leukemia. In B. W. Stewart and P. Kleihues,
    tumor marker behavior in asymptomatic healthy            eds.: World Cancer Report, World Health
    women: Implications for screening. Cancer                Organization. Lyon: IARC Press, 2003, pp.
    Epidemiol Biomarkers Prev 9:1107, 2000.                  242–247.
26. M. Hoglund, D. Gisselsson, G. B. Hansen, T.          41. A. K. Stewart and A. C. Schuh: White cells 2:
    Sall, and F. Mitelman: Ovarian carcinoma de-             Impact of understanding the molecular basis of
    velops through multiple modes of chromosomal             haematological malignant disorders on clinical
    evolution. Cancer Res 63:3378, 2003.                     practice. Lancet 355:1447, 2000.
27. Cancers of the male reproductive tract. In           42. T. R. Golub, D. K. Slonim, P. Tamazo, et al.:
    B. W. Stewart and P. Kleihues, eds.: World               Molecular classification of cancer: Class discov-
    Cancer Report, World Health Organization.                ery and class prediction by gene expression
    Lyon: IARC Press, 2003, pp. 208–214.                     monitoring. Science 286:531, 1999.
112                                                                                      CANCER BIOLOGY

43. B. J. Drucker, S. Tamura, E. Breckdunger, et al.:         peripheral blood lymphocytes. Mutat Res 92:
    Effects of a selective inhibitor of the Abl tyro-         321, 1982.
    sine kinase on the growth of Bcr-Abl positive       60.   F. Veglia, G. Matullo, and P. Vineis: Bulky DNA
    cells. Nat Med 2:561, 1996.                               adducts and risk of cancer: A meta-analysis.
44. A. N. Houghton and D. Polsky: Focus on mel-               Cancer Epidemiol Biomarkers Prev 12:157,
    anoma. Cancer Cell 2:275, 2002.                           2003.
45. Melanoma.In B. W. Stewart and P. Kleihues,          61.   Q. Wei, L. Cheng, C. I. Amos, L. E. Wang,
    eds.: World Cancer Report, World Health                   Z. Guo, et al.: Repair of tobacco carcinogen–
    Organization. Lyon: IARC Press, 2003, pp.                 induced DNA adducts and lung cancer risk: a
    253–256.                                                  molecular epidemiologic study. J Natl Cancer
46. L. M. De Angelis: Brain tumors. N Engl J Med              Inst 92:1764, 2000.
    344:114, 2001.                                      62.               ¨
                                                              J. Stjernsward: Battle against tobacco. J Natl
47. Tumours of the nervous system. In B. W.                   Cancer Inst 81:1524, 1989.
    Stewart and P. Kleihues, eds.: World Cancer         63.   P. A. Wingo, L. A. G. Ries, G. A. Giovino, D. S.
    Report, World Health Organization. Lyon: IARC             Miller, H. M. Rosenberg, et al.: Annual re-
    Press, 2003, pp. 265–269.                                 port to the nation on the status of cancer,
48. A. Behin, K. Hoang-Xuan, A. F. Carpentier, and            1973–1996, with a special section on lung
    J-Y. Delattre: Primary brain tumours in adults.           cancer and tobacco smoking. J Natl Cancer
    Lancet 361:323, 2003.                                     Inst 91:675, 1999.
49. J. Peto: Cancer epidemiology in the last century    64.   A. J. Tuyns, G. Pequinot, and O. M. Jensen: Les
    and the next decade. Nature 411:390, 2001.                cancers de l’oesophage en Ille-et-Villaine en
50. The known causes of cancer. In K. Meister,                function des niveaux de consummation d’alcool
    E. M. Whelan, G. L. Ross, and A. N. Stimola,              et de tabac. Des risques qui se multiplient. Bull
    eds.: America’s War on ‘‘Carcinogens’’. New               Cancer (Paris) 64:45, 1977.
    York: American Council on Science and               65.   B. Herity, M. Moriarity, G. J. Bourke, and L.
    Health, 2005, pp. 59–73.                                  Daly: A case–control study of head and neck
51. R. Doll and R. Peto: The causes of cancer. J              cancer in the Republic of Ireland. Br J Cancer
    Natl Cancer Res 66:1191, 1981.                            43:177, 1981.
52. J. Bradbury: Add colorectal cancer to list of       66.   K. Matsuo, N. Hamajima, M. Shinoda, S.
    smoking-associated cancers, say experts. Lancet           Hatooka, M. Inoue, et al.: Gene–environment
    356:2072, 2000.                                           interaction between an aldehyde dehydroge-
53. The Health Consequences of Smoking, HEW                   nase-2 (ALDH-2) polymorphism and alcohol
    Publ. No. (CDC) 74–8704. Washington, DC:                  consumption for the risk of esophageal cancer.
    U.S. Department of Health, Education, and                 Carcinogenesis 22:913, 2001.
    Welfare, 1974, pp. 35–37.                           67.   J. L. Kelsey and M. D. Gammon: The epide-
54. D. R. Shopland, H. J. Eyre, and T. F. Pechacek:           miology of breast cancer. CA Cancer J Clin
    Smoking-attributable cancer mortality in 1991:            41:146, 1991.
    Is lung cancer now the leading cause of death       68.   T. J. Key, N. E. Allen, E. A. Spencer, and R. C.
    among smokers in the United States? J Natl                Travis: The effect of diet on risk of cancer.
    Cancer Inst 83:1142, 1991.                                Lancet 360:861, 2002.
55. L. Garfinkel and E. Silverberg: Lung cancer          69.   W. C. Willett: Diet and cancer: One view at the
    and smoking trends in the United States over              start of the millennium. Cancer Epidemiol
    the past 25 years. CA Cancer J Clin 41:137,               Biomarkers Prev 10:3, 2001.
    1991.                                               70.   E. Giovannucci and W. C. Willett: Dietary
56. United States Public Health Service: The                  factors and risk of colon cancer. Ann Med 26:
    Health Consequences of Smoking—Cancer: A                  443, 1994.
    Report of the Surgeon General. Rockville, MD:       71.   P. Terry, E. Giovannucci, K. B. Michels, L.
    U.S. Department of Health and Human Ser-                  Bergkvist, H. Hansen, et al.: Fruit, vegetables,
    vices, Office on Smoking and Health, 1982.                 dietary fiber, and risk of colorectal cancer. J
57. R. M. Putzrath, D. Langley, and E. Eisenstadt:            Natl Cancer Inst 93:525, 2001.
    Analysis of mutagenic activity in cigarette smok-   72.   J. H. Cohen, A. R. Kristal, and J. L. Stanford:
    ers’ urine by high performance liquid chemo-              Fruit and vegetable intakes and prostate cancer
    therapy. Mutat Res 85:97, 1981.                           risks. J Natl Cancer Inst 92:61, 2000.
58. G. Obe, H. J. Voght, S. Madle, A. Fahning, and      73.   S. D. Mark, Y-L. Qiao, S. M. Dawsey, Y-P.
    W. D. Heller: Double-blind study on the effect            Wu, H. Katki, et al.: Prospective study of
    of cigarette smoking on the chromosomes of                serum selenium levels and incident esophageal
    human peripheral blood lymphocytes in vivo.               and gastric cancers. J Natl Cancer Inst 92:1753,
    Mutat Res 92:309, 1982.                                   2000.
59. Vijayalaxmi and H. J. Evans: In vivo and in vitro   74.   L. C. Clark, B. Dalkin, A. Krongrad, et al.:
    effects of cigarette smoke on chromosomal dam-            Decreased incidence of prostate cancer with
    age and sister-chromatid exchange in human                selenium supplementation: Results of a double-
THE EPIDEMIOLOGY OF HUMAN CANCER                                                                               113

      blind cancer prevention trial. Br J Urol 81:730,             men: Cross-Canada study of pesticides and
      1998.                                                        health. Cancer Epidemiol Biomarkers Prev 10:
75.   O. P. Heinonen, D. Albanes, J. Virtamo, et al.:              11155, 2001.
      Prostate cancer and supplementation with a-            89.   A. Blair: Herbicides and non-Hodgkin’s lym-
      tocopherol and b-carotene: Incidence and mor-                phoma: New evidence from a study of Sas-
      tality in a controlled trial. J Natl Cancer Inst             katchewan farmers. J Natl Cancer Inst 82:544,
      90:440, 1998.                                                1990.
76.                                        ´
      A. Lacassagne: Apparition d’adenocarcinomes            90.   Environmental pollution. In B. W. Stewart and
      mammaires chez des souris males, traitees par´               P. Kleihues, eds.: World Cancer Report, World
                                ´          ´
      une substance oestrogene synthetique. Compt                  Health Organization. Lyon: IARC Press, 2003,
      Rend Soc Biol 129:641, 1938.                                 pp. 39–42.
77.   J. E. Bruni and D. G. Monetmurro: Effect of            91.   W. Haenszel, D. B. Loveland, and M. G. Sirken:
      pregnancy, lactation, and pituitary isografts on             Lung-cancer mortality as related to residence
      the genesis of spontaneous mammary gland tu-                 and smoking histories. I. White males. J Natl
      mors in the mouse. Cancer Res 31:1903, 1971.                 Cancer Inst 28:947, 1962.
78.   C. J. Bradley, G. S. Kledzik, and J. Meites: Pro-      92.   V. E. Archer, J. D. Gillam, and J. K. Wagoner:
      lactin and estrogen dependency of rat mammary                Respiratory disease mortality among uranium
      cancers at early and late stages of development.             workers. Ann N Y Acad Sci 271:280, 1976.
      Cancer Res 36:319, 1976.                               93.   W. J. Nicolson: Asbestos—The TLV approach.
79.   B. E. Henderson, M. C. Pike, V. R. Gerkins, and              Ann N Y Acad Sci 271:152, 1976.
      J. T. Casagrande: The hormonal basis of breast         94.   E. C. Hammond, I. J. Selikoff, and H. Seidman:
      cancer: Elevated plasma levels of estrogen,                  Asbestos exposure, cigarette smoking, and death
      prolactin, and progesterone. In H. H. Hiatt,                 rates. Ann N Y Acad Sci 330:473, 1979.
      J. D. Watson, and J. A. Winsten, eds.: Origins of      95.   R. L. Jolley: Chlorination effects on organic
      Human Cancer. Cold Spring Harbor, NY: Cold                   constituents in effluents from domestic sanitary
      Spring Harbor Laboratory, 1977, pp. 77–86.                   sewage treatment plants. Publication No. 565.
80.   M. R. Melamed: Diagnostic cytology of urinary                Oak Ridge, TN: Environmental Science Divi-
      tract carcinoma. Cancer 8:287, 1972.                         sion, Oak Ridge National Laboratory, 1973.
81.   International Agency for Research on Cancer,           96.   J. J. Rook: Formation of haloforms during
      Report of an IARC Working Group: An eval-                    chlorination of natural waters. J Soc Water Treat
      uation of chemicals and industrial processes                 Exam 23:234, 1974.
      associated with cancer in humans based on              97.   U.S. Environmental Protection Agency (EPA):
      human and animal data: IARC Monographs Vol-                  Preliminary assessment of suspected carcino-
      umes 1 to 20. Cancer Res 40:1, 1980.                         gens in drinking water. Report to Congress.
82.   M. Aksoy, E. Erdem, and G. Dincol: Leukemia                  Washington, D.C.: EPA, 1975.
      in shoe-workers exposed chronically to benzene.        98.   G. A. Junk and S. E. Stanley: Organics in drink-
      Blood 44:837, 1974.                                          ing water. Part I: Listing of identified chemi-
83.   P. F. Infante, J. K. Wagoner, R. A. Rinsky, and              cals. Springfield, VA: National Technical
      R. J. Young: Leukaemia in benzene workers.                   Information Services, 1975.
      Lancet 2:76, 1977.                                     99.   T. A. DeRouen and J. E. Diem: Relationships
84.   F. J. van Schooten, F. E. van Leeuwen, M. J. X.              between cancer mortality in Louisiana drinking
      Hillebrand, M. E. deRijke, A. A. M. Hart, et al.:            water source and other possible causative
      Determination of benzo[a]pyrene diol epoxide-                agents. In H. H. Hiatt, J. D. Watson, and
      DNA adducts in white blood cell DNA from                     J. A. Winsten, eds.: Origins of Human Cancer.
      coke-oven workers: the impact of smoking. J Natl             Cold Spring Harbor, NY: Cold Spring Harbor
      Cancer Inst 82:927, 1990.                                    Laboratory, 1977, pp. 331–345.
85.   E. S. Johnson, W. Parsons, C. R. Weinberg,            100.                                     ´
                                                                   J. Higginson and A. G. Oettle: Cancer inci-
      D. L. Shore, J. Mathews, et al.: Current serum               dence in the Bantu and ‘‘Cape Colored’’ races
      levels of 2,3,7,8-tetrachlorodibenzo-p-dioxin in             of South Africa: Report of a cancer survey
      phenoxy acid herbicide applicators and charac-               in the Transvaal (1953–55). J Natl Cancer Inst
      terization of historical levels. J Natl Cancer Inst          24:589, 1960.
      84:1648, 1992.                                        101.   J. Higginson and C. S. Muir: Environmental
86.   E. S. Johnson: Human exposure to 2,3,7,8-                    carcinogenesis: Misconceptions and limitations
      TCDD and risk of cancer. Crit Rev Toxicol 21:                to cancer control. J Natl Cancer Inst 63:1291,
      451, 1992.                                                   1979.
87.   H. I. Morrison, K. Wilkins, R. Semenciw, Y.           102.   J. Higginson: Perspectives and future develop-
      Mao, and D. Wigle: Herbicides and cancer. J Natl             ments in research on environmental carcino-
      Cancer Inst 84:1866, 1992.                                   genesis. In A. C. Griffin and C. R. Shaw, eds.:
88.   H. H. McDuffie, P. Pahwa, J. R. McLaughlin,                   Carcinogens: Identification and Mechanism
      J. J. Spinelli, S. Fincham, et al.: Non-Hodgkin’s            of Action. New York: Raven Press, 1979, pp.
      lymphoma and specific pesticide exposures in                  187–208.
114                                                                                         CANCER BIOLOGY

103. J. L. Lyon, M. R. Klauber, J. W. Gardner, M. P.       121. J. P. Neglia, A. T. Meadows, L. L. Robison, T. H.
     H. Melville, and C. R. Smart: Cancer incidence             Kim, W. A. Newton, F. et al.: Second neoplasms
     in Mormans and non-Mormans in Utah, 1966–                  after acute lymphoblastic leukemia in childhood.
     1970. N Engl J Med 294:129, 1976.                          N Engl J Med 325:1330, 1991.
104. A. J. Fox and A. M. Adelstein: Occupational           122. L. B. Travis, D. A. Hill, G. M. Dores, M.
     mortality: Work or way of life? J Epidemol Com-            Gospodarowicz, F. E. van Leeuwen, et al.:
     munity Health 32:73, 1978.                                 Breast cancer following radiotherapy and che-
105. W. Dubreuilh: Epitheliomatose d’origins so-                motherapy among young women with Hodgkin
     laire. Ann Dermatol Syphiliq 8:837, 1907.                  disease. JAMA 290:465, 2003.
106. G. M. Findlay: Ultra-violet light and skin cancer.    123. P. E. Oyer, E. B. Stinson, S. W. Jamieson, S. A.
     Lancet 2:1070, 1928.                                       Hunt, M. Billingham, et al.: Cyclosporin-A in
107. E. L. Scott and M. L. Straf: Ultraviolet radiation         cardiac allografting: A preliminary experience.
     as a cause of cancer. In H. H. Haitt, J. D. Watson,        Transplant Proc 15:1247, 1983.
     and J. A. Winsten, eds.: Origins of Human             124. International Agency for Research on Cancer,
     Cancer. Cold Spring Harbor, NY: Cold Spring                Report on an IARC Working Group: An evalu-
     Harbor Laboratory, 1977, pp. 529–546.                      ation of chemicals and industrial processes as-
108. S. Feygin: Du Cancer Radiologique. Paris:                  sociated with cancer in humans based on human
     Rousset, 1915.                                             and animal data: IARC Monographs, Vols.
109. H. C. March: Leukemia in radiologists. Radi-               1–20. Cancer Res 40:1, 1980.
     ology 43:275, 1944.                                   125. M. C. Pike, M. D. Krailo, B. E. Henderson,
110. H. S. Martland and R. E. Humphries: Osteo-                 J. T. Casagrande, and D. G. Hoel: ‘‘Hormonal’’
     genic sarcoma in dial painters using luminous              risk factors, ‘‘breast tissue age,’’ and the age-
     paint. Arch Pathol 7:406, 1929.                            incidence of breast cancer. Nature 303:767, 1983.
111. J. S. Horta, L. C. Da Motta, and M. H. Tavares:       126. P. C. MacDonald, C. D. Edman, D. L. Hemsell,
     Thorium dioxide effects in man. Epidemologi-               J. C. Porter, and P. K. Siiteri: Effect of obesity
     cal, clinical, and pathological studies. Environ           on conversion of plasma androstenedione to
     Res 8:131, 1974.                                           estrone in postmenopausal women with and
112. L. H. Hempelmann, W. J. Hall, M. Phillips, R. A.           without endometrial cancer. Am J Obstet Gyne-
     Cooper, and W. R. Ames: Neoplasms in persons               col 130:448, 1978.
     treated with X-rays in infancy: Fourth survey in      127. R. K. Ross, A. Paganini-Hill, P. C. Wan, and
     20 years. J Natl Cancer Inst 55:519, 1975.                 M. C. Pike: Effect of hormone replacement
113. A. C. Upton: Radiation effects. In H. H. Haitt,            therapy on breast cancer risk: Estrogen ver-
     J. D. Watson, and J. A. Winsten, eds.: Origins of          sus estrogen plus progestin. J Natl Cancer Inst
     Human Cancer. Cold Spring Harbor, NY: Cold                 92:328, 2000.
     Spring Harbor Laboratory, 1977, pp. 477–500.          128. C. Schairer, J. Lubin, R. Troisi, S. Sturgeon,
114. J. L. Kelsey and M. D. Gammon: The epide-                  L. Brinton, et al.: Menopausal estrogen and
     miology of breast cancer. CA Cancer J Clin                 estrogen–progestin replacement therapy and
     41:146, 1991.                                              breast cancer risk. JAMA 283:485, 2000.
115. N. H. Harley and J. H. Harley: Potential lung-        129. Million Women Study Collaborators: Breast
     cancer risk from indoor radon exposure. CA                 cancer and hormone-replacement therapy in
     Cancer J Clin 40:265, 1990.                                the Million Women Study. Lancet 362:419,
116. J. Schoenberg and J. Klotz: A case–control                 2003.
     study of radon and lung cancer among New              130. K. Armstrong, A. Eisen, and B. Weber: Primary
     Jersey women. New Jersey State Department of               care—Assessing the risk of breast cancer. N
     Health Technical Report, Phase I. Trenton, NJ:             Engl J Med 342:564, 2000.
     New Jersey State Department of Health, 1989.          131. A. L. Herbst, P. Cole, T. Colton, R. E. Scully,
117. W. J. Blot, Z.-Y. Xu, J. D. Boice, Jr., D.-Z. Zhao,        and S. J. Robboy: Age-incidence and risk of DES-
     B. J. Stone, et al.: Indoor radon and lung cancer          related clear cell adenocarcinoma of the vagina
     in China. J Natl Cancer Inst 12:1025, 1990.                and cervix. Am J Obstet Gynecol 128:43, 1976.
118. A. V. Nero, M. B. Schwehr, W. W. Nazaroff, and        132. B. Safai, B. Diaz, and J. Schwartz: Malignant
     K. L. Revzan: Distribution of airborne radon-              neoplasms associated with human immunode-
     222 concentrations in United States homes.                 ficiency virus infection. CA Cancer J Clin 42:74,
     Science 234:992, 1986.                                     1992.
119. Medicinal drugs.InB.W.Stewartand P.Kleihues,          133. W. B. Ershler and D. L. Longo: Aging and can-
     eds.: World Cancer Report, World Health                    cer: Issues of basic and clinical science. J Natl
     Organization. Lyon: IARC Press, 2003, pp.                  Cancer Inst 89:1489, 1997.
     48–50.                                                134. R. A. DePinho: The age of cancer. Nature 408:
120. C. A. DeLaat and B. C. Lampkin: Long-term                  248, 2000.
     survivors of childhood cancer: Evaluation and         135. A. F. Olumi, G. D. Grossfeld, S. W. Hayward,
     identification of sequelae of treatment. CA                 P. R. Carroll, T. D. Tlsty, and G. R. Cunha:
     Cancer J Clin 42:263, 1992.                                Carcinoma-associated fibroblasts direct tumor
THE EPIDEMIOLOGY OF HUMAN CANCER                                                                                  115

     progression of initiated human prostatic epithe-              rates in humans: Implications for carcinogene-
     lium. Cancer Res 59:5002, 1999.                               sis and for the role of exogenous factors in ‘‘spon-
136. D. H. Ly, D. J. Lockhart, R. A. Lerner, and                   taneous’’ germinal mutagenesis. Proc Natl Acad
     P. G. Schultz: Mitotic misregulation and human                Sci USA 89:7036, 1992.
     aging. Science 287:2486, 2000.                         150.   D. G. Hoel, N. L. Kaplan, and M. W. Anderson:
137. C. T. Murphy, S. A. McCarroll, C. I. Bargmann,                Implication of nonlinear kinetics on risk esti-
     A. Fraser, R. S. Kamath, et al.: Genes that act               mation incarcinogenesis. Science219:1032,1983.
     downstream of DAF-16 to influence the life-             151.   S. Woloshin, L. M. Schwartz, and H. G. Welch:
     span of Caenorhabditis elegans. Nature 424:                   Risk charts: Putting cancer in context. J Natl
     277, 2003.                                                    Cancer Inst 94:799, 2002.
138. J. Campisi: Aging, chromatin, and food                 152.   J. Kaiser: No meeting of minds on childhood can-
     restriction—Connecting the dots. Science 289:                 cer. Science 286:1832, 1999.
     2062, 2000.                                            153.   R. Twombly: Federal carcinogen report debuts
139. S. Dukan, A. Farewell, M. Ballesteros, F. Taddei,             new list of nominees. J Natl Cancer Inst 93:
     M. Radman, et al.: Protein oxidation in response              1372, 2001.
     to increased transcriptional or translational          154.   P. Boffetta, A. Agudo, W. Ahrens, E. Benhamou,
     errors. Proc Natl Acad Sci USA 97:5746, 2000.                 S. Benhamou, et al.: Multicenter case–control
140. S. Melov, J. Ravenscroft, S. Malik, M. S. Gill,               study of exposure to environmental tobacco
     D. W. Walker, et al.: Extension of life-span with             smoke and lung cancer in Europe. J Natl Cancer
     superoxide dismutase/catalase mimetics. Sci-                  Inst 90:1440, 1998.
     ence 289:1567, 2000.                                   155.   D. Wartenberg, E. E. Calle, M. J. Thun, C. W.
141. Genetic susceptibility. In B. W. Stewart and                  Heath, Jr., C. Lally, et al.: Passive smoking ex-
     P. Kleihues, eds.: World Cancer Report, World                 posure and female breast cancer mortality. J
     Health Organization. Lyon: IARC Press, 2003,                  Natl Cancer Inst 92:1666, 2000.
     pp. 71–75.                                             156.   P. Reynolds, S. Hurley, D. E. Goldberg, H.
142. A. G. Knudson, Jr.: Genetic predisposition to                 Anton-Culver, L. Bernstein, et al.: Active smok-
     cancer. In H. H. Hiatt, J. D. Watson, and J. A.               ing, household passive smoking, and breast
     Winstein, eds.: Origins of Human Cancer. Cold                 cancer: Evidence from the California Teachers
     Spring Harbor, NY: Cold Spring Harbor Lab-                    Study. J Natl Cancer Inst 96:29, 2004.
     oratory, 1977, pp. 45–52.                              157.   J. E. Entrom and G. C. Kabat: Environmental
143. H. T. Lynch, R. E. Harris, H. A. Giurgis, K.                  tobacco smoke and tobacco related mortality in
     Maloney, L. L. Carmody, et al.: Familial asso-                a prospective study of Californians, 1960–98.
     ciation of breast/ovarian carcinoma. Cancer                   BMJ 326:1–10, 2003.
     41:1543, 1978.                                         158.   J. M. Samet: Indoor radon exposure and lung
144. D. W. Nebert and W. W. Weber: Chapter 7. In                   cancer: Risky or not?—All over again. J Natl Can-
     W. B. Pratt and P. Taylor, eds.: Pharmacoge-                  cer Inst 89:4, 1997.
     netics in Principles of Drug Action: The Basis         159.   H. Frumkin and J. M. Samet: Radon. CA Can-
     of Pharmacology. New York: Churchill Living-                  cer J Clin 51:337, 2001.
     stone, 1990, pp. 469–531.                              160.                        ¨     ¨
                                                                   A. Auvinen, I. Maekelainen, M. Hakama, O.
144a. X. Wu, M. R. Spitz, C. I. Amos, J. Lin, L. Shao,                   ´n,
                                                                   Castre E. Pukkala, et al.: Indoor radon exposure
       et al.: Mutagen sensitivity has high heritability:          and risk of lung cancer: A nested case–control
       Evidence from a twin study. Cancer Res 66:                  study in Finland. J Natl Cancer Inst 88:966, 1996.
       5993, 2006.                                          161.   J. H. Lubin and J. D. Boice, Jr.: Lung cancer
145. R.Doll:Introduction,InH.H.Hiatt,J.D.Watson,                   risk from residential radon: Meta-analysis of
     and J. A. Winstein, eds.: Origins of Human                    eight epidemiologic studies. J Natl Cancer Inst
     Cancer. Cold Spring Harbor, NY: Cold Spring                   89:49, 1997.
     Harbor Laboratory, 1977, pp. 1–12.                     162.   F. Pisa, F. Barbone, A. Betta, et al.: Residential
146. B. N. Ames: Identifying environmental chemi-                  radon and risk of lung cancer in an Italian alpine
     cals causing mutations and cancer. Science                    area. Arch Environ Health 56:208, 2001.
     204:587, 1979.                                         163.   F. Lagarde, G. Axelsson, L. Damber, et al.: Res-
147. B. N. Ames, W. E. Durston, E. Yamasaki, and                   idential radon and lung cancer among never
     F. D. Lee: Carcinogens are mutagens: A simple                 smokers in Sweden. Epidemiology 12:396, 2001.
     test system combining liver homogenates for            164.   J. S. Neuberger and T. F. Gesell: Childhood
     activation and bacteria for detection. Proc Natl              cancers, radon, and g radiation. Lancet 360:1437,
     Acad Sci USA 70:2281, 1973.                                   2002.
148. J. H. Weisburger and G. M. Williams: Carcin-           165.   C. Johansen, J. D. Boice, Jr., J. K. McLaughlin,
     ogen testing: Current problems and new ap-                    and J. H. Olsen: Cellular telephones and
     proaches. Science 214:401, 1981.                              cancer—A nationwide cohort study in Denmark.
149. R. D. Kuick, J. V. Neel, J. R. Strahler, E. H. Y.             J Natl Cancer Inst 93:203, 2001.
     Chu, R. Bargal, et al.: Similarity of spontaneous      166.   M. Feychting and A. Ahibom: Magnetic fields
     germinal and in vitro somatic cell mutation                   and cancer in children residing near Swedish
116                                                                                        CANCER BIOLOGY

     high-voltage power lines. Am J Epidemiol 138:              pounds and the subsequent development of
     467, 1993.                                                 breast cancer. Cancer Epidemiol Biomarkers
167. M. S. Linet, E. E. Hatch, R. A. Kleinerman, L. L.          Prev 8:525, 1999.
     Robison, W. T. Kaune, et al.: Residential expo-     179.   M. D. Gammon, R. M. Santella, A. I. Neugut,
     sure to magnetic fields and acute lymphoblastic             S. M. Eng, S. L. Teitelbaum, et al.: Environ-
     leukemia in children. N Engl J Med 337:1, 1997.            mental toxins and breast cancer on Long Island.
168. M. Longnecker: Alcoholic beverage consump-                 I. Polycyclic aromatic hydrocarbon DNA ad-
     tion in relation to risk of breast cancer: Meta-           ducts. Cancer Epidemiol Biomarkers Preven-
     analysis and review. Cancer Causes Control                 tion 11:677, 2002.
     5:73, 1994.                                         180.   A. S. Robbins, S. Brescianini, and J. L. Kelsey:
169. S. A. Smith-Warner, D. Spiegelman, S. S. Yaun,             Regional differences in known risk factors and
     P. A. van den Brandt, A. R. Folsom, et al.:                the higher incidence of breast cancer in San
     Alcohol and breast cancer in women: A pooled               Francisco. J Natl Cancer Inst 89:960, 1997.
     analysis of cohort studies. JAMA 279:535, 1998.     181.   D. K. Mirick, S. Davis, and D. B. Thomas:
170. C. Byrne, P. M. Webb, T. W. Jacobs, G. Peiro,              Antiperspirant use and the risk of breast cancer.
     S. J. Schnitt, et al.: Alcohol consumption and             J Natl Cancer Inst 94:1578, 2002.
     incidence of benign breast disease. Cancer Epi-     182.   H. Komulainen, V. M. Kosma, S. L. Vaitti-
     demiol Biomarkers Prev 11:1369, 2002.                      nen, T. Vartianen, E. Kaliste-Korhonen, et al.:
171. H. S. Feigelson, C. R. Jonas, A. S. Robertson,             Carcinogenicity of the drinking water mutagen
     M. L. McCullough, M. J. Thun, et al.: Alcohol,             3-chloro-4-(dichloromethyl)–5-hydroxy-2(5H)-
     folate, methionine and risk of incident breast             furanone in the rat. J Natl Cancer Inst 89:848,
     cancer in the American Cancer Society Cancer               1997.
     Prevention Study II Nutrition Cohort. Cancer        183.   R. L. Melnick, G. A. Boorman, and V. Dellarco:
     Epidemiol Biomarkers Prev 12:161, 2003.                    Water chlorination, 3-Chloro-4-(dichloromethyl)-
172. T. Zheng, T. R. Holford, S. T. Mayne, J. Tessari,          5-hydroxy-2(5H)-furanone (MX), and potential
     B. Ward, et al.: Risk of female breast cancer              cancer risk. J Natl Cancer Inst 89:832, 1997.
     associated with serum polychlorinated biphenyls     184.   M. Mahue-Giangreco, G. Ursin, J. Sullivan-
     and 1,1-Dichloro-2,20 -bis(p-chlorophenyl)ethyl-           Halley, and L. Bernstein: Induced abortion, mis-
     ene. Cancer Epidemiol Biomarkers Prev 9:167,               carriage, and breast cancer risk of young women.
     2000.                                                      Cancer Epidemiol Biomarkers Prev 12:209,
173. Breast cancer in New York State by zip code,               2003.
     1993–1997. New York State Department of             185.   M. Camus, J. Siemiatycki, and B. Meek: Nonoc-
     Health, 2000. Access at            cupational exposure to chrysotile asbestos and                                                    the risk of lung cancer. N Engl J Med 338:1565,
174. R. Trombley: Long Island study finds no link                1998.
     between pollutants and breast cancer. J Natl        186.   S. Takayama, S. M. Sieber, R. H. Adamson, U. P.
     Cancer Inst 94:1348, 2002.                                 Thorgeirsson, D. W. Dalgard, et al.: Long-term
175. A. Demers, P. Ayotte, J. Brisson, S. Dodin, J.             feeding of sodium saccharin to nonhuman
     Robert, et al.: Risk and aggressiveness of breast          primates: Implications for urinary tract cancer.
     cancer in relation to plasma organochlorine                J Natl Cancer Inst 90:19, 1998.
     concentrations. Cancer Epidemiol Biomarkers         187.   M. Mitka: Fear of frying: Is acrylamide in foods
     Prev 9:161, 2000.                                          a cancer risk? JAMA 288:2105, 2002.
176. F. Laden, G. Collman, K. Iwamoto, A. J. Alberg,     188.   Health scares based on animal carcinogen
     G. S. Berkowitz, et al.: 1,1-Dichloro-2,2-bis              testing. In K. Meister, E. M. Whelan, G. L. Ross,
     ( p-chlorophenyl)ethylene and polychlorinated              and A. N. Stimola, eds.: America’s War on
     biphenyls and breast cancer: Combined analysis             Cancer. New York: American Council on Sci-
     of five U.S. studies. J Natl Cancer Inst 93:768,            ence and Health, 2005, pp. 117–137.
     2001.                                               189.   H. D. Strickler, J. J. Goedert, S. S. Devesa,
177. E. M. Ward, P. Schulte, B. Grajewski, A. Ander-            J. Lahey, J. F. Fraumeni, et al.: Trends in U.S.
     sen, D. Patterson, Jr., et al.: Serum organochlo-          pleural mesothelioma incidence rates follow-
     rine levels and breast cancer: A nested case–              ing simian virus 40 contamination of early
     control study of Norwegian women. Cancer                   poliovirus vaccines. J Natl Cancer Inst 95:38,
     Epidemiol Biomarkers Prev 9:1357, 2000.                    2003.
177a. D. M. Winn: The Long Island Breast Cancer          190.   E. A. Engels, H. A. Katki, N. M. Nielsen, J. F.
       Study Project. Nature Rev Cancer 5:986,2005.             Winther, H. Hjalgrim, et al.: Cancer incidence
178. K. J. Helzlsouer, A. J. Alberg, H-Y. Huang,                in Denmark following exposure to poliovirus
     S. C. Hoffman, P. T. Strickland, et al.: Se-               vaccine contaminated with simian virus 40. J
     rum concentrations of organochlorine com-                  Natl Cancer Inst 95:532, 2003.

The Biochemistry and Cell
Biology of Cancer

HISTORICAL PERSPECTIVES                               C. elegans found genes involved in apoptosis.12
                                                      The cell cycle began to be studied in great detail
The development of knowledge about the bio-           in lower organisms, and organisms such as clams,
chemistry and cell biology of cancer comes from       yeast, and fruit flies have contributed greatly to
a number of disciplines. Some of this knowledge       our understanding of the cell cycle events.13
has come from research initiated a century or            The findings that simple molecules like cyclic
more ago. There has been a flow of information         AMP could direct a whole panoply of cellular
about genetics into a knowledge base about can-       functions14 led to the discovery of signal trans-
cer, starting with Gregor Mendel and the dis-         duction pathways, which are now becoming fa-
covery of the principle of inherited traits and       vored molecular targets for anticancer drug dis-
leading through Theodor Boveri’s work on the          covery.
chromosomal mode of heredity and chromo-                 Much of what we originally knew about the
somal damage in malignant cells1 to Avery’s dis-      biochemical differences between normal and
covery of DNA as the hereditary principle,2           malignant cells, however, was discovered in their
Watson and Crick’s determination of the struc-        patterns of enzymatic activity. In the 1920s,
ture of DNA,3 the human genome project, DNA           Warburg studied glycolysis in a wide variety of
microarrays, and proteomics. Not only has this        human and animal tumors and found that there
information provided a clearer picture of the         was a general trend toward an increased rate of
carcinogenic process, it has also provided better     glycolysis in tumor cells.15 He noted that when
diagnostic approaches and new therapeutic tar-        normal tissue slices were incubated in a nutrient
gets for anticancer therapies.                        medium containing glucose, but without oxygen,
   Once cell culture techniques were developed4       there was a high rate of lactic acid production
it became possible to test which genes are in-        (anaerobic glycolysis); however, if they were
volved in malignant transformation and progres-       incubated with oxygen, lactic acid production
sion. This field of research led to the discovery of   virtually stopped. The rate of lactic acid pro-
oncogenes5 and tumor suppressor genes.6 He-           duction was higher in tumor tissue slices in the
reditary studies led to the two-hit theory7 and the   absence of oxygen than in normal tissues, and
concept of the hereditary nature of some can-         the presence of oxygen slowed, but did not elim-
cers.8 Chromosomal staining techniques enabled        inate, lactic acid formation in the tumor slices.
Nowell and Hungerford9 and Rowley10 to iden-          Warburg concluded that cancer cells have an ir-
tify chromosomal translocation as a tumor initi-      reversible injury to their respiratory mechanism,
ating event.                                          which increases the rate of lactic acid production
   Studies in yeast produced the concept of cell      even in the presence of oxygen (aerobic glyco-
cycle checkpoints,11 and investigations with          lysis). He regarded the persistence of this type of

118                                                                                  CANCER BIOLOGY

glycolysis as the crucial biochemical lesion in        duction of the end products of glycolysis, lipid
neoplastic transformation. This old idea still has     levels indicative of cell membrane turnover,
some credence in that there are hypoxic areas in       and alterations in amino acids and nucleotide
the core of tumors, where anaerobic metabo-            levels.18
lism predominates. This has clinical implications         Since mitochondria contain the enzymatic cas-
because hypoxic cells do not respond as well to        cades for oxidative metabolism, it has been sug-
certain anticancer drugs or radiation therapy.         gested that damage to mitochondria may be in-
The ability of lactate and pyruvate, end points of     volved in the disruptions of oxidative metabolism
glycolysis, to enhance tumor progression appears       seen in malignant tumors. Mutations of mito-
to be mediated by the activation of hypoxia in-        chondrial DNA (mtDNA) has been observed in
ducible factor-1 (HIF-1).16 In addition to in-         a variety of human cancers, including bladder,
creased activity of enzymes of the glycolytic path-    head and neck, lung,18 and ovarian19 cancers. In-
way, such as hexokinase, phosphofructokinase,          terestingly, in the bladder cancers, the mutation
and pyruvate kinase in cancer cells, hypoxia is        hot spots were primarily in a nicotinamide ade-
also a common feature of many human solid              nine dinucleotide dehydrogenase subunit, a key
cancers. These effects have been linked to tu-         component of the electron transfer machinery.
mor progression, metastasis, and multidrug resis-      This suggests a mechanism for the alterations
tance.17 Interestingly, oncogenes such as ras, src,    in oxidative metabolism seen in malignant cells.
and myc enhance aerobic glycolysis by increasing       Because mitochondrial DNA is exposed to high
the expression of glucose transporters and glyco-      levels of reactive oxygen species generated dur-
lytic enzymes (reviewed in Reference 16).              ing oxidative phosphorylation, it is not surprising
   Cancer cells react to hypoxic conditions by up-     that mtDNA is highly susceptible to mutational
regulating expression of HIF-1, which is a tran-       events. The mutational rate of mtDNA has been
scription factor that in turn up-regulates expres-     estimated to be 10 times higher than that of nu-
sion of genes involved in glycolysis, glucose          clear DNA.19 Mitochondria also play a key role in
transport (GLUT-1), angiogenesis (VEGF), cell          apoptosis (see section on apoptosis below), and
survival, and erythropoiesis. HIF-1 expression         alterations in those mitochondria-mediated events
has been observed in cancers of the brain, breast,     are seen in cancer cells.
colon, lung, ovary, and prostate and their metas-         In the early 1950s, Greenstein formulated the
tases but not in the corresponding normal tis-         ‘‘convergence hypothesis’’ of cancer, which states
sues. Its expression in tumors correlates with         thatthe enzymatic activity of malignantneoplasms
poor prognosis.                                        tends to converge to a common pattern.20 Al-
   Interest in tumor metabolism has been stimu-        though he recognized some exceptions to thisrule,
lated once again by modern techniques such as          he considered the generalization, based mostly
position emission tomography (PET), sensitive          on repeatedly transplanted tumor models, to
mass spectrometry (MS), and high-resolution nu-        be valid. It is now more fully appreciated that
clear magnetic resonance spectroscopy (NMR).           even though cancer cells do have some commonly
PET uses fluorine-18 labeled fluorodeoxyglucose          increased metabolic pathways, such as those in-
(FdG) to detect tissue regions of high glucose up-     volved in nucleic acid synthesis, there is tremen-
take, which is indicative of up-regulated glycolysis   dous biochemical heterogeneity among malig-
and increased metabolic rate. FdG PET imaging          nant neoplasms, and that there are many fairly
has shown that most primary and metastatic hu-         well-differentiated cancers that do not have
man cancers have increased glucose uptake.17           the common enzymatic alterations he suggested.
This finding is indicative of a ‘‘glycolytic switch’’   Thus, cancers do not have a universally uniform
in cancer cells and may be a precursor of tu-          malignant phenotype as exemplified by their en-
mor angiogenesis and metastasis.17                     zyme patterns.
   NMR and MS can now be used to measure                  On the basis of work of about 60 years ago,
mestatic profiles of cancer cells and the meta-         which evolved from studies on the production of
bolic phenotype of tissues and organs. This so-        hepatic cancer by feeding aminoazo dyes, the
called science of ‘‘metabolomics’’ can provide         Millers advanced the ‘‘deletion hypothesis’’ of
metabolic biomarkers of tumors such as pro-            cancer.21 This hypothesis was based on the ob-
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                            119

servation that a carcinogenic aminoazo dye cova-      of cancer, which states that ‘‘the biochemical
lently bound liver proteins in animals undergoing     strategy of the genome in neoplasia could be
carcinogenesis, whereas little or no dye binding      identified by elucidation of the pattern of gene
occurred with the protein of tumors induced by        expression as revealed in the activity, concen-
the dye. They suggested that carcinogenesis re-       tration, and isozyme aspects of key enzymes and
sulted from ‘‘a permanent alteration or loss of       their linking with neoplastic transformation
protein essential for the control of growth.’’        and progression.’’26 Weber proposed three gen-
    About 10 years later, Potter suggested that       eral types of biochemical alterations associated
the proteins lost during carcinogenesis may be        with malignancy: (1) transformation-linked al-
involved in feedback control of enzyme systems        terations that correlate with the events of ma-
required for cell division,22 and he proposed the     lignant transformation and that are probably al-
‘‘feedback deletion hypothesis.’’23 In this hypoth-   tered in the same direction in all malignant cells;
esis, Potter postulated that ‘‘repressors’’ crucial   (2) progression-linked alterations that correlate
to the regulation of genes involved in cell pro-      with tumor growth rate, invasiveness, and me-
liferation are lost or inactivated by the action of   tastatic protential; and (3) coincidental alterations
oncogenic agents on the cell, either by interact-     that are secondary events and do not correlate
ing with DNA to block repressor gene tran-            strictly with transformation or progression. We-
scription or by reacting directly with repressor      ber maintained that key enzymes, that is enzymes
proteins and inactivating them. This prediction       involved in the regulation of rate and direction of
anticipated the discovery of tumor suppressor         flux of competing synthetic and catabolic path-
proteins, such as p53 and RB, by about 25 years.      ways, would be the enzymes most likely to be
    Biochemical studies of cancer were also aided     altered in the malignant process. In contrast,
by the so-called minimal-deviation hepatomas          ‘‘non-key’’ enzymes, that is, enzymes that are not
developed by Morris and colleagues.24 These           rate limiting and do not regulate reversible equi-
tumors were originally induced in rats by feed-       librium reactions, would be of lesser importance.
ing them the carcinogens fluorenylphthalamic           As one would expect, a number of enzyme ac-
acid, fluorenylacetamide compounds, or trime-          tivities that Weber and others have found to be
thylaniline. These hepatocellular carcinomas are      altered in malignant cells are those involved in
transplantable in an inbred host strain of rats       nucleic acid synthesis and catabolism. In gen-
and have a variety of growth rates and degrees of     eral, the key enzymes in the de novo and salvage
differentiation. They range from slowly growing,      pathways of purine and pyrimidine biosynthesis
well-differentiated, karyotypically normal cells      are increased and the opposing catabolic en-
to rapidly growing, poorly differentiated, poly-      zymes are decreased during malignant transfor-
ploid cells. All these tumors are malignant and       mation and tumor progression. Weber noted that
eventually kill the host. The term ‘‘minimal de-      the degree of neoplasia was related to the con-
viation’’ was coined by Potter 23 to convey the       centrations of certain regulators of key metabolic
idea that some of these neoplasms differ only         pathways. The question of why anaplastic, rap-
slightly from normal hepatic parenchymal cells.       idly growing tumors tend to be biochemically
The hypothesis was that if the biochemical            alike, whereas more well-differentiated tumors
lesions present in the most minimally deviated        display a vast array of phenotypic characteristics,
neoplasm could be identified, the crucial chan-        was approached by Knox.27 He thought that the
ges defining the malignant phenotype could be          vast bulk of biochemical components in tumor
determined. As Weinhouse25 indicated, stud-           tissues are ‘‘normal,’’ in the sense that they are
ies of these tumors greatly advanced our knowl-       produced by certain specialized adult normal
edge of the biochemical characteristics of the        cells or by normal cells at some stage of their
malignant phenotype, and they have ruled out          differentiation. In cancer cells, it is the combi-
many secondary or nonspecific changes that relate      nation and proporations of these normal com-
more to tissue growth rate than to malignancy.        ponents that are abnormal. The biochemical di-
    The extensive biochemical analyses of the         versity of cancer cells, then, would depend on the
Morris minimal-deviation hepatomas led Weber          cell of origin of the neoplasm and its degree of
to formulate the ‘‘molecular correlation concept’’    neoplasticity.22 All too frequently, even now, in
120                                                                                    CANCER BIOLOGY

the histopathologic or biochemical character-          be involved in genetic instability (rapidly pro-
ization of cancer, a biochemical component that        liferating cells have less time to repair DNA
is present or absent or increased or decreased is      damage) and tumor vascularization that leads to
not considered in relation to the particular cell of   dysregulated proliferation of cells, which in turn
origin of a tumor, its differentiation state, or its   eats up more oxygen, creates hypoxia, and turns
degree of neoplasticity.                               on HIF-1 and additional angiogenesis. Similarly,
   Taken together, the data on enzyme patterns         genes involved in tumor cell invasion may also
of cancer cells indicate that undifferentiated,        be involved in loss of growth control (invasive
highly malignant cells tend to resemble one an-        cells have acquired the skills to survive in ‘‘hos-
other and fetal tissues more than their adult nor-     tile’’ new environments) and evasion of apoptosis
mal counterpart cells, whereas well-differentiated     (less cell death even in the face of a normal rate of
tumors tend to resemble their cell of origin more      cell proliferation produces more cells). As will be
than other tumors. Of course, between these two        described in more detail below, the molecular
extremes several levels of neoplastic gradation        genetic alterations of cancer cells lead to cells
occur, leading to the vast biochemical heteroge-       that can generate their own growth-promoting
neity of tumors. This heterogeneity also exists for    signals, are less sensitive to cell cycle checkpoint
tumors of the same tissue type arising in different    controls, evade apoptosis, and thus have almost
patients or even in the same patient at different      limitless replication potential. The signal trans-
stages of the disease.                                 duction mechanisms involved in this replication
   The fact that more undifferentiated tumors          potential will also be discussed in this chapter. As
tend to converge to a more fetal-like state is evi-    will become clearer, these signaling pathways are
denced by a frequently observed production of          interlinked. As was not initially realized, cancer
oncodevelopmental gene products. A number of           cells have multiple proliferative pathways and can
cancer cell characteristics, such as invasiveness      bypass an interdiction of one or more of these.
and ‘‘metastasis,’’ are also seen in embryonic tis-    This redundancy makes design of effective signal
sue. For example, the developing trophoblast           transduction-targeted chemotherapeutic drugs
invades the uterine wall during the implantation       that target a single pathway very difficult indeed.
step of embryonic development. During organ-              Cancer cells can also subvert the environment
ogenesis, embryonic cells dissociate themselves        in which they proliferate. Alterations in both cell–
from the surrounding cells and migrate to new          cell and cell–extracellular matrix interactions also
locations, a process not unlike metastasis.            occur, leading to creation of a cancer-facilitating
                                                       environment. For example, a common alteration
                                                       in epithelial carcinomas is alteration of E-cad-
GROWTH CHARACTERISTICS                                 herin expression. E-cadherin is a cell–cell adhe-
OF MALIGNANT CELLS                                     sion molecule found on all epithelial cells.
                                                          Cancer cells exhibit remarkable plasticity.
As will be discussed more in Chapter 5, most           Malignant cells have the ability to mimic some
cancers (other than those for which there is a         of the characteristics of other cell types as they
dominantly inherited cancer susceptibility gene)       progress and became less well differentiated.
are an acquired molecular genetic disease in           For example, cancer cells may assume some of
which a single (or a few) clone(s) of cells accu-      the structure and function of vascular cells.28 As
mulate cellular genetic changes that progress to       cancer cells metastasize, they may eventually
the full-blown cancer phenotype. Cancer can be         take on a new phenotype such that the tissues of
characterized as a disease of genetic instabil-        origin may become unclear—so-called cancers
ity, altered cellular behavior, and altered cell–      of unknown primary site.
extracellular matrix interactions.28 These alter-
ations lead to dysregulated cell proliferation,
                                                       Phenotypic Alterations in Cancer Cells
and ultimately to invasion and metastasis. There
are interactions between the genes involved in         Treatment of animals or cells in culture with
these steps. For example, the genes associated         carcinogenic agents is a means of studying dis-
with loss of control of cell proliferation may also    crete biochemical events that lead to malignant
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                           121

transformation. Studies of cell transformation in     transformed cells that correlate with the growth
vitro, however, have many pitfalls. These ‘‘tissue    of a cancer in vivo. This research has tremen-
culture artifacts’’ include overgrowth of cells not   dously increased our knowledge of the biochem-
characteristic of the original population of cul-     istry of cancer cells. However, many of the bio-
tured cells (e.g., overgrowth of fibroblasts in cul-   chemical characteristics initially thought to be
tures that were originally primarily epithelial       closely associated with the malignant phenotype
cells), selection for a small population of variant   of cells in culture have subsequently been found
cells with continued passage in vitro, or appear-     to be dissociable from the ability of those cells
ance of cells with an abnormal chromosomal            to produce tumors in animals. Furthermore, in-
number or structure (karyotype). Such changes         dividual cells of malignant tumors growing in
in the characteristics of cultured cell popula-       animals or in humans exhibit marked biochemi-
tions can lead to ‘‘spontaneous’’ transformation      cal heterogeneity, as reflected in their cell surface
that mimics some of the changes seen in popu-         composition, enzyme levels, immunogenicity, re-
lations of cultured cells treated with oncogenic      sponse to anticancer drugs, and so on. This has
agents. Thus, it is often difficult to sort out the    made it extremely difficult to identify the essential
critical malignant events from the noncritical        changes that produce the malignant pheno-
ones. Although closer to the carcinogenic pro-        type. Hahn et al.30 have shown that ectopic ex-
cess in humans, malignant transformation in-          pression ofthe human telomerasecatalytic subunit
duced in vivo by treatment of susceptible ex-         (hTERT) in combination with the oncogenes h-
perimental animals with carcinogenic chemicals        ras and SV40 virus large-T antigen can induce
or oncogenic viruses or by irradiation is even        tumorigenic conversion in normal human epi-
more difficult because it is hard to discriminate      thelial and fibroblast cells, suggesting that dis-
toxic from malignant events and to determine          ruption of the intracellular pathways regulated by
what role a myriad of factors, such as the nu-        these gene products is sufficient to produce a
tritional state of the animal, hormone levels, or     malignant cell.
endogenous infections with microorganisms or             Since hyperproliferative conditions in patients,
parasites, might have on the in vivo carcinogenic     such as inflammatory bowel disease or psoriasis,
events. Moreover, tissues in vivo are a mixture of    may mimic some of the characteristics of malig-
cell types, and it is difficult to determine in        nant cells, it is important to use a number of cri-
which cells the critical transformation events are    teria that define the malignant state. The evidence
occurring and what role the microenvironment          that these phenotypic properties found in trans-
of the tissue plays. Thus, most studies designed      formed cells in culture are related to malignant
to identify discrete biochemical events occur-        neoplasia in vivo is discussed below.
ring in cells during malignant transformation
have been done with cultured cells, since clones
                                                      Immortality of Transformed
of relatively homogeneous cell populations can
                                                      Cells in Culture
be studied and the cellular environment defined
and manipulated. The ultimate criterion that          Most normal diploid mammalian cells have a
establishes whether cells have been transformed,      limited life expectancy in culture. For example,
however, is their ability to form a tumor in          normal human fibroblast lines may live for 50 to
an appropriate host animal. The generation of         60 population doublings (the ‘‘Hayflick index’’),
immortalized ‘‘normal’’ cell lines of a given dif-    but then viability begins to decrease rapidly un-
ferentiated phenotype from human embryonic            less they transform spontaneously or are trans-
stem cells has enhanced the ability to study cells    formed by oncogenic agents. However, malig-
of a normal genotype from a single source.29          nant cells, once they become established in
Such cell lines may also be generated by trans-       culture, will generally live for an indefinite num-
fection of the telomerase gene into cells to          ber of population doublings, provided the right
maintain chromosomal length.                          nutrients and growth factors are present. It is not
   Over the past 60 years, much scientific ef-         clear what limits the life expectancy of normal
fort has gone into research aimed at identify-        diploid cells in culture, but it may be related
ing the phenotypic characteristics of in vitro–       to the continual shortening of chromosomal
122                                                                                    CANCER BIOLOGY

telomeres each time cells divide. Transformed          grow well when they are suspended in fluid or a
cells are known to have elevated levels of telo-       semisolid agar gel. If these cells make contact
merase that maintain telomere length. Trans-           with a suitable surface, however, they attach,
formed cells that become established in culture        spread, and proliferate. This type of growth is
also frequently undergo karyotypic changes, usu-       called anchorage-dependent growth. Many cell
ally marked by an increase in chromosomes              lines derived from tumors and cells transformed
(polyploidy), with continual passage. This sug-        by oncogenic agents are able to proliferate in sus-
gests that cells with increased amounts of certain     pension cultures or in a semisolid medium (meth-
growth-promoting genes are generated and/or            ylcellulose or agarose) without attachment to a
selected during continual passage in culture. The      surface. This is called anchorage-independent
more undifferentiated cells from cancers of ani-       growth. This property of transformed cells has
mals or patients also often have an atypical kar-      been used to develop clones of malignant cells.35
yology, thus the same selection process may be         This technique has been widely used to compare
going on in vivo with progression over time of         the growth properties of normal and malignant
malignancy from a lower to a higher grade.             cells. Another advantage that has been derived
                                                       from the ability of malignant cells to grow in soft
                                                       agar (agarose) is the ability to grow cancer cells
Decreased Requirement
                                                       derived from human tumors to test their sensi-
for Growth Factors
                                                       tivity to chemotherapeutic agents and to screen
Other properties that distinguish transformed          for potential new anticancer drugs.36
cells from their nontransformed counterparts are
decreased density-dependent inhibition of pro-
                                                       Loss of Cell Cycle Control and
liferation31 and the requirement for growth fac-
                                                       Resistance to Apoptosis
tors for replication in culture. Cells transformed
by oncogenic viruses have lower serum growth           This topic will be discussed in more detail below.
requirements than do normal cells.32 For exam-         Suffice it to say here that normal cells respond
ple, 3T3 fibroblasts transformed by SV40, poly-         to a variety of suboptimal growth conditions
oma, murine sarcoma virus, or Rous sarcoma             by entering a quiescent phase in the cell division
virus are all able to grow in a culture medium that    cycle, the G0 state. There appears to be a deci-
lacks certain serum growth factors, whereas un-        sion point in the G1 phase of the cell cycle, at
infected cells are not (reviewed in Reference 33).     which time the cell must make a commitment to
   Cancer cells may also produce their own growth      continue into the S phase, the DNA synthesis
factors that may be secreted and activate prolif-      step, or to stop in G1 and wait until conditions are
eration in neighboring cells (paracrine effect) or,    more optimal for cell replication to occur. If this
if the same malignant cell type has both the re-       waiting period is prolonged, the cells are said to
ceptor for a growth factor and the means to            be in a G0 phase. Once cells make a commitment
produce the factor, self-stimulation of cell prolif-   to divide, they must continue through S, G2, and
eration (autocrine effect) may occur. One exam-        M to return to G1. If the cells are blocked in S,
ple of such an autocrine loop is the production of     G2, or M for any length of time, they die. The
tumor necrosis factor-alpha (TNF-a) and its re-        events that regulate the cell cycle, called cell cycle
ceptor TNFR1 by diffuse large cell lymphoma.34         checkpoints, are defined in more detail below.
Co-expression of TNFa and its receptor are neg-        This loss of cell cycle check point control by
ative prognostic indicators of survival, suggesting    cancer cells may contribute to their increased
that autocrine loops can be powerful stimuli for       susceptibility to anticancer drugs. Normal cells
tumor aggressiveness and thus potentially impor-       have mechanisms to protect themselves from
tant diagnostic and therapeutic targets.               exposure to growth-limiting conditions or toxic
                                                       agents by calling on these check point control
                                                       mechanisms. Cancer cells, by contrast, can con-
Loss of Anchorage Dependence
                                                       tinue through these checkpoints into cell cycle
Most freshly isolated normal animal cells and          phases that make them more susceptible to the
cells from cultures of normal diploid cells do not     cytotoxic effects of drugs or irradiation. For ex-
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                            123

ample, if normal cells accrue DNA damage due            mal cells have been shown to react with Lewis
to ultraviolet (UV) or X-irradiation, they arrest in    blood group antigens, such as Lex, Lea, Leb, or
G1 so that the damaged DNA can be repaired              their analogues.37
prior to DNA replication. Another check point in           The biochemical characterization or the aber-
the G2 phase allows repair of chromosome breaks         rant glycosylation of glycoproteins was also dem-
before chromosomes are segregated at mitosis.           onstrated in earlier studies. The presence of
Cancer cells, which exhibit poor or absent check        high-molecular-weight glycopeptides with altered
point controls, proceed to replicate the damaged        glycosylation patterns was detected on trans-
DNA, thus accounting for persisting and accu-           formed cells in early studies before they were
mulating mutations.                                     clearly chemically identified.38,39 Later, the chem-
                                                        ical basis for some of the changes in tumor cell
                                                        glycoproteins was attributed to the fact that the
Changes in Cell Membrane
                                                        N-linked oligosaccharides of tumor cells contain
Structure and Function
                                                        more multiantennary structures than the oligo-
The cell surface membrane (plasma membrane)             saccharides derived from normal cells.40
plays an important role in the ‘‘social’’ behavior of      Tumor-associated carbohydrate antigens can
cells, that is, communication with other cells, cell    be classified into three groups37: (1) epitopes
movement and migration, adherence to other              expressed on both glycolipids and glycoproteins,
cells or structures, access to nutrients in the mi-     (2) epitopes expressed only on glycolipids, and
croenvironment, and recognition by the body’s           (3) epitopes expressed only on glycoproteins. To
immune system. Alterations of the plasma mem-           the first group belongs the lacto-series structure
brane in malignant cells may be inferred from a         that is found in the most common human cancers,
variety of properties that characterize their           such as lung, breast, colorectal, liver, and pan-
growth and behavior, for example, the loss of           creatic cancers. The common backbone structure
density-dependent inhibition of growth, de-             for these epitopes is Ga1b1?3G1cNAcb1?
creased adhesiveness, loss of anchorage depen-          3Ga1 (type 1 blood group) or Ga1b1?4 G1cNac
dence, and invasiveness through normal tissue           b1?3 Ga1 (type 2 blood group). The second
barriers. In addition, a number of changes in the       group of epitopes, expressed exclusively on gly-
biochemical characteristics of malignant cells’         colipids, is mostly on the ganglio- or globo-series
surfaces have been observed. These include ap-          structures. This series of epitopes is expressed
pearance of new surface antigens, proteoglycans,        abundantly only on certain types of human can-
glycolipids, and mucins, and altered cell–cell and      cers, such as melanoma, neuroblastoma, small cell
cell–extracellular matrix communication.                lung carcinoma, and Burkitt’s lymphoma. The
                                                        third group of epitopes, seen only on glycopro-
                                                        teins, consists of the multiantennary branches of
     Alterations in Cell Surface
                                                        N-linked carbohydrates and the alterations of O-
     Glycolipids, Glycoproteins,
                                                        linked carbohydrate chains seen in some mucins.
     Proteoglycans, and Mucins
                                                           Tumor-associated carbohydrate antigens can
Aberrant glycosylation was first suggested as            also be classified by the cell types expressing
the basis for the tumor-associated changes in           them, as those (1) expressed on only certain types
glycolipids by the finding of a remarkable accu-         of normal cells (often only in certain develop-
mulation of fucose-containing glycolipids found         mental stages) and greatly accumulated in tumor
in human adenoarcinomas, some of which were             cells; (2) expressed only on tumor cells, for ex-
identified as lactofucopentaose-III-ceramide,            ample, altered blood group antigens or mucins;
lactofucopentaose-II-ceramide (Lewis A blood            and (3) expressed commonly on normal cells but
group glycolipid), and lactodifucohexose and la-        present in much higher concentrations on tumor
ctodifucooctose ceramide (Lewis B glycolipid).37        cells, for example, the GM ganglioside in mela-
These identifications were confirmed once mono-           noma and Lex in gastrointestinal cancer.37
clonal antibodies (mAbs) were used to iden-                A variety of chemical changes that can explain
tify antigens definitively. A number of mAbs with        the altered glycosylation patterns in tumor cells
preferential reactivity for tumor cells over nor-       have been identified. These result from three
124                                                                                     CANCER BIOLOGY

kinds of altered processes: (1) incomplete syn-          glycolipids observed in cancer cells. Increased
thesis and/or processing of normally existing            branching of asparagine-linked oligosaccharides
carbohydrate chains and accumulation of the              and incomplete processing of these oligosaccha-
resulting precursor form; (2) ‘‘neosynthesis’’ re-       rides have also been noted in certain cell-surface
sulting from activation of glycosyltransferases          as well as secretory glycoproteins.44,45 The in-
that are absent or have low activity in normal           creased activity of specific N-acetylglucosaminyl
cells; and (3) organizational rearrangement of           transferases in tumor cells appears to be responsi-
tumor cell membrane glycolipids.41                       ble for the appearance of tri- and tetra-antennary
   Moreover, the glycosyl epitopes found in              structures, whereas the analogous glycoprotein in
glycolipids and glycoproteins make up micro-             normal cells is often a biantennary structure. Un-
domains that are involved in cell adhesion and           usually high expression of N-acetylglucosaminyl-
signal transduction events. They function as a           transferase-IVa has been observed in human
‘‘glycosynapse’’ (analogous to the ‘‘immunological       choriocarcinoma cell lines and may be the enzy-
synapse’’) in mediating these events.42 The cell         matic basis for the formation of abnormal bian-
motility, altered adhesive properties, and inva-         tennary sugar chains on human chorionic go-
siveness observed in cancer cells are regulated by       nadotropin (hCG) produced by these cells.46
these glycosynapse complexes.42                          Similarly, the extra fucosylations that appear on
   Interest in the carbohydrate components of            membrane glycoproteins and glycolipids have
cell surface glycolipids, glycoproteins, and pro-        been associated with the induction of an unusual
teoglycans has been heightened by the fact that          a-fucosyltransferase in chemical carcinogen–
many of the monoclonal antibodies developed to           induced precancerous rat liver and in the result-
tumor cell–associated antigens recognize these           ing hepatomas.47 These investigations strongly
carbohydrate moieties or peptide epitopes ex-            suggest that the regulation of glycosyltransferase
posed by altered glycosylation. Moreover, many of        genes is important in malignant transformation.
these have turned out to be blood group–specific          Other changes in glycosyl transferase activities
antigens or modifications of blood group–specific          include a decrease in b1, 3-galactosyl transferase
antigens, some of which are antigens seen at cer-        b3 Gal-T5 in human adenocarcinomas compared
tain stages of embryonic development and thus fit         to normal colon.48
the definition of oncodevelopmental antigens.                All these data strongly support the idea that
Thus, the field of chemical glycobiology is making        glycosylation patterns change during transfor-
significant contributions to our understanding of         mation of normal cells into malignant ones. Be-
the cell surface biochemistry of normal and ma-          cause cell–cell interactions, adhesion to extra-
lignant cells. The aberrantly expressed glycans on       cellular matrices, regulation of cell proliferation,
tumor cells regulate a number of aspects of tumor        and recognition by the host’s immune system are
progression, including cell proliferation, tumor         all profoundly affected by the composition of the
invasion, angiogenesis, and metastasis.43 Some           cell surface, the entire social behavior of a cell
of these altered glycans are detectable in the           could be altered by such changes.
bloodstream and can be used as biomarkers of                Additional evidence for the importance of
tumor burden and response to therapy.43 In ad-           glycosylation patterns of cell surface glycopro-
dition, there are some glycan-targeted anticancer        teins and glycolipids in the malignant phenotype
therapeutics being tested in clinical trials. There is   comes from the use of glycosylation inhibitors
some evidence that tumor glycosphingolipids              and oligosaccharide-processing inhibitors. For
shed into the bloodstream may impair host im-            example, tunicamycin, an inhibitor of addition of
munity to some tumors.43                                 N-linked glycans to nascent polypeptide chains,
                                                         castanospermine, an inhibitor of glucosidase,
                                                         and KI-8110, an inhibitor of sialyltransferase
      Role of Glycosyl Transferases and
                                                         activity, all reduce the number of lung metastases
      Oligosaccharide Processing Enzymes
                                                         in murine experimental tumor models.49–51 In
The substitution of additional carbohydrate moi-         addition, swainsonine, an inhibitor of mannosi-
eties on blood group–related structures is not the       dase II, was shown to reduce the rate of growth of
only aberrant modification of glycoproteins or            human melanoma xenografts in athymic nude
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                              125

mice,52 and castanospermine was observed to in-         processed mucins of cancer cells, may have
hibit the growth of v-fms oncogene-transformed          some utility as tumor vaccines. Clinical trials of
rat cells in vivo.53 These results support the hy-      mucin-derived vaccines have been initiated.57,58
pothesis that the synthesis of highly branched          Some mucin antigens are shed from tumor cells
complex-type oligosaccharides are associated with       and can be detected in the sera of patients with
the malignant phenotype and may provide tumor           pancreatic, ovarian, breast, and colon cancers.
cells with a growth advantage.                          These include CA19-9, CA125, CA15-3, SPan-1,
                                                        and DuPan-2, which are currently being used as
                                                        tumor markers.58
                                                           The membrane-associated mucins of tumor
Mucins are a type of highly glycosylated glyco-         cells have multiple roles in cancer biology.58 They
proteins that a variety of secretory epithelial cells   interact with and modify the microenvironment
produce. They are 50% to 80% carbohydrate by            in which tumors grow. They provide an inter-
weight and function to lubricate and protect duc-       mediate signaling pathway by exposing, through
tal epithelial cells. They contain O-linked glycans     conformational changes, active growth factor
(serine- and threonine-linked) of various lengths       domains and recruitment of signal-transducing
and structures, depending on the tissue type in         molecules such as epidermal growth factor.
which they are produced. They are made in a wide        Muc1 (and Muc4) play a role in regulation of
variety of tissues, including the gastrointestinal      cancer cell proliferation and differentiation.
tract, lung, breast, pancreas, and ovary. Tumors        There is some evidence that the overexpression
arising in these organs may have an altered gly-        of Muc1 on tumor cells decreases cell–cell and
cosylationpatternthatdistinguishes them fromthe         cell–substratum interactions and that may pro-
normal mucins and renders them immunogenic.             mote tumor invasion and metastasis. Finally,
   Total expression of the mucins is increased in       there is evidence that tumor-derived mucins can
many cancers and up-regulated in some normal            modulate the immune response and may be im-
tissues under different physiologic states (e.g.,       munosuppressive for T-lymphocyte responses.
lactating mammary gland).54 Increased expres-
sion of the mucin 1 gene (muc1) has been ob-
served in most adenocarcinomas of the breast,
lung, stomach, pancreas, prostate, and ovary.           The proteoglycans are high-molecular-weight
Although muc1-encoded mucin has been the                glycoproteins that have a protein core to which
most extensively studied, cancer-related alter-         are covalently attached large numbers of side
ations in other mucins have been observed.              chains of sulfated glycosaminoglycans as well as
Moreover, it appears that some cells, both nor-         N-linked and /or O-linked oligosaccharides. They
mal and cancer, can express more than one mucin.        are categorized on the basis of their glycos-
Focal aberrant expression of muc2 and muc3 has          aminoglycans into several types, including
been frequently observed in a variety of adeno-         heparan sulfate, chondroitin sulfate, dermatan
carcinomas.55 However, in general, mucin genes          sulfate, and keratan sulfate.59 The glycosami-
appear to be independently regulated and their          noglycans have different repetitive disaccha-
expression is organ and cell type specific.55            ride units bound to the core protein through a
   There is evidence for host immune recognition        common glycosaminoglycan linkage region:
of the breast cancer mucin, in that cytotoxic T         G1cNAcb1?3Ga1b1?3Ga1b1?4Xy1b1-O-Ser.
lymphocytes isolated from breast cancer patients        The structure of the sulfated glycopeptides from
recognize a mucin epitope expressed on the              the carbohydrate-protein linkage region of some
breast cancer cells.56 The immune-recognized            of the proteoglycans has been determined.60
epitope involves the core protein that appears to          Proteoglycans interact via their multiple bind-
be selectively exposed on breast, ovarian, and          ing domains with many other structural macro-
other carcinomas. It has also been demonstrated         molecules, giving them the capacity ‘‘to function
that patients can produce antibodies to cancer          as a multipurpose ‘glue’ in cellular interactions.’’61
mucins,54 and this is the basis for the proposal        They bind together extracellular matrix (ECM)
that glycopeptides, because of the aberrantly           components, such as hyaluronic acid, collagen,
126                                                                                    CANCER BIOLOGY

laminin, and fibronectin; mediate binding of cells      the free passage of cells across them, but there
to the ECM; act as a reservoir for growth factors;     are mechanisms that permit the passage of in-
and ‘‘present’’ growth factors to growth factor        flammatory cells. It is also clear that basement
receptors on cells. The proteoglycans also act as      membranes act as regulators of cell attachment,
cell adhesion factors by promoting organization        through cellular receptors called integrins (see
of actin filaments in the cell’s cytoskeleton. Pro-     below). There is also ‘‘cross talk’’ between epi-
teoglycans have been shown to undergo both             thelial cells and their ECM to create a micro-
quantitative and qualitative changes during ma-        environment for accurate signal transduction for
lignant transformation, and alterations have           growth factors and other regulatory molecules. It
been reported in breast, colon, and liver carci-       has been shown, for example, that exogenous
nomas, in glioma cells, and in transformed mu-         reconstituted basement membranes stimulate
rine mammary cells and 3T3 fibroblasts.                 specific differentiation of a variety of cell types,
   Two putative tumor suppressor genes are             including mammary cells, hepatocytes, endo-
glycosyl transferases required for the biosyn-         thelial cells, lung alveolar cells, uterine epithelial
thesis of the proteoglycan heparan sulfate.62          cells, Sertoli cells, and Schwann cells.64
Mutations of these genes, called ext1 and ext2,           The basement membrane barrier can be
have been associated with the development of           breeched by tumor cells that release a variety of
skeletal dysplasias, and these findings suggest         proteases, glycosidases, and collagenases that
that alterations in the synthesis of heparan sul-      have the ability to degrade various components
fate precursor polysaccharide are involved in          of the matrix and thus allow tumor cells to in-
dysregulation of heparan sulfate production and        vade through tissue barriers and blood vessel
function in tumor formation.                           and lymph channel walls. In addition, malignant
                                                       cells themselves have receptors for and /or can
                                                       produce certain components of the matrix; this
Modification of Extracellular
                                                       capability enables them to bind to the vascular
Matrix Components
                                                       endothelium and may be involved in their ability
The ECM plays a key role in regulating cellular        to metastasize. Tumor cells may also release
proliferation and differentiation. In the case of      polypeptide factors that can modulate the type
tumors, it is now clear that development of a          of proteoglycans produced by host mesenchy-
blood supply and interaction with the mesen-           mal cells. The tumor stromal cells, in turn, can
chymal stroma on which tumor cells grow are            release factors that favor tumor cell proliferation
involved in their growth, invasive properties,         and invasiveness. For example, activated fibro-
and metastatic potential. This supporting stro-        blasts in the tumor stroma release a number of
mal structure is continuously remodeled by the         growth factors that stimulate cell proliferation,
interaction between the growing tumor and host         inhibit apoptosis, and alter cell differentiation
mesenchymal cells and vasculature. About 80%           and that up-regulate proteases involved in de-
of the cells within a tumor are stromal cells, in-     grading the ECM (reviewed in References
cluding fibroblasts, non-tumor epithelial cells,        65 and 66). These factors include hepatocyte
mast cells, and macrophages.                           growth factor (HGF), insulin-like growth factors
   The ECM components include collagen, pro-           (IGF)-1 and -2, EGF, TGF-a, TGF-b, inter-
teoglycans, and glycoproteins, such as fibronec-        leukin-6, fibroblast growth factors (FGF)-2 and
tin, laminin, and entactin. The ECM forms the          -10, and matrix metalloproteases-1 and -7.65
milieu in which tumor cells proliferate and pro-       These multiple effects of the tumor stroma on
vides a partial barrier to their growth. Basement      cancer growth and progression provide a num-
membranes are a specialized type of ECM.               ber of potential targets for anticancer therapy.67
These membranes serve as a support structure
for cells, act as a ‘‘sieving’’ mechanism for trans-
                                                       Cell-Extracellular Matrix
port of nutrients, cellular metabolic products,
                                                       and Cell-Cell Adhesion
and migratory cells (e.g., lymphocytes), and play
a regulatory role in cell proliferation and differ-    Cells in tissues are attached to one another and
entiation.63 Basement membranes also prevent           to the ECM. Disruption of these adhesion events
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                          127

leads to increased cell motility and potential       dependent transmembrane glycoproteins that
invasiveness of cells through the ECM. In ad-        play an essential role in the initiation and sta-
dition, most cell types require attachment to the    bilization of cell-cell contacts. Regulation of
ECM for normal growth, differentiation, and          cadherin-mediated cell–cell adhesion is impor-
function. This attachment is responsible for         tant in embryonic development and maintenance
what is termed anchorage dependence. Normal          of normal tissue differentiation.71,72
cells that are detached from their binding to           The extracellular domain of various cadherins
the ECM undergo apoptosis, whereas tumor             is responsible for cell–cell homotypic binding (a
cells that are less dependent on this attach-        given cadherin domain for a given cell type), and
ment are free to proliferate, wander, and invade     the conserved cytoplasmic domains interact with
tissues.                                             cytoplasmic proteins called catenins. Each cad-
   Cell adhesion to the ECM is mediated by cell      herin molecule can bind to either b-catenin or g-
surface receptors called integrins. Integrins are    catenin, which in turn bind a-catenin. a-Catenin
a family of proteins consisting of ab heterodi-      links the cadherin complex to the actin cyto-
mers that are integral membrane proteins with a      skeleton. Cell lines that lack a-catenin lose
specific arginine, glycine, aspartic acid (RGD)       normal cell-cell adhesiveness, and tumor cells
amino acid sequence involved in binding to the       with mutated or down-regulated a-catenin have
ECM.68 Integrins also link the external ECM          increased invasiveness.73
cytoskeleton to the intracellular actin cytoskel-       E-cadherin is the predominant type of cad-
eton, and via this connection a linkage to control   herin expressed in epithelial tissue. Alterations of
of gene expression in the cell nucleus is es-        E-cadherin expression and function have been
tablished. In this way, cell-ECM interactions        observed in human cancers.74 In addition, down-
can control gene read-out involved in cell dif-      regulation of E-cadherin correlates with in-
ferentiation and function. Cell–ECM interac-         creased invasiveness, metastasis, and poor prog-
tions occur via focal adhesions that consist of      nosis in cancer patients. Suppression of this
clusters of ECM-bound integrins, and these in        invasive phenotype can be achieved by transfec-
turn connect to actin fibrils and the signal          tion of E-cadherin cDNA into carcinoma cells,
transduction machinery inside the cell. These        and contrarily, invasiveness of E-cadherin gene-
signaling pathways include the focal adhe-           transfected cells can be restored by exposure
sion kinase (FAK) pathway that participates          of the cells to E-cadherin antibodies or an E-
in the control of anchorage dependence, and          cadherin antisense RNA.74 Germline mutations
growth factor signaling pathways, such as the        of the E-cadherin gene (cdh1) have been found in
ras-raf-mitogen-activated kinase, protein ki-        New Zealand Maori families with a dominantly
nase C, and phosphatidylinositol 3-kinase path-      inherited susceptibility to gastric cancer.75
ways.69 Thus, integrins cooperate with growth           The cell surface receptor for E-cadherin is
factors to enhance mitogenic signaling. Alter-       b-catenin. Early mutations in the human colon
ations in integrin receptor expression have          cancer progression pathway affect the cellular
been observed in chemically transformed hu-          distribution of b-catenin. In patients with colon
man cells and in human colon and breast cancer       cancer, the normal colonic epithelial cells adja-
tissue.70                                            cent to neoplastic lesions had mostly cell surface
   Cell-cell interactions are also important for     membrane expression of b-catenin, whereas cy-
the normal regulation of cell proliferation and      toplasmic expression of b-catenin was observed
differentiation. These interactions are mediated     in aberrant crypt foci.76 Nuclear expression was
by a family of molecules called cell adhesion        observed in more advanced dysplasias and in-
molecules (CAMs), which act as both receptors        creased as adenomas progressed to carcinomas.
(on one cell) and ligands (for another cell). The    These latter changes are also observed in less
expression of CAMs is programmed during de-          well-differentiated areas of tumors and are ac-
velopment to provide positional and migratory        companied by loss of E-cadherin expression at
information for cells. A large family of CAMs        the invasive front of breast carcinomas, possi-
has been identified. One group of these, called       bly due to hypermethylation of the E-cadherin
cadherins, comprise a superfamily of Ca2þ-           promoter.77
128                                                                                   CANCER BIOLOGY

CELL PROLIFERATION VERSUS                              to have undergone malignant transformation.
DIFFERENTIATION                                        Somehow the genes controlling cell proliferation
                                                       are locked in the ‘‘on’’ position when they should
A cancer develops from cells that are capable of       be in the ‘‘off’’ position, and the genes control-
dividing. All tissues in the body contain some         ling differentiation are either not expressed or are
cells that can divide and renew themselves. A          expressed only imperfectly. What we need to
subset of the cell population in any tissue can        know to understand carcinogenesis and to de-
differentiate into the functional cells of that tis-   velop ways of preventing or curing cancer, then,
sue. The normal process of cellular differentia-       is contained in the mechanisms of normal cellu-
tion ultimately leads to an adult, fully differenti-   lar differentiation. Only by understanding these
ated, ‘‘dead-end’’ cell that cannot, under ordinary    mechanisms can the manner in which cells are
circumstances, divide again. These fully differ-       altered during malignant transformation be as-
entiated cells are the workhorse cells in most         certained.
tissues of the body. They are the neurons in the          Differentiation is the sum of all the processes
brain controlling ideation and behavior; the liver     by which cells in a developing organism achieve
cells that manufacture enzymes to metabolize           their specific set of structural and functional
substrates needed for growth, produce plasma           characteristics. By the acquisition of these spe-
proteins, and clear the blood of potentially toxic     cial traits, progeny cells are distinguishable from
substances; the pancreatic cells that manufacture      their parent cells and from each other. Somatic
insulin and the enzymes necessary for digestion;       cells that share a set or a subset of structural and
the kidney cells that filter, secrete, or reabsorb      functional characteristics become organized into
substances and fluid in the formation of urine;         tissues in higher organisms. Tissues are arranged
the polymorphonuclear white blood cells that           as organs, and organs make up the organism.
phagocytize and destroy bacteria; and so on.           Indeed, cellular differentiation is the sine qua
Under circumstances that are not clearly under-        non of multicellular life.
stood, cells that have the potential to divide can        Differentiation requires a progressive restric-
be changed by interaction with carcinogenic            tion of genomic expression in the pathway from
agents into a cell type that is capable of continued   the totipotential fertilized ovum to the unipotent
proliferation and thereby is prevented from            cells of specialized organs (Fig. 4–1).78 The to-
achieving the normal state of complete differ-         tipotentiality of cells starts to change very early
entiation. The carcinogen-altered cell is said         after fertilization as the developing embryo

               Figure 4–1. Schematic model of the various stages of differentiation in
               multicellular organisms. (Adapted from Rutter et al.,78 with permission.)
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                            129

proceeds through the blastula stage, where it           phibians, for example, the cell cycle prior to the
is already evident that certain cells are pre-          blastula stage is abbreviated; there is no inter-
destined to form certain tissues. This process          mitotic G1 phase and G2 is short. During blas-
appears to occur by means of a combination of           tulation, the G2 period is prolonged, and a short
internal reprogramming (which perhaps occurs            G1 appears. Thereafter, the G1 period between S
as early as the first cell division) and external        (DNA replication) phases lengthens until the
stimuli by neighboring cells, proximity to the          tail-budding stage is reached, at which time the
maternal circulation, and gradients of growth           cell cycle approaches that of adult proliferating
factors, oxygen, nutrients, ions, and so on that        cells. The cell cytoplasm appears to exert control
result from the position in which a cell finds           over the timing of nuclear DNA synthesis since
itself as the early embryonic cells continue to         nuclei from adult tissues in which DNA syn-
divide. The whole microenvironment of the cell          thesis is rare can be induced to synthesize DNA
determines its developmental destiny. The               by injection into unfertilized ova.
process of differentiation appears to be fairly            In mammalian cells, also, growth arrest is
permanent in that, as tisues develop, some cells        coordinated with expression of the differenti-
retain the capacity to divide, whereas others           ated phenotype—for example, in hematopoi-
divide and then differentiate into cells with a         etic cells and in epithelial cells of the skin and
more restricted phenotype. These latter cells           gastrointestinal tract as well as in such cell cul-
are then said to be pluripotent rather than             ture systems as the preadipocyte mouse 3T3
totipotent, that is, they are now committed to          lines that can be induced to differentiate into fat
develop into one of the cell types peculiar to          cells. Cells transformed by carcinogenic agents
their tissue of origin. For example, a pluripotent      or oncogenic viruses lose this ability to become
stem cell of the bone marrow may differenti-            growth arrested and to become terminally dif-
ate into an erythrocyte, a polymorphonuclear            ferentiated.
leukocyte, or a megakaryocyte. It used to be
thought that only a few proliferative tissues in
                                                        Mechanisms of Cellular Differentation
the body contained stem cells that had unlim-
ited proliferative capacity, e.g., the bone marrow      Much of what we have learned about the cel-
hematopoietic stem cells, basal cells of the skin       lular mechanism of differentiation has come
and crypt cells of the gastrointestinal tract. It is    from studies of lower organisms, including
now known that, in fact, all tissues in the body,       yeast, slime molds, round worms, sea urchins,
including the central nervous system, have stem         fruitflies, zebrafish, and chickens. As more has
cells that can re-enter the cell proliferation cycle    become known about the genomes of these or-
if a tissue is damaged. These tissue stem cells are     ganisms and of humans, we have also learned
probably the ones most susceptible to carcino-          that there are orthologous genes and proteins
genic stimuli.                                          that have similar sequence and function and that
   An organism could not develop, of course,            can be traced back in evolution. A number of
without vigorous cell replication. Nor could it         these orthologs are involved in cell differentia-
survive without continued cell division. How-           tion processes in all species. As an example, 50%
ever, there is normally a well-controlled balance       of the genes of fruitflies have human equivalents
between cell division, cell differentiation, and cell   and almost every human gene has a counterpart
death. It is this delicate balance that is disrupted    in the mouse.
in cancer tissue. Cell differentiation usually pro-        The nematode (round worm) Caenorhabditis
duces, ultimately, a cell that no longer has the        elegans (C. elegans) was the second eukaryotic
capacity to divide, but many cells in the process       organism that had its genome completely se-
of differentiating continue to divide. Hence, the       quenced; the first was the yeast Saccharomyces
two processes are not mutually exclusive.               cerevisiae (S. cerevisiae). This sequencing al-
   The period following fertilization up to late        lowed for the first time a direct comparison of
blastula is a period of intense cell division with      orthologous genes from widely divergent organ-
very little cell growth between mitoses. In am-         isms and provided background information on
130                                                                                       CANCER BIOLOGY

               Figure 4–2. Distribution of core biological functions conserved in both yeast
               and worm. Yeast and worm protein sequences were clustered into closely re-
               lated groups (BLASTP P<1 Â10–50, with the >80% aligned length constraint).
               Each sequence group (including groups with two or more sequences) was
               assigned into a single functional category, relying primarily on the functional
               annotations for the yeast genes in Saccharomyces Genome Database (SGD)
               when available. The unclassified category contains groups of sequences with-
               out annotation. The boxed number within each category reflects the ratio of
               worm to yeast proteins for that category. (From Chervitz et al.,79 reprinted with
               permission from the American Association for the Advancement of Science.)

how similar genes in humans may function.79              to two yeast ras genes (ras1 and ras2). The cell
These shared genes carry out fundamental bio-            cycle checkpoint genes cdc 28 in yeast and ncc-1
logical processes such as intermediary metabo-           in C. elegans are orthologous pairs of the cyclin-
lism; DNA and RNA synthesis and processing;              dependent kinase family that are functionally
and protein folding, trafficking, and degradation.        interchangeable. Mammalian cells have similar
The percent conservation of these functions              cell cycle checkpoint control proteins.
between yeast and nematode is shown in Figure               A Hedgehog-related protein in yeast (HINT)
4–2 (see color insert). There has, of course, been       has binding domains similar to nematode
some sequence and functional divergence dur-             Hedgehog, although the yeast gene has a dif-
ing evolution, yet a surprising number of similar        ferent function. Hedgehog is a key regulator of
gene sequences related to similar functions have         positional orientation in insect and vertebrate
been passed down to humans. A few examples               development (see below). The DNA binding
will demonstrate this.                                   domains of these yeast and worm proteins are
   Functional conservation of proteins from              specifically related to the helix-turn-helix do-
widely divergent species was first shown exper-           mains of the transposases of animal and bacte-
imentally by demonstrating that the mammalian            rial transposons, which provides an idea of how
RAS protein could substitute for yeast RAS in a          evolutionarily old these protein domains are.
RAS-deficient yeast strain (reviewed in Refer-               Thus, the comparison of the complete protein
ence 79). The C. elegans ras gene is homologous          sets of organisms like C. elegans and S. cerevisiae,
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                          131

based on correlation of sequence and function        vergent species. In addition, this network of
data, provides a powerful tool for understanding     genes contained many cell proliferation and cell
the functional role of their orthologous genes in    cycle genesacrossmultiplespecies.Thesedatanot
human cells.                                         only help to define a previously unknown function
   If we start with the lowest level and work our    of genes but also help to define what other genes
way up to mammals and humans, we can see             are linked to a functional network. This type of
what this comparitive study can teach us about       study will be a great tool to help define upstream
how differentiation and development can go           and downstream targets for therapeutic inter-
awry along the way. In addition, some interest-      vention against malignant disease.
ing targets for drug discovery may come to light.
Even though the genome sequences of humans
                                                          Slime Molds
and several lower organisms have established a
fairly detailed list of genes required for key       In 1942, Ernest Runyon discovered that if he put
functions, an important next step is to deter-       samples of the slime mold Dictyostelium dis-
mine the functions of the large pool of genes for    coidium on both sides of a piece of porous cel-
which no function is currently known and to          lophane, the individual cells began to move ‘‘tail
learn about how genes are networked to carry         to head’’ in streams toward a central point (re-
out specific biological processes. Gene expres-       viewed in Reference 81). Remarkably, the cells
sion data from DNA microarrays gives clues           on the other side of the cellophane lined up and
about this, but a more powerful tool is to de-       moved in the same way, forming mirror-image
termine how genes are co-regulated and whe-          streams. Runyon hypothesized that a diffusible
ther this co-regulation has been conserved over      molecule penetrated the cellophane and at-
evolution.                                           tracted the cells on the other side. This was the
   The availability of genomic sequence and          first example of a ‘‘chemoattractant’’ that could
DNA microarray data for humans and organisms         regulate cell movement and migration. This
such as yeast, C. elegans, and Drosophila allows     chemoattractant turned out to be cyclic AMP.81
determination of evolutionarily conserved ex-        The tail-to-head orientation is achieved by the
pression over a large phylogenetic scale. Stu-       distribution of adenylyl cyclase that produces
art et al.80 have identified pairs of genes co-       cAMP at the rear of the migrating cells, thus
expressed over 3182 DNA microarrays from             explaining how the cells line up—i.e., cAMP
humans, Drosophila, C. elegans, and S. cerevi-       is produced at one end of the migrating cells,
siae. They found 22,163 co-expression relation-      is secreted into the environment, and binds to
ships, each of which have been conserved across      cAMP receptors in the trailing cells. This process
these species. This finding implies that there is a   in turn leads to the accumulation of factors in-
selective advantage for these gene sets and that     volved in cellular movement in the membrane at
these genes are functionally related. One inter-     the front edge of the trailing cells. The induced
esting gene set is comprised of five metagenes        ability ‘‘to march in close-order files’’ also facili-
involved in cell proliferation and cell cycle reg-   tates the laying down of an extracellular matrix
ulation that were not previously known to be         over which the cells can move smoothly. Cellular
so involved.80 A metagene is defined as a set of      movement also requires an intact internal cel-
genes across multiple organisms whose protein        lular skeleton of actin fibers. Mutations in ade-
sequences reflect the best reciprocal connection,     nylyl cyclase prevent this smooth concatenation
based on a function. Expression levels of these      of events. Thus, even this primitive organism
genes were measured in human pancreatic              processes properties similar to certain human
cancer cells and normal pancreas and in loss-of-     developmental and functional properties—cell
function mutants of C. elegans. All five gene sets    movement, response to chemoattractants, de-
were overexpressed in pancreatic cancers com-        position and modulation of an extracellular
pared to normal tissue. In C. elegans loss-of-       matrix, and signaling through an internal cyto-
function mutants, germline cell proliferation was    skeleton.
also suppressed, providing evidence that these          Because cAMP appears to play a key role in a
genes are functionally linked in two widely di-      large number of cellular processes (Table 4 –1)
132                                                                                                         CANCER BIOLOGY

          Table 4–1. Effects of cAMP on Various Cellular Processes
                                                                                               Change in Activity
          Enzyme or Process Affected                       Tissues or Organism                 or Rate*
          Protein kinase{                                  Several                             þ
          Phosphorylase                                    Several                             þ
          Glycogen synthetase                              Several                             À
          Phosphofructokinase                              Liver fluke                          þ
          Lipolysis                                        Adipose                             þ
          Clearing factor lipase                           Adipose                             À
          Amino acid uptake                                Adipose                             À
          Amino acid uptake                                Liver and uterus                    þ
          Synthesis of several enzymes                     Liver                               þ
          Net protein synthesis                            Liver                               À
          Gluconeogenesis                                  Liver                               þ
          Ketogenesis                                      Liver                               þ
          Steroidogenesis                                  Several                             þ
          Water permeability                               Epithelial                          þ
          Ion permeability                                 Epithelial                          þ
          Calcium resorption                               Bone                                þ
          Renin production                                 Kidney                              þ
          Discharge frequency                              Cerebellar Purkinje                 À
          Membrane potential                               Smooth muscle                       þ
          Tension                                          Smooth muscle                       À
          Contractility                                    Cardiac muscle                      þ
          HCI secretion                                    Gastric mucosa                      þ
          Fluid secretion                                  Insect salivary glands              þ
          Amylase release                                  Parotid glad                        þ
          Insulin release                                  Pancreas                            þ
          Thyroid hormone release                          Thyroid                             þ
          Calcitonin release                               Thyroid                             þ
          Research of other hormones                       Anterior pituitary                  þ
          Histamine release                                Mast cells                          À
          Melanin granule dispersion                       Melanocytes                         þ
          Aggregation                                      Platelets                           À
          Aggregation                                      Cellular slime molds                þ
          Messenger RNA synthesis                          Bacteria                            þ
          Synthesis of several enzymes                     Bacteria                            þ
          Proliferation                                    Thymocytes                          þ
          Cell growth                                      Tumor cells                         À
          *þ, increase; À, decrease.
           Stimulation of protein kinase is known to mediate the effects of cAMP on several systems, such as the glycogen
          synthetase and phosphorylase systems, and may be involved in many or even most of the other effects of cAMP.
          (From Sutherland82)

including cell proliferation and differentiation, a                and a decreased rate of cell proliferation.
number of investigators have speculated that                       However, not all transformed cells respond to
alterations in the cyclic nucleotide–generating                    cAMP treatment in this way. The response is
or response systems may be altered during ma-                      determined by the cell of origin of the trans-
lignant transformation. A considerable amount                      formed cell line. Fibroblastic cells, in general,
of work has been done to show this, and there                      tend to respond to the cAMP in the previously
does appear to be some correlation, but the                        described manner, whereas epithelial cells often
exact role of cyclic nucleotides in this process                   do not. Experiments with various clones of rat
has not yet been defined. A number of studies                       kidney cells, for example, showed that in a fibro-
have shown that transformed fibroblasts regain a                    blastic clone, intracellular levels of cAMP rose
number of the characteristics of untransformed                     as the cells reached confluency, but this was not
cells after treatment with cAMP analogues.                         the case in an epithelial clone.83 Murine sar-
These characteristics include a more flattened                      coma virus–transformed fibroblastic clones of
morphology, an increased adhesion to the sub-                      rat kidney cells did not have elevated intracel-
stratum, a decreased agglutinability by lectins,                   lular cAMP at confluency, and they responded
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                               133

to treatment with exogenous cAMP analogues            antibodies to bombesin inhibit tumor cell prolif-
by exhibiting a slower growth rate and a flat-         eration. Some pituitary, adrenal cortical, and ovar-
tened cell morphology. Neither the growth nor         ian tumors have point mutations in G proteins
the morphology of the epithelial clone was af-        coupled to adenylyl cyclase that could lead to
fected by exogenous cAMP.                             constitutive overproduction of cAMP.
   It appears that certain parts of the response         The a-1b-adrenergic receptor is a member of
system for cAMP differ in certain kinds of cells.     the G protein–coupled receptor superfamily
In support of this idea, experiments with the S49     and activates PI hydrolysis, a signaling pathway
lymphoma cell line showed that the proliferation      that is activated by a number of growth factors
of these cells was inhibited by cAMP but was          and that plays a crucial role in mitogenesis.
not inhibited in a mutant S49 cell line defective     Mutation of three amino acids residues in the
in cAMP-dependent protein kinase.84 Similar           third intracellular loop (see Fig. 4-3)88 increases
results have been obtained for mutant Chi-            the binding affinity of norepinephrine and its
nese hamster ovary cells with a variant cAMP-         ability to stimulate PI hydrolysis by two to three
dependent protein kinase. There may also be           orders of magnitude.89 Moreover, this activating
alterations in the cAMP-binding regulatory sub-       mutation renders the receptor constitutively
unit of cAMP-dependent protein kinase or in           active, stimulating PI turnover even in the ab-
the translocation step involved in the nuclear        sence of ligand. When the wild-type gene for the
uptake of protein kinase, a step that appears to      a1b receptor is transfected into rat or NIH 3T3
be required for response to cAMP in some              fibroblasts, the cells express high levels of this
cells. Decreased binding of cAMP and an altered       receptor, become transformed in response to
cAMP-binding protein have been demonstrated           norepinephrine, and form tumors when injected
in a cAMP-unresponsive line of Walker 256 car-        into nude mice. When the mutated gene is
cinosarcoma cells compared with the responsive        transfected into fibroblasts, the cells spontane-
parent line.85 The nuclear translocation of cAMP-     ously form transformed foci in the absence of
binding proteins and protein kinase was also          ligand and have an enhanced ability to form tu-
markedly diminished in the unresponsive tumor         mors in nude mice. Thus, the a1b adrenergic
cells after treatment with dibutyryl camp.86
   It is clear that cAMP affects the proliferation
rate of some normal and transformed cultured
cells and that cAMP levels are lower in some
transformed cell lines. It is not clear if changes
in cAMP lebels or the cAMP-response system
are responsible for the appearance of the trans-
formed phenotype and, more importantly, for
the loss of normal growth control. In some cells,
alterations in intracellular cAMP appear to be
more closely related to the morphologic char-
acteristics of the transformed phenotype than to
growth control; in fact, the two events are clearly
dissociable in certain cell types. Nevertheless, it   Figure 4–3. Schematic representation of the mem-
is clear that induction of cAMP in several types      brane organization of plasma membrane receptors
of cultured neoplastic cells induces a more dif-      (such as adrenergic receptors, substance K receptors,
                                                      or opsins) that are linked to G proteins. An extra-
ferentiated, less transformed phenotype.              cellular amino-terminal region with sites of glyco-
   Several lines of evidence implicate G protein–     sylation on asparagines residues is followed by seven
coupled receptors in malignant transformation.87      membrane-spanning domains (M1 to M7) inter-
Overexpression of acetylcholine or serotonin          spersed with three intercellular and three extracel-
receptors in NIH 3T3 cells causes ligand-             lular loops and then an intracellular carboxy terminus.
                                                      The consensus sequences expected at sites for phos-
dependent transformation. Bombesin-like pep-          phorylation are found in the third intracellular loop and
tides are secreted by some small-cell lung car-       carboxyl-terminal regions. (From Taylor and Insel,88
cinoma cells and stimulate their growth, and          with permission.)
134                                                                                CANCER BIOLOGY

receptor gene acts like a proto-oncogene and         partners.91 Moreover, the proteins in interacting
when activated or overexpressed is a transform-      sets presumably act in the same functional path-
ing oncogene. These data suggest that other G        ways in different organisms, and although not
protein–coupled receptors of this type can act as    clearly demonstrated across widely divergent
oncogenes in certain cell types. This further        species, these sets most likely have similar func-
suggests a host of strategies for chemothera-        tional duties in different organisms.
peutic interdiction of this system, for example,        Some of the striking correlates between yeast
the design of specific antagonists of the G           and higher organisms involve the cell cycle check-
protein–coupled receptors that may be activated      point pathways, which were originally demon-
or overexpressed in tumor cells.                     strated in yeast and now have been shown to be
   There is also evidence that alteration of G       analogous in mammalian, including human, cells
protein subunits themselves can cause alter-         (see Cell Cycle Regulation, below).
ations in fibroblast growth characteristics. For
example, transfection and overexpression of a
                                                          Sea Urchin
mutated G protein a1b subunit gene, a gene
shown to be involved in proliferation of fibro-       Development of a ‘‘body plan’’ for multicellular
blasts and differentiation of myeloid cells, in fi-   organisms requires a precise interaction of
broblasts produces increased cell proliferation      multiple factors and is controlled by what Eric
and anchorage-independent growth, indicating         Davidson and colleagues call a ‘‘gene regulatory
a role for this G-protein subunit in regulation of   network’’ (GRN).92 Expression of genes in this
fibroblast cell proliferation and in transforma-      complex network is in turn replicated by ‘‘waves’’
tion events.90                                       of transcription factors and repression that turn
                                                     genes on and off in an exquisitely orchestrated
                                                     pattern. This complex array of events is not easily
                                                     studied in complex animal systems such as mam-
Since the yeast genome has been sequenced, a         mals, but in simple organisms such as sea ur-
number of functional and evolutionary correla-       chins it can be analyzed in reasonable detail.
tions have become more evident. RNA interfer-           The ‘‘heart’’ of developmentally regulated
ence (RNAi) has been used to block gene expres-      gene networks is the genes encoding transcrip-
sion, thus allowing the ‘‘knock out’’ of specific     tion factors and cis-regulatory elements that
sequences and the determination of what the key      control the expression of these genes. Each of
functional defect is. Silencing of genes to detect   the cis-regulatory elements gets multiple inputs
function has also been used in yeast, C. elegans,    from transcription factors (TFs) encoded by
and mammalian cells (see Chapter 5).                 other genes in the network. These TFs in turn
   Recent advances in detecting protein-protein      recognize specific gene sequences in the cis-
interacting networks, in addition to complete        regulatory elements. These cis-regulatory sys-
genome sequencing, provide another tool to           tems at the center of a developmental GRN
look at evolutionary changes over time and spe-      receive carefully timed input information based
ciation. Fraser et al.91 have used this concept to   on the rise and fall of the TFs to which they
catalog 3541 interactions between 2445 yeast         respond. Another key element is the signal
proteins to estimate evolutionary rates for these    transduction machinery that produces the sig-
protein sets. Comparison of orthologous se-          nals for TF expression. This machinery is turned
quences between S. cerevisiae and C. elegans         on by chemostimulants in the local environ-
provided a method to compare differences and         ment, which are also present in kinetic waves of
hence delineate conservation of protein in-          production and release from cells in the local
teracting networks during evolution. Although        cellular milieu. What turns on the expression of
protein-protein interacting networks evolve          these chemoregulants at the appropriate time
more slowly than individual proteins, such in-       still remains to be completely understood. The
teracting proteins appear to co-evolve, in that      chemoregulants that specify developmentally
substitutions in one protein results in selective    regulated events must of necessity operate along
pressure for reciprocal changes in interacting       gradients in the cellular environment of an
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                                       135

embryo. Otherwise, all cells would be exposed                       because the most closely examined examples of a
to the same signals and no cell type specific                        cis-regulatory information-processing system is
differentiation could occur, given that all the                     the developmental expression of the endo 16
cells in a given organism carry the same genome.                    gene of the sea urchin.94 Endo 16 encodes a
    This differentiation is achieved by the release                 polyfunctional protein that is secreted into the
of cellular diffentiating signals, called morpho-                   lumen of the embryonic and larval midgut. It is
gens, by certain ‘‘pacemaker’’ cells in the local                   expressed first in progenitor cells of the endome-
embryonic environment, similar to the way that                      soderm, then throughout the gut, and finally only
cyclic AMP regulates cell orientation, move-                        in the midgut. Early- and late-expression phases
ment, and differentiation in slime molds as de-                     are controlled by two different regulatory gene
scribed above. Morphogens are defined as                             sequences, each several hundred base-pairs long.
‘‘signaling molecules that are produced in a re-                       Control of gene expression regulated by these
stricted region of a [developing] tissue and move                   sequences is subject to input from nine DNA-
away from their source to form a long-range                         sequence specific TFs as time-varying regulatory
concentration gradient.’’93 Cells then differen-                    signals. The end result of all this is the devel-
tiate in response to local morphogen concen-                        opment and differentiation of the primitive gut
tration depending on how far away they are                          in the sea urchin. The steps in this process are
from the morphogen source. Examples of mor-                         instructive about what goes on in the develop-
phogens are Wingless, Hedgehog, and Dpp,                            ment of higher organisms and is worth cataloging
which form gradients and elicit differentiation                     here (Table 4–2).
responses across distances during wing and                             Another key point about developmentally
leg development in Drosophila. Multiple target                      regulated GRNs is that they are subject to both
genes are induced depending on the local con-                       positive and negative regulation. The develop-
centration of these ligands and binding to their                    mental process is moved forward by intergenic
responsive (receptor-bearing) cells.                                feedback loops, both positive and negative, that
    The reason that sea urchin development is an                    lock in a given developmental state. These seldom
ideal system in which to study these programs is                    go backwards once stabilized (with the possible

Table 4–2. Phenomenological Aspects of Endomesoderm Specification in Sea Urchin Embryos: Developmental
1. Autonomouse cues of maternal origin
   Nuclearization of b-catenin in micromeres (by fourth cleavage) and veg2 cells (from sixth cleavage on)
   Exclusion of ectodermal transcription factors from vegetal-most cell nuclei
   Nuclearization of Otx factor in micromeres at fourth cleavage
2. Early micromere signal
   Micromere signal to veg2 (fourth through sixth cleavage) required for normal endomesodermal specification
3. Wnt8/Tcf loop
   Wnt8 ligand expressed throughout endomesodermal domain maintains and strengthens b-catenin/ Tcf input in these
   B-catenin/Tcf input required for endomesoderm specification
4. Late micromere signal
   Expression of Delta ligand in micromeres
   Activation of Notch signal transduction in veg2 descendants adjacent to micromeres that receive Delta signal
5. Skeletogenesis
   Skeletogenic functions expressed after ingression of skeletogenic cells in late blastula
6. Specification of veg2 mesoderm and endoderm
   Segregation of cell type precursors within vegetal plate complete by late blastula
   Mesoderm cells turn off endoderm genes, leaving endoderm genes expressed in peripheral veg2 cells
7. Specification of veg1 endoderm
   Wnt8 signal from veg2 to veg1 and activation of b-catenin nuclearization in abutting veg1 cells
8. Invagination of archenteron
   veg2 mesoderm carried inward at tip of archenteron on gastrulation
   Followed by roll-in of veg1 endoderm, contributing mainly hindgut
From Davidson et al.,94 reprinted with permission from the America Association for the Advancement of Science
136                                                                                    CANCER BIOLOGY

exception of stem cell transdifferentiation or         suggesting that the two proteins are functional
carcinogenic insult (see Stem Cells, below). This      homologs as well as sequence homologs. Mu-
feedback network for intergenic loops used to          tations in the human lkb1 gene cause Purtz-
stabilize a developmental phenotype is not pe-         Jeghers syndrome, which is characterized by the
culiar to sea urchins. It is also seen, for example,   formation of intestinal polyps and a high inci-
in the hox gene network that controls rhombo-          dence of adenocarcinomas, as well as a number
mere development in the mouse hindbrain (re-           of other epithelial cancers. Thus, lkb1 has been
viewed in Reference 92). Thus, mammals use             classified as a tumor suppressor gene, whose
similar developmentally regulated systems.             function may relate to a role in cell cycle regu-
   Once the cis-regulatory network is used to          lation or apoptosis. It is hypothesized that disrup-
achieve a partially complete differentiated state,     tion of cellular polarity of epithelial tissues leads
later epigenetic processes such as genetic im-         to polyp and tumor formation.96 This is another
printing, DNA methylation, and regulation of           example of how disruption of normal tissue
chromatin packaging contribute to further dif-         developmental processes can lead to abnormal
ferentiation and stabilization of a differentiated     cellular behavior and potentially to cancer.
tissue phenotype (see Chapter 5).
      Drosophila Melanogaster
                                                       Another morphogen gradient, one that involves
Regulation of cellular polarity and ‘‘position-        mRNA decay, has been found in mouse em-
ality’’ is key to determining the cell motility, po-   bryos.97 Head-to-tail patterning requires a strict
sitioning, and orientation of body axes (e.g.,         time-dependent coordination and involves fi-
anterior-posterior [A-P], head-to-tail). Genes         broblast growth factor (FGF) signaling. The in-
involved in this key developmental process are         vestigators showed that transcription of FGF-8
evolutionarily conserved to a large extent. Re-        mRNA was progressively degraded in the newly
cently, it was found that homologs of genes that       formed tissues, thus producing a gradient of
control polarity of hairs on the epidermal cells of    FGF-8 mRNA in the posterior part of the em-
Drosophila (fruit fly) wings also control polarized     bryo. The mRNA gradient correlates with a
cell motility that determines cellular movements       similar gradient of FGF-8 protein, which in turn
underlying body shape orientation in verte-            correlates with a gradient of phosphorylation of
brates.95 These movements (called morphogenic          Akt kinase, which is a downstream signal trans-
movements) determine the mediolateral and A-P          duction effector of FGF-8. This provides an-
orientation in the vertebrate embryo as well as in     other morphogenic mechanism, in addition to a
the fruit fly. These mechanisms regulate body           pacemaker cell type that produces a growth fac-
axis orientation in all chordate species examined,     tor such as FGF, transforming growth factor b
including sea squirts, teleost (bony) fish, amphi-      (TGF-b), or other morphogen, creating a diffu-
bians, birds, and mammals (reviewed in Refer-          sion gradient for a regulatory protein.
ence 95).
   Some of the genes involved in A-P axis for-
                                                            Pathways: Getting to Know
mation and epithelial polarity have been iden-
                                                            All the Players
tified. Two of these genes are serine/threonine
kinases that have sequence homology between            The process of early development is a compli-
C. elegans and Drosophila, and with the human          cated one, and there are some similarities and
tumor suppressor gene LKB1. The kinase do-             some differences among various multicellular
main of the Drosophila LKB1 gene has 66%               organisms. The biochemical signals and genes
amino acid identity to human LKB1.96 Inter-            involved, as noted above, show a lot of evolu-
estingly, both the fruit fly and human LKB1             tionary conservation. Various polypeptide growth
proteins have conserved prenylation and protein        factors have been shown to play a role in early
kinase A phosphorylation sites that are essential      morphogenesis.98 For example, in early Xenopus
to the in vivo function of both proteins, strongly     development, there are a series of inductive
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                          137

events that involve growth factors, whose ac-         which the genes involved in this late phase are
tions lead to differentiation of mesoderm at the      regulated is not totally clear, but it determines
interface between the animal and vegetal poles        what regulates final organ size and the relation-
of the embryo. This induction is most efficiently      ship of organ size to the overall size of the de-
achieved by a combination of members of the           veloping embryo.
FGF and TGF-b families of growth factors. In              The relationship of these processes to cancer
Xenopus, Drosophila, and developing chick limb        is intriguing. Alterations of these events occur in
buds the role for members of the FGF and TGF-         malignancy: a turn-on of genes leading to cell
b families of polypeptide growth factors in early     proliferation, an alteration in the balance of cell
development appears to be regulation of ex-           proliferation and apoptosis, and a lack of feed-
pression of hox genes.98–101 For example, growth      back controls to limit organ size. Thus, an un-
factors regulate expression of a hox gene called      derstanding of the regulation of these develop-
xhox3 in Xenopus that is required for A-P pat-        mental events should go a long way toward
terning. Similar observations have been made in       understanding what goes wrong in the biochem-
Drosophila. Since hox genes themselves code for       istry of the cancer cell.
transcriptional regulators that can turn genes            Some cellular developmental pathways have
on or off, some of which may code for growth          been closely linked to human cancers, e.g., the
factor–like substances, one can visualize a cas-      Hedgehog and Wnt pathways. Sonic hedgehog
cade of events in which a local concentration         (Shh) is the mammalian version of the Dro-
of growth factor turns on a hox gene, which, in       sophila morphogen and is known to mediate
turn, activates another growth factor that turns on   epithelial–mesenchymal interaction in lung de-
another hox gene in a responding cell. This pro-      velopment. Loss of Shh function results in severe
cess suggests a way that pattern formation could      lung defects as demonstrated by failure of bron-
be transmitted from one cell region to another.       chial tree branching. This information has led
   The activation of hox genes, however, does         to the hypothesis that signaling via the Shh-
not clearly explain how, for example, within a        mediated pathway is important in airway epi-
given mesodermal area, different mesodermal           thelium repair after toxic damage or carcinogenic
cell types arise because hox genes are expressed,     insult.103 Watkins et al.103 have shown that the
albeit perhaps at different times and levels,         human hedgehog (Hh) pathway is extensively
throughout the mesodermal layer. Thus, addi-          activated during repair of acute airway injury.
tional genes must be expressed in a carefully         Activation of this pathway was also seen in hu-
regulated way to lead to further ‘‘subspecial-        man small cell lung carcinoma (SCLC). It was
ization’’ or differentiation events. One well-        shown that these cancers maintain their malig-
studied example of this is the expression of genes    nant phenotype via Hh pathway activation, thus
involved in the muscle differentiation pathway,       suggesting that SCLC may develop by recapitu-
such as the myogenic genes myo D and myogenin.        lating an early Hh-mediated event in airway ep-
   Another example is limb bud formation,             ithelial differentiation.
studied in vertebrates.102 This occurs in several         Members of the Hh family of molecules are
stages. The first phase involves the establishment     secreted during early development and have es-
of signaling centers within the bud primordium.       sential roles in tissue patterning in organisms
These signaling centers have positional deter-        from Drosophila to humans. Although Hedgehog
minants in the embryo: anterior-posterior,            proteins are known to affect cell fate, they can
dorsal-ventral, and medial-lateral. The second        also stimulate cell proliferation. Mutations that
phase is usually associated with increased cell       aberrantly activate Hh signaling have been ob-
proliferation mediated by various mitogens such       served in basal cell carcinomas and other human
as members of the FGF and Sonic hedgehog              cancers.104 In Drosophila, Hh acts on somatic
family of gene products. Ultimately, limb bud         ovarian stem cells to stimulate their prolifera-
outgrowth ceases because of decreased release         tion.105 Moreover, high production and signaling
of mitogens and a balance between cell prolif-        of Hh produces an abnormal number of ovarian
eration and programmed cell death. The way in         stem cells. The authors of this study postulate
138                                                                                   CANCER BIOLOGY

that Hh is a stem cell proliferation factor and that    catenin activates a number of genes that control
its excessive signaling causes abnormal expansion       cell fate. As noted above in the section on cell–
of stem cell pools in human tissues, which may          ECM and cell–cell adhesion, dysregulation of the
be part of the aberrant signaling during the ma-        b-catenin pathway is observed in colon, breast,
lignant transformation process.                         and other human cancers.
   It has been shown that a wide range of di-              The Notch gene–regulated pathway also plays
gestive tract cancers, including carcinomas of          a central role in many developmental processes.
the esophagus, stomach, biliary tract, and pan-         Examples are peripheral neurogenesis in Dro-
creas (but not colon), have increased activity of       sophila, vulval development in C. elegans, and
the Hh pathway. Interestingly, cyclopamine, an          lymphoid development in mammals (reviewed
Hh pathway antagonist, suppresses cell prolif-          in Reference 108). In zebrafish, activation of the
eration in gastrointestinal (GI) tract cell lines       Notch signaling pathway facilitates heart regen-
and causes regression of GI tract cancer xeno-          eration.109 Zebrafish have a remarkable ability
grafts in vivo in athymic (nude) mice.106 In ad-        to regenerate various organs including heart and
dition, cell proliferation of digestive tract cancer    fins. These processes involve increased expres-
cells in culture is enchanced by endogenous             sion of the notch lb gene and subsequent ac-
expression of Hh ligands (as determined by RT-          tivation of homeolox genes msxB and msxC.
PCR) and blocked by a Hh-neutralizing mono-             Notch activation is a key link in the decision-
clonal antibody,106 a finding suggesting that the        making process for proliferation and differenti-
Hh pathway would be a good target for anti-             ation of stem cells in hematopoietic, neural,
cancer drug development.                                gastrointestinal, and skeletal muscle lineages of
   The Wnt pathway is another important de-             various organisms.
velopmental pathway involved in human can-                 In a subset of acute lymphoblastic leukemias,
cers. The Wnt family of proteins are involved           there is a chromosomal translocation involving
in cell–cell signaling and adhesion during many         the human notch 1 gene and the T-cell receptor
steps in animal development, including forma-           b locus. This rearrangement leads to constitu-
tion of the embryonic axes to end stage devel-          tive expression of a truncated notch allele, which
opment of organs such as the kidney. This is the        behaves like an oncogene in this situation (re-
case both for invertebrates and humans. There           viewed in Reference 108). Dysregulated Notch
are at least 19 genes encoding Wnt proteins in          signaling has now been implicated in the path-
humans and ten receptors for these proteins.            ogenesis of a broad range of human cancers.
   There are three cellular signaling cascades          Somewhat surprising is the observation that
activated by Wnt proteins that separately regu-         Notch acts as a tumor suppressor in skin carci-
late cell differentiation, cell polarity, and cell      nogenesis, where it interacts with the Hedgehog
adhesion (reviewed in Reference 107). These             and Wnt pathways. This demonstrates a princi-
three cascades are (1) regulation of b-catenin          ple that is common in cancer biology—i.e., that
intracellular location and function; (2) activation     the consequences of activation or modification
of a planar cell polarity pathway mediated via          of a signal transduction pathway are dependent
the disheveled gene, cdc 42, and Jnk signal             on cellular context.
transduction; and (3) activation of an increase in         There are, of course, a number of other tissue
intracellular calcium levels and protein kinase         development and cell differentiation pathways
C. All of these pathways are mediated by Wnt            that involve developmentally regulated genes
proteins interacting with G protein–coupled             and their encoded proteins. In fact, each differ-
receptors.                                              entiated tissue in an adult multicellular organism
   Details of the Wnt /b-catenin pathway came           uses a combination of complex signaling path-
from Drosophila, where this pathway functions           ways to achieve its final destination in the body
in patterning body segments and appendages.             and final functional state. Many of these path-
Activation of this pathway by a Wnt ligand sta-         ways are ones that are disrupted or re-activated
bilizes b-catenin and facilitates its transfer to the   during the carcinogenic process. The take-home
nucleus, where in combination with chromatin-           message here is that most of these pathways
associated high-mobility group proteins, b-             were originally discovered and characterized in
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                              139

lower organisms where the biology can be stud-              Treatment of cancer through induction of
ied in a much less complicated way.                      cellular differentiation is an attractive idea be-
   Other important parameters of morphogenesis           cause the therapy could be target-cell specific
include the ability of like cells to cluster together    and most likely be much less toxic then standard
and ‘‘talk to each other’’ and the ability of cells to   chemotherapeutic agents. The best example of
produce and interact with a specific tissue type          this is the treatment of acute promyelocytic leu-
ECM. Thus, the ability to regulate cell-cell and         kemia in patients with all-transretinoic acid. A
cell-ECM (cell-substratum) interactions is also          more recent example is induction of solid tumor
key to normal development and cellular differ-           differentiation by the peroxisome proliferator-
entiation. Two families of adhesion molecules are        activated receptor-g (PPAR-g) ligand troglitazone
involved: cell-cell adhesion molecules, or CAMs,         in patients with liposarcoma.112 PPAR-g is a nu-
and cell-substratum adhesion molecules, or               clear receptor that forms a heterodimeric complex
SAMs.110,111 CAMs produce cell-cell contact be-          with the retinoid X receptor (RXR). This complex
tween like-minded cells that foster their interac-       binds to specific recognition sequences on DNA
tions and cell sorting into homogeneous popu-            and, after binding ligands for either receptor,
lations. As noted above, CAMs, or cadherins as           enhances transcription differentiation–inducing
they are also called, are large transmembrane            genes, including those for the adipocyte-specific
proteins that interact through cytoplasmic con-          pathway. PPAR-g appears to act as a tumor sup-
nections called catenins that link cadherins to the      pressor in the prostate and thyroid gland, but not
cell cytoskeleton, thus providing an internal sig-       in the colon, where its actions are more com-
naling process for CAMs that are in contact with         plex.113 Nevertheless, agents that can exploit the
the extracellular environment. These interac-            proliferation-inhibiting effects of PPAR-g in can-
tions are capable of modulating formation of             cer tissue and have minimal metabolic side effects
actin cables in the cytoplasm, and, thus, of af-         may be good targets for drug discovery.
fecting cell migration and cell surface polarity.
   Thus, a number of key interactions among
                                                         Stem Cells
growth factors, hox genes, CAMs, SAMs, the
ECM, and specific genes involved in cell lineage–         There has been tremendous excitement, not
specific pathways occur during early develop-             without some controversy of an ethical, politi-
ment and early differentiation. Although mostly          cal, and scientific nature, about stem cells. Until
studied in lower organisms, all of these genes           recently, it had always been thought that stem
have homologous counterparts in mammalian,               cells, those self-renewing, pluripotent cells that
including human, cells.                                  exist in an embryo, were only present in highly
                                                         proliferative tissues such as the bone marrow.
                                                         Even in tissues with a high cellular turnover
     Stimulation of Cancer
                                                         rate, such as skin and gastrointestinal mucosa or
     Cell Differentiation
                                                         in tissues such as liver that can regenerate, these
There are a number of examples of animal ma-             cells were thought to be limited in their ability to
lignant tumors or human cancer cells in culture          generate cells of different lineages with various
that can be induced to lose their malignant phe-         differentiated phenotypes. In some tissues such
notype by treatment with certain differentiation-        as the brain and the heart, self-renewing stem
inducing agents. These include induction of dif-         cells were thought to be nonexistent. It has been
ferentiation of the Friend virus–induced murine          a true revolution in cell biology to find out that
erythroleukemia by dimethylsulfoxide (DMSO);             self-renewing, multipotent, and perhaps pluri-
differentiation of murine embryonal carcinoma            potent stem cells exist in every organ in the body
cells by exposure toretinoicacid, cAMPanalogues,         of mammals and most likely of humans. These
hexamethylbisacetamide, or sodium butyrate; and          cells presumably are called on to proliferate in
differentiation of human acute promyelocytic             response to tissue injury and are involved in
(HL-60) cells in culture by a number of antican-         tissue repair.
cer drugs, sodium butyrate, DMSO, vitamin D3,               A ‘‘true’’ stem cell must satisfy the following
phorbol esters, or retinoic acid analogues.111           criteria: (1) it must have unlimited self-renewal
140                                                                                          CANCER BIOLOGY

capacity; and (2) it must be able to divide into two   Table 4–3. Potential Plasticity of Stem Cells
types of daughter cells—one that replicates a          Location of
self-renewing phenotype and one that attains the       Stem Cell               Type of Cells Generated
phenotype of a subsequent differentiated state.        Brain                   Neurons, oligodendrites, skeletal
   Pluripotent stem cells can originate from the                               muscle, blood cells
inner cell mass of an embryonic blastocyst or          Bone marrow             Endothelial cells, blood cells,
from the fetal gondal ridge of 8- to 10-week                                   cartilage, bone, adipocytes,
                                                                               cardiac muscle, skeletal muscle,
embryos. Although such cells, known as embry-                                  neuronal cells, skin, oval cells,
onic stem (ES) cells and embryonic germ (EG)                                   gastrointestinal tract cells,
cells, respectively, have been known about in                                  thymus, pulmonary epithelial cells
mice for many years, it was only recently that ES      Skeletal muscle         Skeletal muscle, bone, cartilage, fat,
                                                                               smooth muscle
cells114 and EG cells115 were isolated from hu-
                                                       Myocardium              Myocytes, endothelial cells
man sources. Even more recently it has became
                                                       Skin                    Keratinocytes
known that most, and probably all, mammalian
                                                       Liver                   Liver cells
tissues also contain stem cells of amazing plas-
                                                       Testis and ovaries      Gonads
ticity. These not only proliferate and renew dam-
                                                       Pancreatid ducts        Islet cells
aged tissues but in some cases can even differ-
                                                       Fatty tissue            Fat, muscle, cartilage, bone
entiate into cell types of another tissue. ES cells
can be grown in cell culture and are capable of        From Rosenthal,116 reprinted with permission from the Massachu-
                                                       ssetts Medical Society
producing multiple cell types including vascular,
neuronal, pancreatic, and cardiac muscle cells         that can regenerate function in an animal model
(reviewed in Reference 116). EG cells have also        of Parkinson’s disease119 or can produce insulin-
been shown to produce multiple lineages, in-           secreting structures similar to pancreatic islets,120
cluding neurons, glial cells, vascular endothe-        it is much less expected that bone marrow–
lium, hematopoietic cells, cardiomyocytes, and         derived cells can do the same thing. Although
glucose-responsive, insulin-secreting cells.115 In     some reports indicate that hematopoietic stem
effect, ES and EG cells are capable of producing       cells (HSCs) from the bone marrow or cytokine
virtually any type of tissue, given the appropriate    mobilized peripheral HSCs are capable of multi-
culture conditions.                                    ple cell lineage differentiation, most likely bone
   The really astonishing thing is that some stem      marrow mesenchymal stem cells are the cell type
cells derived from adult tissues are also capable      that has this broad pluripotency. Nevertheless,
of generating other tissue cell types in addition to   cytokine mobilized HSCs from the bone marrow
their own (Table 4–3). Although the mechanism          have been shown to repair myocardial infarctions
for how this comes about is somewhat con-              in a mouse model,l21 a finding suggesting that
troversial (see below), there are extensive data       HSCs themselves can transdifferentiate into mul-
showing that it can happen. For example, it has        tiple cell types and could be a relatively avail-
been observed that humans receiving nonautol-          able source of cells for tissue regeneration. Bone
ogous bone marrow transplants have cells of            marrow–derived cells of stromal origin have been
donor origin in nonhematopoietic tissues such as       shown to initiate pancreatic regeneration122 and
the liver.117                                          to produce neurons.123
   The ability of bone marrow–derived stem cells          Jiang et al.124 reported that mesenchymal stem
and stem cells from other tissue sources to gen-       cells derived from the bone marrow of mice could
erate mature functional cell types has led to the      differentiate into cells with visceral mesoderm,
exciting possibility of tissue regeneration and re-    neuroectoderm, and endoderm characteristics
pair for a variety of human diseases or conditions,    in vitro. When single cells from this source were
including Parkinson’s disease, diabetes, myocar-       injected into early blastocysts, they contributed
dial infarction, congestive heart failure, chronic     to most somatic cell types in the developed an-
limb ischemia, liver cirrhosis, and chronic lung       imal. In addition, when these cells were trans-
disease.118 While it is perhaps less surprising,       planted into a non-irradiated host, they en-
given their greater plasticity that ES cells can       grafted and differentiated into hematopoietic
differentiate into dopamine-producing neurons          lineage cells and the epithelium of the liver,
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                            141

lung, and gut. Thus bone marrow mesenchymal            tissue stem cells. What cancer cells lose, how-
stem cells may be an ideal source for therapy of       ever, is the feedback systems to know when to
human inherited or degenerative diseases. This         stop proliferating and to start differentiating.
laboratory has also reported that co-culturing         Thus, it is not unexpected that a number of
of mouse multipotent adult progenitor cells,           signal transduction pathways and their regula-
derived from the bone marrow, with astrocytes          tory mechanisms are shared by stem cells and
facilitated neuronal differentiation, suggesting       cancer cells. Such shared regulatory pathways
that astrocyte-derived factors are required for        include expression of genes involved in pre-
this process.125                                       venting apoptosis, e.g., bcl-2, and the develop-
   A caveat to this work is that a number of           mentally regulated genes, Shh, Wnt, and Notch
laboratories have not been able to reproduce           (Fig. 4–4; see color insert). Notch and Shh are
the extreme plasticity of bone marrow–derived          involved in self-renewal of HSCs and also
cells,126,127 and there is evidence to indicate that   dysregulated in cancer cells. The Wnt pathway
the appearance of differentiated cell types from       has been shown to regulate both self-renewal
either bone marrow stem cells or ES cells is due       and oncogenesis in various organs. Data from
to fusion of these cells with adult differentiated     transgenic mice suggest that activation of the
cells in the organ in which these stem cells seed      Wnt signaling pathway in epidermal stem cells
out. Regardless of the mechanism, however, it is       produces epidermal cancers.
exciting to think that stem cells from a donor            These data lead to an obvious question: Are
could induce the production of differentiated          stem cells the targets for carcinogenic transfor-
cells that can regenerate or repair a tissue. Such     mation? If so, what dysregulates the normal
cells could also be used to deliver genes as a         feedback-regulated self-renewal process in cells
form of gene therapy.128                               undergoing malignant transformation? One
   A number of questions and caveats remain,           could further ask whether the steps leading to
however. For example, if adult stem cells are          this dysregulation are likely to be the best targets
pluripotent and have unlimited self-renewal            for therapeutic intervention. Reya et al.132 pos-
capacity, what prevents them from ‘‘taking off ’’      tulate that stem cells are the targets for transfor-
on their own and escaping the body’s feedback          mation for two reasons: (1) stem cells already
regulatory systems that stop them from becoming        have the machinery for self-renewal turned on
tumors? Since stem cells and cancer cells have a       and it would require fewer genetic or epigenetic
number of characteristics in common, are the           manipulations for a cell to become a cancer cell
stem cells present in various tissues the targets      than if they had to turn all these genes on de
for carcinogenic agents? We already know that          novo; and (2) stem cells by their vary nature are
one of the drawbacks of ES cells as a source for       set up to proliferate for several population doubl-
tissue regeneration is that they can form terato-      ings and thus have greater opportunity for car-
mas, i.e., tumors that are made up of a wide va-       ginogenic mutations to accumulate than in most
riety of cell types without any organized organ        mature cell types.
structure. They are merely diffuse masses of              Both normal cells and cancer cells have the
cells. In addition, many of the cell surface re-       ability to generate heterogenous cell types. In the
ceptors expressed on stem cells are also found on      former case, these multiple cell types in a tissue
cancer cells116,129 and the profile of expressed        assume some functional differentiated state pecu-
genes determined by DNA microarray for mu-             liar to that tissue. In the latter case, a cancer be-
rine and human HSCs overlaps with genes ex-            comes a heterogeneous mass of cells with little or
pressed in cancer cells.130 Also, a ras-like gene,     no differentiated function. Although some of the
Eras, is expressed in mouse ES cells, which may        cellular heterogenecity observed in malignant
give these cells tumor-like properties.131             tumors is a result of genetic instability and the
   Since normal stem cells and cancer cells share      resultant continuing mutagenesis, it is likely that
a number of characteristics that facilitate the        some of this heterogeneity is due to aberrant dif-
capacity for unlimited self-renewal, it seems          ferentation of cancer cells.132 Cancers often con-
likely that cancer cells acquire the machinery for     tain a mixture of cells, some of which have a par-
cell proliferation that is expressed in normal         tially differentiated phenotype, often reflective
142                                                                                    CANCER BIOLOGY

                Figure 4–4. Signaling pathways that regulate self-renewal mechanisms dur-
                ing normal stem cell development and during transformation. Wnt and Notch
                pathways have been shown to contribute to the self-renewal of stem cells
                and /or progenitors in a variety of organs, including the hematopoietic and
                nervous systems. When dysregulated, these pathways can contribute to on-
                cogenesis. Mutations of these pathways have been associated with a number
                of human tumors, including colon carcinoma and epidermal tumors for Wnt,
                mdeulloblastoma, basal cell carcinoma, and T-cell leukemias for Notch. (From
                Reya et al.,132 reprinted with permission from Macmillan Publishers Ltd.)

of an early developmental stage, and as tumor           and when transplanted in vivo, only 1% to 4%
progression occurs, cancers become less well            formed spleen colonies.133 For solid cancers,
differentiated. Also, there is variable expression of   similar data have been obtained: only 1 in 1000
cellular differentiation markers in cancers. For        to 1 in 5000 lung cancer, ovarian cancer, or neu-
example, there is variable expression of myeloid        roblastoma cells were able to form colonies in
markers in chronic myeloid leukemia, of neuronal        soft agar,134 a finding again suggesting that there
markers in neuroectodermal tumors, and of estro-        is a subpopulation of cancer cells that proliferate
gen receptor in breast cancer (reviewed in Ref-         to maintain progressive tumor growth. It has
erence 132).                                            now been possible to distinguish the genetic and
   Another implication of the idea that tissue          phenotypic characteristics of the subset of cells
stem cells are the target for carcinogenic attack       that are the more aggressive, self-renewing cells
is that, in a given cancer the cancer stem cells        in a cancer. Al-Hajj et al.135 found that when
maintain the proliferative capacity of the tumor        human breast cancer cells derived from breast
and not the whole cancer cell mass. There is            cancer patients were grown in immunocom-
good evidence to support this concept. For ex-          promised (SCID) mice, only a minority of breast
ample, when mouse myeloma cells were placed             cancer cells were able to form tumors. As few as
in an in vitro colony-forming assay, only one in        100 out of tens of thousands of cells were able to
several thousand cells was able to form colonies,       do this. The tumorigenic subpopulation was
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                              143

identified by their cell surface markers and             used to distinguish premalignant, preinvasive,
identified as having a CD44þ/CD24À pheno-                and invasive stages of human breast cancer.
type. When these cells were passaged into               Interestingly, genes highly expressed at the in-
additional mice, tumors were generated that             vasive stage were already expressed in preinva-
contained both CD44þ/CD24À cells and non-               sive stages, suggesting that the cancer stem cell
tumorigenic cells. These data demonstrate that          population may be present early in tumor devel-
only a few cells from human breast cancers have         opment. In addition, the expression of a subset
the ability to proliferate extensively, whereas the     of genes was quantitatively correlated with the
majority of cells from these tumors have only           transition from preinvasive to invasive growth. It
limited proliferative capacity in vivo. Similarly, a    has been shown that the hedgehog and Bmi-1
cancer stem cell population has been identified          signaling pathways are activated in human breast
in human brain tumors.136 These data are con-           cancer stem cells.
sistent with the concept that a cancer stem
population lurking within a human cancer con-
tains the cells responsible for the aggressive          CELL CYCLE REGULATION
growth of cancers. Further, this may be the cell
population for which biochemical markers need
                                                        Historical Perspectives
to be developed and implemented clinically to
discern which breast neoplasms to treat ag-             The development of our knowledge about cell
gressively and which may be more indolent and           cycle regulation is itself a fascinating story and
less dangerous. This would be a big help, for           takes us through a tale of fundamental dis-
example, in discriminating which breast ductal          coveries in yeast, sea urchins, clams, fruit flies,
carcinomas in situ (DCIS tumors) should un-             frogs, mice, and humans. This story serves as a
dergo more extensive surgery and chemother-             wonderful example of why fundamental, basic
apy or hormonal therapy and which may be                research should be supported for its own sake,
managed less aggressively. Similarly, such mark-        even though its primary aim at the time may
ers could be used to determine which prostate           simply be the pursuit of knowledge.
cancers should be excised, irradiated, or left for         The story of the factors involved in cell cycle
‘‘watchful waiting.’’                                   regulation goes back many years. Definition of
   In addition, it is the cancer stem cell popu-        distinct phases of a division cycle, i.e., G1, S, G2,
lation for which therapies should be targeted           and M, became established in the mid- to late
and developed. Currently available chemother-           1950s when tritum labeling and cell synchroni-
apeutic drugs were developed largely on the             zation techniques became available to score
basis of their ability to shrink a tumor mass in an     mitoses and to measure the time between one
experimental model and in a human clinical              mitotic wave and another in cycling cells (re-
trial. Since most cells in a cancerous tissue have      viewed in Reference 138). This method consists
limited proliferative potential, the ability of a       of labeling cells with a pulse of tritiated thymi-
drug to decrease a tumor mass largely reflects           dine, taking an aliquot of cells at various times
the ability of the drug to kill this less aggressive,   after the labeling, fixing the cells, and counting
potentially less dangerous type of cell, leav-          the percentage of mitoses that are labeled
ing behind the more proliferative clones. Thus,         in autoradiograms from each time point. The
drugs more specifically targeted to the cancer           percentage of labeled mitoses will rise from zero
stem cell population should result in more ef-          to a peak as the cells that were in S phase at the
fective and durable responses.                          time of the pulse go through mitosis. Following
   Ways of determining the genetic and pheno-           the peak, there will be a ‘‘trough’’ as the cells in
typic markers for cancer cells are becoming             G1 at the time of label go through M. A second
more apparent. Gene expression arrays and               cycle will then show a similar peak of labeled M
proteomic analyses are beginning to tell us what        phase cells as the first ‘‘wave’’ comes back
these markers are (see Chapter 7). For example,         through the cell cycle, but the peak will be lower
Ma et al.137 have shown that laser capture              because of dilution of the [3H] thymidine as
microdissection and DNA microarrays can be              another round of DNA synthesis occurs and
144                                                                                   CANCER BIOLOGY

because of some spread of cell cycle times as           tion promoting factor (MPF) in unfertilized frog
synchrony is diminished (Fig. 4–5). In this way,        (Xenopus) eggs, originally identified by Masui
the classic cell cycle of G1 ? S ? G2 ? M was           and Markert (reviewed in Reference 142), in-
established.                                            duced immature oocytes to undergo mitotic
   It became clear from early studies of yeast          division and was conserved in oocytes of dis-
mutants that certain genetically controlled fac-        tantly related species such as starfish. MPF was
tors played a key role in regulating the cell cycle.    found to be a protein complex containing a
In the 1970s, Lee Hartwell and colleagues               factor identical to the cdc2 gene product of
identified mutants of the budding yeast Saccha-          S. pombe (reviewed in Reference 143). It was
romyces cerevisiae that had defects at specific          also later shown that a homologous human gene
stages of the cell cycle. Temperature-sensitive         could substitute for a defective cdc2 gene in S.
mutants that were defective in initiation of DNA        pombe,144 indicating the evolutionary linkage
replication, DNA elongation, DNA ligation, tu-          for this key cell cycle gene.145
bulin assembly, spindle elongation, chromatin              Another key to the puzzle was discovered
assembly, sister chromated separation, nuclear          when Tim Hunt and his students in the Physi-
division, and cytokinesis were identified (re-           ology Course at Woods Hole Marine Biological
viewed in Reference 139). This led to the con-          Laboratory were looking for a simple experi-
cept of ‘‘checkpoints’’ in the cell cycle that sense    ment to study sea urchin egg development (re-
the completion of one event before allowing the         viewed in Reference 142 and 145). They labeled
cell to proceed to the next event. (This will be        the eggs continuously with [35S] methionine
discussed in more detail below.) Hartwell et al.        after fertilization and analyzed by SDS-gel elec-
identified a series of genes called cdc genes,           trophoresis the pattern of labeled proteins over
whose mutation produced defects in cell cycle           time. They found that one protein about 55 kDa
progression. The mutated genes were postulated          was strongly labeled after fertilization but that
to be the ones involved in the checkpoints. One         its presence seemed to oscillate with the division
gene, called cdc28, appeared to control entry           cycle in that it built up during interphase and
into mitosis. A similar gene, cdc2, was later dis-      was lost at about the time the egg divided. Hunt
covered in the fusion yeast S. pombe by Nurse           and colleagues made similar observations in a
and Bisset.140 That these genes had the same            second species of sea urchins and in the clam
function was shown by the observation that the          Spisula solidissima, and they named this protein
cdc28 gene could substitute for cdc2 mutants            ‘‘cyclin.’’146
of S. pombe in allowing the cell cycle to pro-             What remained was to link all these pieces
ceed.141                                                together. The linkage of MPF to cdc2 and then
   Another piece fell into place when it was re-        to cyclin came from several approaches (re-
alized that a cytoplasmic factor called matura-         viewed in Reference 142). Purification of MPF
                                                        from frog eggs showed that it contained CDC2,
                                                        the gene product of cdc2, and a cyclin called
                                                        cyclin B. CDC2 was shown to have a protein
                                                        kinase activity that phosphorylated histone
                                                        H1 and that oscillated in activity with cell cycle
                                                        phase. Reconstitution experiments using frog
                                                        egg cytoplasm depleted of endogenous tran-
                                                        scripts showed that the presence of a cyclin
                                                        whose synthesis varied with the cell cycle was
                                                        tied to progression of cell division. Identification
                                                        of cyclin homologs in the yeast S. cerevisiae
                                                        (called CLN 1, 2, and 3), whose levels vary with
Figure 4–5. Diagram of labeled mitoses (metaphases)     various phases of the cell cycle and which bind
in successive cell samples after a pulse of tritiated   to the CDC2 homolog CDC28 and regulate its
thymidine. (From Mitchinson,138 with permission.)       kinase activity (as well as similar data from other
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                        145

              Figure 4–6. The cyclin–cdc2 cycle. During interphase, cyclin B accumulates
              and associates with cdc2 to form pre-MPF. This is then sequentially phos-
              phorylated by wee1 (a cyclin-cdc2-specific protein tyrosine kinase) on Tyr15 of
              cdc2, then by CAK (cdc2-activating kinase) on Thr161: cdc25, a protein ty-
              rosine phosphatase then dephosphorylates Tyr15, leaving active MPF. This
              triggers mitosis and activates cyclin protease. As cyclin is broken down, MPF
              disperses and the cyclin-degrading enzymes are inactivated. Thus, cyclin be-
              gins to accumulate once more. (From Kirschner,142 with permission.)

organisms), provided an early picture of the cell     ology or Medicine in recognition of their pio-
cycle as shown in Figure 4–6.                         neering work on cell cycle regulation. Hartwell
   Unfortunately or fortunately, depending on         was recognized for identification of yeast (S.
your point of view, the story has become much         cerevisiae) genes involved in cell cycle regula-
more complicated, even in yeast. Genetic stud-        tion, his lab identifying over 100 of them, in-
ies in S. pombe have identified a negative reg-        cluding cdc28, also called start. He also intro-
ulator of CDC2, called wee 1, which is a tyrosine     duced the concept of cell cycle checkpoints and
kinase that phosphoryates a specific tyrosine on       identified five genes involved in those cell cycle
CDC2, tyrosine 15, leading to inactivation of         stop signals.
CDC2. CDC2 is re-activated by a phosphatase,             Paul Nurse, working with another species of
called CDC25 in S. pombe, which removes the           yeast, S. pombe, identified mutants that showed
phosphate at tyrosine 15. A second phosphory-         which genes were important in cell cycle regu-
lation step, phosphorylation of threonine 167, is     lation, and showed that one he identified as cdc2
required for activation of CDC2 kinase activity.      was functionally equivalent to start. These two
Thus, the phosphorylation state of CDC2 is            genes were later shown to encode enzymes be-
important for its regulation. Similar activation      longing to the family of cyclin-dependent ki-
and inactivation events for CDC2-like kinases         nases (CDKs; see below). Homologs of the yeast
have been described in Drosophila and mam-            CDKs were also found in human cells, indicat-
malian cells. Indeed, CDC2-like kinases are key       ing that these important cell cycle regulatory
cell cycle regulators in all cell types examined,     genes were conserved during evolution.
making it the mother of all cell cycle kinases.          Tim Hunt’s contribution was the discovery of
   Lee Hartwell, Paul Nurse, and Tim Hunt             cyclins and how their levels varied during phases
were awarded the 2001 Nobel Prize for Physi-          of the cell cycle.
146                                                                                      CANCER BIOLOGY

The Molecular Players                                                 Assembly &          p27Kip1     p27Kip1     P
      Cyclin-Dependent Protein Kinases                                                        E-cdk2

Cyclin-dependent protein kinases (CDKs), of            Mitogenic
                                                                       Cyclin D-
                                                                                       Rb              Cyclin E
which CDC2 is only one, are crucial regulators                        Dependent
                                                        Signals                        E2F
of the timing and coordination of eukaryotic cell                                               E2F
cycle events. Transient activation of members of
this family of serine/threonine kinases occurs at                                                     Cyclin A
                                                                                                     & Other S
specific cell cycle phases. In budding yeast, G1                                                     Phase Genes
cyclins encoded by the CLN genes (see above)
interact with, and are necessary for the activa-      Figure 4–7. Restriction point control and the G1-S
tion of, the CDC2 kinase (also called p34cdc2),       transition. As cells enter the division cycle from qui-
                                                      escence, the assembly of cyclin D–dependent kinases
driving the cell cycle through a regulatory point     in response to mitogenic signals requires Cip/Kip
called START (because it is regulated by the          proteins, which are incorporated into catalytically ac-
cdc2 or start gene) and committing cells to enter     tive holoenzyme complexes. The cyclin D–dependent
S phase. START is analogous to the G1 restric-        kinases initiate Rb phosphorylation, releasing E2F
tion point in mammalian cells.147 The CDKs            from negative constraints and facilitating activa-
                                                      tion of a series of E2F-responsive genes, the prod-
work by forming active heterodimeric com-             ucts of which are necessary for S-phase entry.
plexes following binding to cyclins, their regu-      Activation of cyclin E by E2F enables formation of the
latory subunits (reviewed in Reference 148).          cyclin E–cdk2 complex. This is accelerated by the con-
CDK2, 4, and 6, and possibly CDK3 cooperate           tinued sequestration of Cip/Kip proteins into com-
to push cells through G1 into S phase. CDK4           plexes with assembling cyclin D–cdk complexes.
                                                      Cyclin E–cdk2 completes the phosphorylation of Rb,
and CDK6 form complexes with cyclins D1, D2,          further enabling activation of E2F-responsive genes,
and D3, and these complexes are involved in           including cyclin A. Cyclin E–cdk2 also phosphory-
completion of G1. CyclinD–dependent kinases           lates p27Kip1, targeting it for ubiquitination and pro-
accumulate in response to mitogenic signals,          teasonmal degradation. The initiation of the self-
and this leads to phosphorylation of the Rb           reinforcing E2F transcriptional program together with
                                                      degradation of p27Kip1 alleviates mitogen dependency
protein. This process is completed by the cyclin      at the restriction point and correlates with the com-
E1- and E2-CDK2 complexes. Once cells enter           mitment of cells to enter S phase. In subsequent cycles,
S phase, cyclin E is degraded and A1 and A2           cyclin D–dependent kinases remain active as long as
cyclins get involved by forming a complex with        mitogens are present, and levels of p27Kip1 remain low.
CDK2. There are a number of regulators of             All p27Kip1 in cycling cells is complexed with cyclin D–
                                                      cdk complexes. Mitogen withdrawal results in cyclin D
CDK activities; where they act in the cell cycle is   degradation, liberating p27Kip1 from this latent pool.
depicted in Figure 4–7.                               The resulting inhibition of cyclin D– and E–dependent
                                                      kinases leads to cell cycle arrest, usually within a single
                                                      cycle. (From Sherr,149 reprinted with permission from
      CDK Inhibitors                                  the American Association for Cancer Research.)

The inhibitors of CDKs include the Cip/Kip
and INK4 family of polypeptides (reviewed in          and promotes cell cycle arrest. The INK4-
Reference 149). The Cip/Kip family includes           directed arrest of the cell cycle in G1 keeps Rb in
p21cip1, p27kip1, and p57kip2. The actions of these   a hypophosphorylated state and represses the
proteins are complex. Although the Cip/Kip            expression of S-phase genes.149
proteins can inhibit CDK2, they are also involved        Four INK4 proteins have been identified:
in the sequestration of cyclin D-dependent ki-        p16INK4a, p15INK4b, p18INK4c, and p19INK4d.
nases that facilitates cyclin E-CDK2 activation       INKA4a loss of function occurs in a variety of
necessary for G1 /S transition.                       cancers including pancreatic and small cell lung
   The INK4 proteins target the CDK4 and              carcinomas and glioblastomas. INK4a fulfills
CDK6 kinases, sequester them into binary CDK-         the criteria of a tumor suppressor and appears to
INK4 complexes, and liberate bound Cip/Kip            be the INK4 family member with the most ac-
proteins. This indirectly inhibits cyclin E–CDK       tive role in this regard.149
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                                147

   The INK4a gene encodes another tumor                      of the cell cycle. Although both cyclins A and B
suppressor protein called ARF (p14ARF). Mice                 interact with p34cdc2 to induce maturation and
with a disrupted ARF gene have a high propen-                mitosis in Xenopus oocytes, the synthesis and
sity to develop tumors, including sarcomas, lym-             destruction of cyclin A occurs earlier in the cell
phomas, carcinomas, and CNS tumors. These                    cycle than cyclin B. Cyclin A is first detected
animals frequently die at less than 15 months                near the G1 /S transition and cyclin B is first
of age. ARF and p53 act in the same pathway to               synthesized during S phase and accumulates in
insure growth arrest and apoptosis in response to            complexes with p34cdc2 as cells approach the
abnormal mitogenic signals such as myc-induced               G2-to-M transition. Cyclin B is then abruptly
carcinogenesis (Fig. 4–8).                                   degraded during mitosis. Thus, cyclins A and B
                                                             regulate S and M phase but do not appear to
                                                             play a role in G1 control points such as the re-
                                                             striction point (R point), which is the point
The originally discovered cyclins, cyclin A and B,           where key factors have accumulated to commit
identified in sea urchins, act at different phases            cells to enter S phase.

               Figure 4–8. ARF tumor surveillance. When induced by inappropriate mito-
               genic signals, ARF antagonizes Mdm2 to activate p53. Hence, hyperproli-
               ferative signals are countered by ARF-dependent p53 induction, which diverts
               incipient cancer cells to undergo growth arrest and /or apoptosis. Loss of the
               ARF checkpoint (indicated by the vertical barrel) subverts this form of cell-
               autonomous tumor surveillance and allows proteins such as Ras, Myc, E1A, and
               E2F to function as ‘‘pure’’ proliferation enhancers. DNA damage signals engage
               various ARF-independent signaling pathways (shown collectively by the hori-
               zontal barrel) that stabilize p53, most commonly by inducing post-translational
               modifications in p53 and /or Mdm2 that prevent their interaction. Although ARF
               is not directly activated by ionizing radiation or various genotoxic drugs, it is still a
               potent modifier of the DNA damage response. ARF induction sensitizes cells to
               DNA damage signals; conversely, ARF loss increases the Mdm2 response and
               severely dampens the p53 response. All proteins enclosed by shaded boxes are
               potential oncogenes, whereas those illustrated by unfilled boxes are tumor
               suppressors. (From Sherr,149 reprinted with permission from the American
               Association for Cancer Research.)
148                                                                                 CANCER BIOLOGY

   Three more recently discovered mammalian           the first pieces of data suggesting that the mam-
cyclins, C, D1, and E, are the cyclins that reg-      malian G1 cyclin-dependent kinases are targets
ulate the key G1 and G1 /S transition points          for negative regulators of the cell cycle.155
(reviewed in Reference 150). Unlike cyclins A
and B, cyclins C, D1, and E are synthesized
                                                           Cell Cycle Checkpoints
during the G1 phase in mammalian cells. Cyclin
C levels change only slightly during the cell         The role of various CDKs, cyclins, and other
cycle but peak in early G1. Cyclin E peaks at the     gene products in regulating checkpoints at G1 to
G1–S transition, suggesting that it controls entry    S, G2 to M, and mitotic spindle segregation have
into S. Three distinct cyclin D forms, D1, 2, and     been described in detail elsewhere.156–158 Al-
3, have been discovered and are differentially        terations of one or more of these checkpoint
expressed in different mouse cell lineages.           controls occur in most, if not all, human cancers
These D cyclins all have human counterparts.          at some stage in their progression to invasive
Cyclin D levels are growth factor dependent in        cancer. Examples of some of these alterations
mammalian cells: when resting cells are stimu-        are given below.
lated by growth factors, D-type cyclin levels rise       A key player in the G1–S checkpoint system is
earlier than cyclin E levels, implying that they      the retinoblastoma gene rb. Phosphorylation
act earlier in G1 than E cyclins. Cyclin D levels     of the Rb protein by cyclin D–dependent kinase
drop rapidly when growth factors are removed          releases Rb from the transcriptional regulator
from the medium of cultured cells. As noted           E2F and activates E2F function. Inactivation of
above, all of these cyclins (C, D, and E) form        rb by genetic alterations occurs in retinoblas-
complexes with, and regulate the activity of,         toma and is also observed in other human can-
various CDKs and these complexes control the          cers, for example, small cell lung carcinomas
various G1, G1–S1, and G2–M transition points.        and osteogenic sarcomas.
   A variety of viral oncogenic proteins also get        The p53 gene product is an important cell
into the act here. The adenovirus E1A protein         cycle checkpoint regulator at both the G1–S
binds the cyclin A-p34cdc2 complex, via its inter-    and G2–M checkpoints but does not appear to
action with p107 and Rb151 and may act to ‘‘strip’’   be important at the mitotic spindle checkpoint
Rb or p107 from the cyclin–CDK complex, aid-          because gene knockout of p53 does not alter
ing in the CDK activation. Introduction of a          mitosis. The p53 tumor suppressor gene is the
constitutively acting c-myc gene into BALB/3T3        most frequently mutated gene in human cancer,
mouse fibroblasts activated cyclin A expression        indicating its important role in conservation of
and produced a growth factor–independent as-          normal cell cycle progression. One of p53’s es-
sociation of cyclin A-CDK2 with the transcrip-        sential roles is to arrest cells in G1 after geno-
tion factor E2F, which correlated with an in-         toxic damage, to allow for DNA repair prior to
crease in E2F transcriptional activity.152 In this    DNA replication and cell division. In response
model system, myc-transformed cells reduced cy-       to massive DNA damage, p53 triggers the apo-
clin D1 expression in early G1. In addition, both     ptotic cell death pathway. Data from short-term
the src gene product p60c-src 153 and SV 40 large     cell-killing assays, using normal and minimally
T antigen154 are phosphorylated by p34cdc2,           transformed cells, have led to the conclusion
which suggests that this phosphorylation event is     that mutated p53 protein confers resistance to
involved in the effects of these oncogenic pro-       genotoxic agents.
teins on DNA replication and cell proliferation.         The spindle assembly checkpoint machinery
   Interestingly, negative growth regulators also     involves genes called bub (budding uninhibited
interact with the cyclin-CDK system. For exam-        by benomyl) and mad (mitotic arrest defi-
ple, TGF-b1, which inhibits proliferation of epi-     cient).158 There are three bub genes and three
thelial cells by interfering with G1-S transition,    mad genes involved in the formation of this
reduced the stable assembly of cyclin E-CDK2          checkpoint complex. A protein kinase called
complexes in mink lung epithelial cells, and          Mps1 also functions in this checkpoint function.
prevented the activation of CDK2 kinase activity      The chromosomal instability, leading to aneu-
and the phosphorylation of Rb. This was one of        ploidy in many human cancers, appears to be
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                      149

due to defective control of the spindle assembly      (also called Cds1), are involved in checkpoint
checkpoint. Mutant alleles of the human bub1          controls that affect a number of genes involved
gene have been observed in colorectal tumors          in maintenance of genome integrity (Fig. 4–9;
displaying aneuploidy. Mutations in these spin-       see color insert). Chk1 and Chk2 are activated
dle checkpoint genes may also result in in-           by DNA damage and initiate a number of cel-
creased sensitivity to drugs that affect microtu-     lular defense mechanisms that modulate DNA
bule function because drug-treated cancer cells       repair pathways and slow down the cell divi-
do not undergo mitotic arrest and go on to die.       sion cycle to allow time for repair. If DNA is
   Maintaining the integrity of the genome is         not successfully mended, the damaged cells
a crucial task of the cell cycle checkpoints.         usually undergo cell death via apoptosis (see
Two checkpoint kinases, called Chk1 and Chk2          below). This process prevents the defective

              Figure 4–9. Chk1 and Chk2 as mediators of the checkpoint signaling net-
              work. Following their activation, Chk1 and Chk2 phosphorylate unique (green
              and red, respectively) and overlapping (blue) downstream effectors that fur-
              ther propagate the checkpoint signaling. Depending on the type of stress,
              velocity of DNA damage, and cellular context, this leads to (1) switch to the
              stress-induced transcription program (E2F1, Brca1, p53); (2) direct or indi-
              rect initiation of DNA repair (BRCA1, p53); (3) acute delay (degradation of
              Cdc25A) and /or sustained block (Cdc25C, p53, E2F1); (4) apoptosis (Pml1,
              p53, E2F1); and (5) modulation of the chromatin remodeling pathways (Tlk
              1/2). The known target sites of Chk1 (green), Chk2 (red), and both Chk1 and
              Chk2 (blue) on the individual substrates are shown. Some of the Chk1 and Chk2
              downstream effectors are classified as proto-oncogenes (PO) or tumor sup-
              pressors (TS), as indicated. (From Bartek and Lucas,159 reprinted with per-
              mission from Elsevier.)
150                                                                                CANCER BIOLOGY

genome from extending its paternity into daugh-      syndrome (LFS), thus Chk2 mutations may be an
ter cells.                                           alternative or overlapping genetic defect along
   Upstream elements activating the checkpoint       with p53 mutations in these patients. Since LFS
signaling pathways such as those turned on by        patients are susceptible to develop multiple types
irradiation or agents causing DNA double-            of tumors, including a predominant incidence of
strand breaks include the ATM kinase, a mem-         breast cancers and sarcomas, the Chk2 path may
ber of the phosphatidylinositol 3-kinase (PI3K)      also be an important tumor suppressor for these
family, that activates Chk2 and its relative ATR     tumors in non-LFS patients. Chk2 mutations
kinase that activates Chk1. There is also cross      have also been found in small subsets of ‘‘spora-
talk between ATM and ATR that mediates these         dic’’ human cancers, including carcinomas of
responses (reviewed in Reference 159). Chk1          the breast, lung, vulva, urinary bladder, colon,
and Chk2 phosphorylate CDC25A and C, which           and ovary as well as in osteosarcomas and lym-
inactivate them. In its dephosporylated state        phomas.159 In contrast, cancer-associated ge-
CDC25A activates the CDK2-cyclin E complex           netic defects in Chk1 are rare but have been
that promotes progression through S phase. It        observed in carcinomas of the colon, stomach, and
should be noted that this is an example of           endometrum.
dephosphorylation rather than phosphorylation           As illustrated in Figure 4–9, there are interac-
activating a key biological function. This is in     tions between the Chk kinases and the p53
contrast to most signal transduction pathways,       pathway. Chk2 phosphorylates threonine-18 or
where the phosphorylated state of a protein (of-     serine-20 on p53, which attenuates p53’s inter-
ten a kinase) is the active state and the dephos-    action with its inhibitor MDM2, thus contribut-
phorylated state is the inactive one (see Signal     ing to p53 stabilization and activation. However,
Transduction Mechanisms, below). In addition,        Chk2 and p53 only have partially overlapping
Chk1 renders CDC25A unstable, which also di-         roles in checkpoint regulation because not all
minishes its activity (reviewed in Reference 160).   DNA-damaging events activate both pathways.
   CDC25A also binds to and activates CDK1-          For example, some types of DNA damage that
cyclin B, which facilitates entry into mitosis.      activate p53 do not activate Chk2 and vice versa.
G2 arrest induced by DNA damage induces              Thus, the two pathways are partly redundant and
CDC25A degradation and, in contrast, G2 arrest       overlapping but not totally so, as evidenced by
is lost when CDC25A is overexpressed.                the fact that in Chk2-deficient cells, Chk1 can
   A number of proteins are now known to act         still phosphorylate and activate p53.
as mediators of checkpoint responses by imping-
ing on the Chk1 and 2 pathways. These include
                                                          Cell Cycle Regulatory Factors as
the BRCT domain–containing proteins 53BP1,
                                                          Targets for Anticancer Agents
BRCA1, and MDC1. These proteins are involved
in activation of Chk1 and Chk2 by acting through     The commonly observed defects in cell cycle
protein–protein interactions that modulate the       regulatory pathways in cancer cells distinguishes
activity of these checkpoint kinases. In genereal,   them from normal cells and provides potential
these modulators are thought to be tumor sup-        targets for therapeutic agents. One approach is
pressors.                                            to inhibit cell cycle checkpoints in combination
   Chk1 and 2 have overlapping roles in cell cycle   with DNA-damaging drugs or irradiation. The
regulation, but different roles during develop-      rationale for this is that normal cells have a full
ment. Chk1 but not Chk2 is essential for mam-        complement of checkpoint controls, whereas tu-
malian development, as evidenced by the early        mor cells are defective in one or more of these and
embryonic lethality of Chk1 knockout mice.           thus are more subject to undergoing apoptosis
Chk2-deficient mice are viable and fertile and do     in response to excessive DNA damage. This has
not have a tumor-prone phenotype unless exposed      been accomplished by combining ATM/ATR
to carcinogens, and this effect is more evident      inhibitors such as caffeine or Chk1 inhibitors
later in life (reviewed in Reference 159). Rare      in combination with DNA-damaging drugs. So
germline mutations of Chk2 have been observed        far this approach hasn’t been demonstrated
in cancer-prone patients with the Li-Fraumeni        clinically, and indeed is somewhat counterintu-
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                          151

itive, since p53 mutant tumor cells are more         to interdict the overactive cell proliferation path-
resistant to many chemotherapeutic drugs. p53        ways in cancer cells. Thus, inhibition of cyclins
is a key player in causing cell death in drug-       D1 and E and CDKs, especially CDK4 and
treated, DNA-damaged cells (one exception to         CDK6, could be targets for inhibiting growth of
that is the microtubule inhibitor paclitaxel), and   cancers. As more knowledge of the complicated
active, unmutated p53 is needed for this re-         steps in cell cycle regulation is gained, more
sponse.                                              potential targets become available. For example,
   Another approach is to target the cyclin-         Bettencourt-Dias et al.164 used RNA-mediated
dependent kinases directly. Alteration of the G1–    interference (RNAi) to carry out a genome-wide
S checkpoint occurs in many human cancers.           survey of protein kinases required for cell cycle
Cyclin D1 gene amplification occurs in a subset       progression in Drosophila and have found as
of breast, esophageal, bladder, lung, and squa-      many as 80 protein kinases, a number of them
mous cell carcinomas. Cyclins D2 and D3 are          previously unknown, that regulate the cell divi-
overexpressed in some colorectal carcinomas. In      sion cycle. A number of these will no doubt have
addition, the cyclin D–associated kinases CDK4       human orthologs.
and CDK6 are overexpressed or mutated in some
cancers. Mutations or deletions in the CDK4 and
CDK6 inhibitor INK4 have been observed in            APOPTOSIS
familial melanomas, and in biliary tract, esopha-
geal, pancreatic, head and neck, non–small cell      Apoptosis (sometimes called programmed cell
lung, and ovarian carcinomas. Inactivating mu-       death) is a cell suicide mechanism that enables
tations of CDK4 inhibitory modulators p15, p16,      multicellular organisms to regulate cell number
and p18 have been observed in a wide variety of      in tissues and to eliminate unneeded or aging
human cancers. Cyclin E is also amplified and         cells as an organism develops. The biochemistry
overexpressed in some breast and colon carci-        of apoptosis has been well studied in recent years,
nomas and leukemias.                                 and the mechanisms are now reasonably well
   Human cancers have a variety of mutations in      understood.165–167 The enzymatic machinery for
cell cycle regulatory genes (reviewed in Refer-      this was first discovered in the nematode C. ele-
ence 148). This includes overexpression of D1        gans, and later the homologues of these genes
and E1 cyclins and CDKs (mainly CDK4 and             and their products were identified in mammalian
CDK6) as noted above. Loss of CDK inhibitory         cells, including human cells. The apoptosis path-
functions (mainly INK4a and 4b and Kip1) also        way involves a series of positive and negative reg-
occurs, as does loss of Rb, one of the first tumor    ulators of proteases called caspases, which cleave
suppressor genes identified (see Chapter 5). Loss     substrates, such as poly (ADP-ribose) polymer-
of Kip1 function and overexpression of cyclin E1     ase, actin, fodrin, and lamin. In addition, apo-
occur frequently and are associated with poor        ptosis is accompanied by the intranucleosomal
prognosis in breast161 and ovarian cancers.162       degradation of chromosomal DNA, producing
   The mitogen-stimulated proliferation of cells     the typical DNA ladder seen for chromatin iso-
is mediated via a retinoblastoma (Rb) pathway        lated from cells undergoing apoptosis. The en-
that involves phosphorylation of Rb, its dissoci-    donuclease responsible for this effect is called
ation from and activation of the E2F family of       caspase-activated DNase, or CAD.
transcription factors, and subsequent turn-on of        A number of ‘‘death receptors’’ have also
genes involved in G1–S transition and DNA            been identified.168 Death receptors are cell sur-
synthesis (reviewed in Reference 163). Disrup-       face receptors that transmit apoptotic signals
tion of this pathway by overexpression of cyclin     initiated by death ligands. The death receptors
D1, loss of the INK4 inhibitor p16, mutation of      sense signals that tell the cell that it is in an
CDK4 to a p16-resistant form, or loss or muta-       uncompromising environment and needs to die.
tion of Rb is frequently seen in cancer cells. The   These receptors can activate the death caspases
activation of CDK inhibitory factors such as         within seconds of ligand binding and induce
p16INK4 or p27kip1 and inhibition of cyclin-         apoptosis within hours. Death receptors belong
dependent kinases are therefore potential ways       to the tumor necrosis factor (TNF) receptor
152                                                                                   CANCER BIOLOGY

gene superfamily and have the typical cystine-          the existence of two morphologically distinct
rich extracellular domains and an additional            types of cell death was obtained by Kerr170 in
cytoplasmic sequence termed the death do-               1965 from histochemical studies of ischemic
main. The best-characterized death receptors            injury to rat liver. Some cell death occurred with
are CD95 (also called Fas or Apo1) and TNF              the typical changes seen in tissue necrosis:
receptor TNFR1 (also called p55 or CD120a).             clumping of chromatin into ill-defined masses,
   The importance of the apoptotic pathway in           swelling of organelles, flocculent densities in
cancer progression is seen when there are mu-           the matrix of mitochondria, membrane disinte-
tations that alter the ability of the cell to undergo   gration, and infiltration of inflammatory cells.
apoptosis and allow transformed cells to keep           Cells in some areas of the damaged liver, how-
proliferating rather than die. Such genetic alter-      ever, died a different death. They contained
ations include the translocation of the bcl-2 gene      chromatin compacted into sharply delineated
in lymphomas that prevents apoptosis and pro-           masses, condensation of the cytoplasm, and out-
motes resistance to cytotoxic drugs. Other genes        cropping of cytoplasmic ‘‘blebs’’ or proturber-
involved as players on the apoptosis stage include      ances that became pinched off (apoptotic bod-
c-myc, p53, c-fos, and the gene for interleukin-        ies) and released, to be devoured by tissue
1b-converting enzyme (ICE). Various oncogene            phagocytic cells. No inflammatory reaction,
products can suppress apoptosis. These include          however, was noted around cells dying by this
adenovirus protein E1b, ras, and n-abl.                 second mechanism. Further studies by Currie
   Mitochondria play a pivotal role in the events       and Wyllie171 showed that this second mecha-
of apoptosis by at least three mechanisms:              nism of cell death occurs as tissues undergo
(1) release of proteins, e.g., cytochrome c, that       remodeling during development and in this sense
triggers activation of caspases, (2) alteration of      is a ‘‘physiologic cell death.’’ The original term
cellular redox potential, and (3) production and        for this phenomenon, shrinkage necrosis, didn’t
release of reactive oxygen species after mito-          seem an appropriate one for this process, so
chondrial membrane damage.169 Another mi-               Kerr et al. searched for another one. A colleague
tochondrial link to apoptosis is implied by the         of theirs at the University of Aberdeen, profes-
fact that Bcl-2, the anti-apoptotic factor, is a        sor James Carmack of the Department of Greek,
mitochondrial membrane protein that appears             suggested the term apoptosis (pronounced
to regulate mitochondrial ion channels and pro-           ´ ¯
                                                        apqto, with the second p silent), meaning ‘‘fall-
ton pumps.                                              ing off ’’ of petals from a flower or leaves from a
   Apoptosis occurs in most, if not all, solid can-     tree. The term has stuck ever since. The concept
cers. Ischemia, infiltration of cytotoxic lympho-        of apoptosis however, was largely ignored until
cytes, and release of TNF may all play a role in        the mid- to late 1980s when the discovery of the
this. It would be therapeutically advantageous to       ced genes in the roundworm Caenorhabditis
tip the balance in favor of apoptosis over mitosis      elegans and of the bcl-2 gene in B lympho-
in tumors, if that could be done. It is clear that a    cytes (see below) put the field on a solid genetic
number of anticancer drugs induce apoptosis in          basis.
cancer cells. The problem is that they usually do          A number of the cellular morphological and
this in normal proliferating cells as well. There-      biochemical changes that occur during the ap-
fore, the goal should be to manipulate selectively      optosis have been worked out. These include
the genes involved in inducing apoptosis in tu-         morphological changes that can be observed
mor cells. Understanding how those genes work           both at the light and electron microscopic levels
may go a long way to achieving this goal.               (Figs. 4–10 and 4–11).172 One of the most ex-
                                                        tensively studied biochemical events in apopto-
                                                        sis is the double-strand cleavage of nuclear DNA
Historical Perspectives
                                                        that occurs at linker strands between nucleo-
The study of cell death is a lively field, as evi-       somes, producing fragments that are multiples
denced by the tremendous spate of recent pub-           of about 185 base pairs. These fragments can be
lications and scientific meetings covering the           observed as characteristic apoptotic DNA ‘‘lad-
subject. But it wasn’t always so. Evidence for          ders’’ by agarose gel electrophoresis.
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                           153

               Figure 4–10. Apoptosis of murine NS-1 cell occurring spontaneously in cul-
               ture. Note the discrete nuclear fragments with characteristic segregation of
               compacted chromatin, the crowding of organelles, and the marked convolu-
               tion of the cellular surface. (From Kerr and Harmon,172 with permission.)

   Other biochemical changes that appear to           Ca2þ precede apoptosis induced by a number of
correlate with the induction of the apoptotic         agents or conditions, and apoptosis is delayed or
cascade of events include CA2þ ion influx into         inhibited when Ca2þ is depleted from the cel-
cells, induction of transglutaminase that catalyzes   lular growth medium. In addition, Ca2þ iono-
the formation of e (g-glutamyl) lysine cross-lines    phores such as A23187, which carries Ca2þ into
between proteins, disruption of microtubules,         cells, induces apoptosis in some cell types.
induction of calmodulin expression, loss of cell         Induction of transglutaminase accompanies
membrane phospholipid asymmetry resulting in          apoptosis in several cell types. There is evidence
exposure of phosphatidylserine on the cell sur-       that this enzyme causes protein-protein cross-
face, activation of the cell surface receptor Fas/    linking to produce a protein net or meshwork
Apo1, activation of protein kinase C (some cells      that holds the cell intact and prevents leakage of
only), and induction of a neutral sphingomyeli-       cellular contents until the apoptotic bodies bud
nase in the plasma membrane that releases phos-       off and are consumed by phagocytes.172
phocholine and ceramide.173 The biochemical              Loss of microtubular structures and reorgani-
mechanisms involved in apoptosis are described        zation of the cytoskeleton occurs in apoptotic cells.
below.                                                Vinca alkaloids such as vincristine, which cause
   Apoptosis can be triggered by a number of          microtubule dissolution, can trigger apoptosis.
agents or stimuli, including events triggered in      However, cytochalasin B, an inhibitor of actin
tissue differentiation during development; re-        polymerization, inhibits formation of apoptotic
moval of growth factors or hormones required          bodies but not the other features of apoptosis.174
for cell survival; exposure to TGF-b, TNF, or
glucorticoids in cell types sensitive to their
                                                      Biochemical Mechanism of Apoptosis
negative regulation; and exposure to DNA-
damaging anticancer drugs or environmental            Multicellular organisms, from the lowest to the
toxins. Sustained increases in intracellular free     highest species, must have a way to get rid of
154                                                                                       CANCER BIOLOGY

                Figure 4–11. Apoptotic body containing well-preserved rough endoplasmic
                reticulum and four nuclear fragments, which has been phagocytosed by an
                intraepithelial macrophage in the rat ventral prostate 2 days after castration.
                (From Kerr and Harmon,172 with permission.)

excess cells or cells that are damaged in order          effects facilitating caspase-3 activation, which is
for the organism to survive. Apoptosis is the            where the two pathways intersect. There are
mechanism that they use to do this. It is the way        both positive and negative regulators that also
that the organism controls cell numbers and              interact on these pathways. For example, the
tissue size and protects itself from ‘‘rogue’’ cells.    Smac/DIABLO protein blocks activation of
   A simplified version of the apoptotic pathways         inhibitor of apoptosis (IAP) proteins, which in-
can be visualized in Figure 4–12 (see color in-          hibit caspase-3. Cross talk between the two cell
sert). The death receptor–mediated pathway is            death pathways is mediated by Bid, a pro-
turned on by members of the death receptor               apoptotic member of the Bcl-2 family. Bid is
superfamily of receptors including Fas receptor          cleaved and activated by caspase-8. Bid then
(CD95) and TNF receptor 1, which are acti-               acts to promote cytochrome c release from mi-
vated by Fas ligand and TNF, respectively.               tochondria. Thus, one can see what a complex
Interaction of these ligands with their receptors        and tightly regulated pathway apoptosis is. Gi-
induces receptor clustering, binding of the re-          ven the literal life-and-death importance of this
ceptor clusters to Fas-associated death domain           pathway, it is easy to see why this is so.
protein (FADD), and activation of caspase-8.
This activation step is regulated by c-FLIP.
Caspase-8, in turn, activates caspase-3 and other
‘‘executioner’’ caspases, which induce a number          Caspases are a family of cysteine proteases that
of apoptotic substrates (see below).                     are activated specifically in apoptotic cells. This
   The DNA damage–induced pathway invokes                family of proteases is highly conserved through
a mitochondrial-mediated cell death pathway              evolution all the way from hydra and nematodes
that involves pro-apoptotic factors like Bax             up to humans. Over 12 caspases have been
(blocked by the anti-apoptotic protein Bcl-2).           identified and although most of them appear to
This results in cytochrome c release from the            function during apoptosis, the function of all of
mitochondria and triggering of downstream                them is not yet clear. The caspases are called
Figure 4–12. The roads to ruin: two major apoptotic pathways in mammalian
cells. The death-receptor pathway is triggered by members of the death-
receptor superfamily (such as CD95 and tumor necrosis factor receptor I).
Binding of CD95 ligand to CD95 induces receptor clustering and formation
of a death-inducing signaling complex. This complex recruits, via the adaptor
molecule FADD (Fas-associated death domain protein), multiple procas-
pase-8 molecules, resulting in capase-8 activation through induced proximity.
Caspase-8 activation can be blocked by recruitment of the degenerate caspase
homologue c-FLIP. The mitochondrial pathway (right) is used extensively in
response to extracellular cues and internal insults such as DNA damage. The
death-receptor and mitochondrial pathways converge at the level of caspase-3
activation. Caspase-3 activation and activity is antagonized by the inhibitors of
apoptosis (IAP) proteins, which themselves are antagonized by the Smac /
DIABLO protein released from mitochondria. Downstream of caspase-3, the
apoptotic program branches into a multitude of subprograms, the sum of
which results in the ordered dismantling and removal of the cell. Cross-talk
and integration between the death-receptor and mitochondrial pathways is
provided by Bid, a pro-apoptotic Bcl-2 family member. Caspase-8-mediated
cleavage of Bid greatly increases its pro-death activity, and results in its
translocation to mitochondria, where it promotes cytochrome c exit. Note that
under most conditions, this cross-talk is minimal, and the two pathways operate
largely independently of each other. Clearly, additional death-inducing path-
ways must exist, as developmental apoptosis is by and large normal in mice
defective in the caspase-8 and caspase-9 pathways. (From Hengartner,166
reprinted by permission from Macmillan Publishers Ltd.)
156                                                                                 CANCER BIOLOGY

cysteine-proteases because they have a cysteine       form heterodimers, which can be looked on as
in the active site that cleaves substrates after      reservoirs of plus and minus apoptotic factors
asparagines in a sequence of asp-X, with the four     waiting for the appropriate signals to be re-
amino acids amino-terminal to the cleavage site       leased. In this scenario, the one who wins de-
determining a caspase’s substrate specificity.         pends on the relative amounts of the pro- or
   The importance of the caspases in apoptosis is     anti- apoptotic factors. This ‘‘war’’ of factors
demonstrated by the inhibitory effects of muta-       occurs at the mitochondrial membrane, where
tion or drugs that inhibit their activity. Caspases   they compete to regulate release of cytochrome
can either inactivate a protein substrate by          c (see below). However, in healthy cells the pro-
cleaving it into an inactive form or activate a       apoptotic proteins Bax and Bak are not found in
protein by cleaving a pro-enzyme negative reg-        association with Bcl-2 family members. Rather,
ulatory domain. In addition, caspases themselves      Bax and Bak appear to be directly activated
are synthesized as pro-enzymes and are activated      under conditions of limiting survival signals.167
by cleavage at asp-x sites. Thus, they can be ac-
tivated by other caspases, producing elements of
                                                           Role of Mitochondria in Apoptosis
the ‘‘caspase cascade’’ shown in Figure 4–12.
Also, as illustrated in Figure 4–12, caspases are     When cells are functioning under normal ho-
activated in a number of steps by proteolytic         meostatic conditions, apoptosis is suppressed by
cleavage by an upstream caspase or by protein–        stict compartmentalization of the cell death me-
protein interactions, such as that seen for the       diators. Mitochondria are key to this in that they
activation of caspase-8 and the interaction of        contain and hold the cell death regulatory bio-
cytochrome c and Apaf-1 in the activation of          chemical signals (reviewed in Reference 177). The
caspase-9. A number of important substrates           major apoptotic pathway activator cytochrome c
of caspases have been identified, including the        is confined to the mitochondrial intermembrane
caspase-activated DNase (CAD), noted above,           space. Upon activation of cell death signals lead-
which is the nuclease responsible for the DNA         ing to permeabilization or rupture of the outer
ladder of cells undergoing apoptosis. Activation      mitochondrial membrane, cytochrome c is re-
of CAD is mediated by caspase-3 cleavage of           leased and binds to the cytosolic factor Apaf-1,
the CAD-inhibitory subunit. Caspase-mediated          leading to the allosteric activation of caspase-9,
cleavage of other specific substrates has been         which in turn activates caspase-3, as shown in
shown to be responsible for other typical changes     Figure 4–12. In addition, Smac/DIABLO is re-
seen in apoptotic cells, such as the cleavage of      leased by mitochondrial permeabilization and
nuclear lamins required for nuclear shrinkage         neutralizes the cytosolic IAPs, thus releasing
and budding, loss of overall cell shape by cleav-     their IAP inhibition of caspases-3 and -9. The
age of cytoskeleton proteins, and cleavage of         apoptosis-inducing factor (AIF) and the endonu-
PAK2, a member of the p21-activated kinase            clease CAD are also released from the mitochon-
family, that mediates the blebbing seen in dying      drial intermembrane space, travel to the nucleus,
cells (reviewed in Reference 166).                    and degrade chromatin to produce the typical
                                                      DNA ladders.
                                                         Mitochondrial permeabilization can be induced
      Bcl-2 Family
                                                      by a number of pro-apoptotic second messengers
Mammalian Bcl-2 was first identified as anti-           such as CA2þ, reactive oxygen species, lipids such
apoptotic protein in lymphomas cells.175 It           as ceramide and ganglioside GD3, and stress ki-
turned out to be a homolog of an anti-apoptotic       nases. CA2þ release from the endoplasmic re-
protein called Ced-9 described in C. elegans.176      ticulum CA2þ ‘‘storage depot’’ appears to be a
In C. elegans, Ced-9 protects from cell death by      key pro-apoptotic event, and its uptake by mito-
binding to the pro-apoptotic factor Ced-4.            chondria is mediated by the pro-apoptotic pro-
Similarly, in mammalian cells, Bcl-2 binds to a       teins Bax and Bak of the Bcl-2 family.178 Con-
number of pro-apoptotic factors such as Bax and       versely, cells are protected by appropriate levels
Apaf-1. One concept is that pro- and anti- apo-       of the anti-apoptotic proteins Bcl-2 or Bcl-XL,
ptotic members of the Bcl-2 family of proteins        which prevent activation of Bax and Bak.
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                             157

  The release of mitochondrial-associated cell          Both of these steps protect cells from anoikis,
death factors appears to be controlled by the           whereas inhibition of the PI3-kinase pathway
amount of ‘‘openness’’ of mitochondrial mem-            induces anoikis (reviewed in Reference 180).
brane ‘‘pores’’ such as the voltage-dependent           Disruption of cell-matrix interactions also turns
anion channels (VDACs) in the outer mem-                on the JNK /p38 pathway, a stress-activated
brane and the adenine-nucleotide translocase            protein kinase. The mitogen-activated kinase
(ANT) in the inner membrane. The openness of            system may also be involved, since caspase-
those pores is increased by activated Bax/Bak           mediated cleavage of MEKK-1 occurs in cells
and inhibited by Bcl-2 and Bcl-XL, which pre-           undergoing anoikis.
vent activation of Bax and Bak.179                         As stated earlier, one of the hallmarks of
                                                        malignantly transformed cells growing in cul-
                                                        ture is their ability to grow in an anchorage-
                                                        independent manner, whereas normal cells do
Anoikis is a form of apoptosis that occurs in           not. Thus, cancer cells may develop resistance
normal cells that lose their adhesion to the sub-       to anoikis. This may be a way that metastatic
strate or extracellular matrix (ECM) on which           cancer cells can survive in the bloodstream until
they are growing. Adherence to a matrix is cru-         they seed out in a metastatic site. Indeed, there
cial for the survival of epithelial, endothelial, and   is some evidence for this in that selection of a
muscle cells. Prevention of their adhesion usu-         Cloudman melanoma cell line resistant to an-
ally results in rapid cell death, which occurs via      oikis had enhanced metastatic potential.180
apoptosis. Thus, anoikis is a specialized form of
apoptosis caused by prevention of cell adhesion.
                                                        Resistance to Apoptosis in Cancer
The term anoikis means ‘‘homelessness’’ in
                                                        and Potential Targets for Therapy
Greek. Although the observation of this phe-
nomen occurs only with cultured cells, it is likely     It would be a mistake to portray apoptosis as
to occur also in vivo because it is known that cell-    only a mechanism to kill cells damaged by some
cell and cell-ECM interactions are crucial to cell      exogenous insult such as DNA-damaging toxins,
proliferation, organ development, and mainte-           drugs, or irradiation. Apoptosis is, in fact, a nor-
nance of a differentiated state (see Cell Prolif-       mal mechanism used by all multicellular or-
eration versus Differentation, above). This may         ganisms to facilitate normal development, se-
be a way that a multicellular organism protects         lection of differentiated cells that the organism
itself from free-floating or wandering cells (such       needs, and control of tissue size. For example,
as occurs in tumor metastasis). The basic rule for      studies of nematodes (C. elegans), fruit flies, and
epithelial and endothelial cells appears to be          mice indicate that apoptotic-mediated mecha-
‘‘attach or die.’’180 Interestly, cells that normally   nisms similar to those described here are in-
circulate in the body such as hematopoietic cells       trinsic and required for normal development.
do not undergo anoikis.                                 Dysfunction of these pathways results in devel-
   As noted earlier in this chapter, cell attach-       opmental abnormalities and disease states.181
ment is mediated by integrins, and ECM-                    In the human, development of the immune
integrin interactions transduce intracellular sig-      system is perhaps the best example of the role
naling pathways that activate genes involved in         for apoptosis in normal development.182 In the
cell proliferation and differentiation. Although        immune system, apoptosis is a fundamental
the cell death pathways induced by disruption of        process that regulates T- and B-cell proliferation
these cell attachment processes aren’t clearly          and survival and is used to eliminate immune
worked out, cell detachment–induced anoikis             cells that would potentially recognize and de-
does result in activation of caspases-8 and -3 and      stroy host tissues (‘‘anti-self ’’). Mechanisms in-
is inhibited by Bcl-2 and Bcl-XL, indicating            volving Apo-1/FAS (CD95)-mediated signaling
some similarities to the typical apoptosis mech-        of the caspase cascade as described above are
anisms. In addition, integrin-ECM interaction           employed in lymphocytic cell selection.
activates focal adhesion kinase (FAK) and                  In the case of T lymphocytes, pre-T cells are
attachment-mediated activation of PI3-kinase.           produced in the bone marrow and circulate to
158                                                                                  CANCER BIOLOGY

the thymus where they differentiate and re-           and Bcl-XL. Mutations or altered expression of
arrange their T-cell receptors (TCRs). Those          p53 downstream effectors (PTEN, Bax, Bak, and
cells that fail to rearrange appropriately their      Apaf-1) or upstream regulators (ATM, Chk2,
TCR genes, and thus cannot respond to self–           Mdm2 and p19ARF) also occur in human can-
major histocompatibility complex (MHC)–               cers. Overexpression of inhibitors of apoptosis
peptide complexes, die by ‘‘neglect.’’ Those T        proteins (IAPs) and heat shock proteins (Hsps),
cells that pass the TCR selection tests mature and    which can inhibit caspase-9 activation, have also
leave the thymus to become the adult peripheral       been observed in human cancers.
T-cell pool. The mature T-cell pool thus passes          The above tumor-related disruptions occur in
through a number of selection steps to ensure         the so-called intrinsic, mitochondrial-mediated
self-MHC restriction and self-tolerance. Apo-         pathway. Tumorigenic disruptions also occur,
ptosis also is used to delete mature peripheral T     though less frequently, in the death-receptor
cells that are insufficiently stimulated by positive   mediated pathway. For example, mutations in
growth signals, and this is a mechanism to down-      Fas (CD95) and TNF-related apoptosis-inducing
regulate or terminate an immune response.             ligand (TRAIL) receptors and downstream sig-
   B lymphocytes undergo selection and matu-          naling pathways of these receptors have been
ration in the bone marrow and germinal centers        seen in human cancers. Inactivation of the CD95
of the spleen and other secondary lymphoid            and TRAIL pathways may also allow tumors to
organs. Those with low antigen affinity or that        escape from immune responses and thus pro-
are autoreactive are eliminated by apoptosis.         mote tumor expansion and metastasis.183
Those that pass this test mature into memory B           It is important that inhibition of apoptosis oc-
cells and long-lived plasma cells.                    curs at different steps in different tumor types, as
   The ability of lymphoid progeny cells to avoid     this may be responsible for variability in drug
apoptosis may lead to lymphatic leukemias or          responses in different cancers. A key to tumor-
lymphomas. For example, follicular lymphomas          specific, effective therapy will be understanding
have a Bcl-2 translocation into the immunoglob-       which steps have gone awry in which tumors.
ulin heavy-chain locus that dysregulates Bcl-2           The cancer-related alterations in the apoptotic
expression. As noted above, Bcl-2 overexpres-         pathway provide a number of cancer chemo-
sion suppresses apoptosis and enhances cell           therapeutic targets.184,185 There have been no
proliferation. In addition, cancers develop mul-      clinical breakthroughs yet, but there are number
tiple mechanisms to evade destruction by the          of Bcl-2 antagonists (both small molecules and
immune system such as decreased expression            anti-sense oligonucleotides) in preclinical devel-
of MHC molecules on cancer cell surfaces and          opment as well as TRAIL agonists and IAP
production of immunosuppressive cytokines             antagonists such as Survivin. Survivin has been
(see Chapter 6).                                      shown to be overexpressed in a wide variety of
   Several cell proliferation–promoting events        human cancers and provides an interesting target
take place in cancer cells as they evolve over        because it is not only anti-apoptotic but also up-
time into growth dysregulated, invasive, meta-        regulated in angiogenically stimulated endothe-
static cell types. These events include activation    lium (reviewed in Reference 186). Thus, target-
of proliferation-promoting oncogenes such as          ing Survivin may also facilitate involution of new
ras and myc (see Chapter 5), overexpression of        blood vessels in tumors.
cell cycle regulatory factors such as cyclin D,
increased telomerase to overcome cell senes-
cence, and increased angiogenesis to enhance          GROWTH FACTORS
blood supply to tumor tissue. In addition, a
number of mutations in apoptotic factors and
                                                      Historical Perspectives
up-regulation of anti-apoptotic factors occur in
cancer cells during progression (reviewed in          Many ‘‘factors’’ that affect the proliferation of
Reference 183). These include mutation or             eukaryotic cells have been identified (Table 4–4).
inactivation of p53 and overexpression of Bcl-2       A growth factor is usually defined as a substance
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                                                      159

Table 4–4. Characteristics of Some Representative Growth Factors
Factor                           Original Source                    Target Cell                                  Molecular Weight
Insulin                          Beta cells of pancreas             General                                       6000
Insulin-like growth factors
  IGF-1                          Human plasma                       General                                       7650
  IGF-2                          Human plasma                       General                                       7500
Basic fibroblast growth factor    Bovine pituitary                   Fibroblasts, myoblasts, smooth muscle,       14,000
  (FGF-2)                                                             chondrocytes, glial cells, vascular
Nerve growth factor (NGF)        Mouse submaxillary gland,          Sympathetic ganglia cells and                26,500
                                  snake venoms, cultured              sensory neurons
Epidermal growth factor          Mouse submaxillary gland,          Epidermal cells, various epithelial cells,    6000
  (EGF)                           human urine                         vascular endothelial cells,
                                                                      chondrocytes, fibroblasts, glial cells
Platelet-derived growth factor   Human platelets                    Fibroblasts, glial cells, arterial           24,000–31,000
   (PDGF) AA, AB, and BB                                              smooth muscle cells
Transforming growth factors
  TGF-a                          Various virally transformed cell   Similar to EGF                               ^6000
                                   types and cancer cells
  TGF-b                          Various transformed cell types,    Similar to EGF                               25,000
                                   normal placenta, kidney, and
Angiogenin                       Human colon carcinoma cell line    Capillary endothelium                        14,000
Colony-stimulating factors
  GM-CSF                         Placenta                           Granulocyte/macrophage                       22,000
  G-CSF                          Placenta                           Granulocyte progenitor cells                 20,000
  M-CSF                          Urine                              Macrophage progenitor cells                  45,000
  IL-1                           Normal and malignant               T lymphocytes, B lymphocytes,                15,000
                                    macrophages, keratinocytes,        fibroblasts, hepatocytes, chondrocytes,
                                    astrocytes                         hypothalamic cells (fever center)
  IL-2 (TCGF)                    T lymphocytes                      T lymphocytes (that become T-helper          23,000
                                                                       cells or cytotoxic T cells)
  IL-3 (Multi-CSF)               T lymphocytes                      Eosinophil, mast cell, granulocyte, and      28,000
                                                                       macrophage progenitors; T lymphocytes
Mammary-derived growth           Human milk and mammary             Normal mammary epithelial cells,             62,000
 factor (MDGF)                     tumors                             epidermoid carcinoma cells
Uterine-derived growth factor    Pregnant sheep uterus              Rat mammary and uterine tumor cells          4000–6000

that stimulates cell proliferation and often also                   may help to elucidate the altered growth control
promotes cell differentiation of specific target                     of malignant cells because transformed cells
cells. Excluded from this class of agents are                       usually have a lower growth requirement for se-
substances that are simply nutrients, such as                       rum and some transformed cells appear to pro-
glucose, essential amino acids, vitamins, and key                   duce their own growth factors.
minerals. The impetus for studies of growth                           Growth factors are now established as im-
factors largely derives from earlier observations                   portant regulators of embryonic growth and
that most mammalian cells growing in culture                        development, cellular proliferation and differ-
require the presence of animal serum to grow                        entiation, and tumor cell proliferation. While
and proliferate. Numerous attempts have been                        the modern era of growth factors research dates
made to isolate growth factors from serum. The                      to the late 1940s and early 1950s with the
isolation and characterization of these factors                     seminal work of Rita Levi-Montalcini, Stanley
160                                                                                  CANCER BIOLOGY

Cohen, and Viktor Hamburger, research in this             A large number of growth-promoting factors
field actually started about 50 years earlier than      have now been found. In general, these are
that (reviewed in Reference 187) with the work         polypeptides of relatively low molecular weight
of a young scientist named Thorburn Brailsford         (6000 to 30,000 Daltons). Specific receptors (high
Robertson, who came from the University of             affinity, saturable) have been identified for many
Adelaide in 1905 to study with Jacob Loeb at the       of them and these receptors are usually cell sur-
University of California. Loeb was studying the        face receptors, a number of which have been
effects of chemicals and salt solutions on cellular    shown to possess endogenous protein kinase ac-
motility of Paramecium. This work stimulated           tivity (e.g., insulin, IGFs, EGF, PDGF, FGFs,
Robertson’s interest in how biological functions       and TGF-a). Some of these receptors undergo
can be modified by chemical substances and this         receptor-mediated endocytosis, which may be in-
in turn led to his interest in regulation of growth    volved in down-regulation of receptor-mediated
of developing organisms. In a series of papers         action, transference of a receptor-mediated signal
published in 1916 in The Journal of Biological         to the cell nucleus, or both. New growth factors
Chemistry, Robertson described the growth              continue to be discovered; only some of the more
stimulatory effects of an anterior pituitary ex-       well-characterized ones will be discussed here.
tract on the growth of juvenile mice from 20           The almost ubiquitous presence of growth factors
to 60 weeks of age. The active component was           in a wide variety of tissues leads one to specu-
extracted from desiccated pituitary glands with        late that each tissue may have its own growth-
alcohol and precipitated in ether. He called this      modulating substances. During development and
substance ‘‘tethelin,’’ from the Greek word for        tissue differentiation, these substances probably
‘‘growing.’’ Interest in tethelin as a pharmaceu-      act locally, either as paracrine- or autocrine-
tical apparently stimulated production of it by a      regulatory chemical messengers. They continue
drug manufacturer in Philadelphia, H. L. Mul-          to act as needed during adult life for tissue re-
ford’s (the first biotech company!). There ap-          newal and wound repair. In cancerous tissue, the
pears to be no record of its use in humans.            secretion of certain of these chemical messengers
   Other early work on ‘‘growth factors’’ included     becomes unregulated and cellular proliferation
that of Alexis Carrel at the Rockefeller Institute     continues unabated.
and Eric Horning in Melbourne. In 1928, Carrel            Some of the factors discussed here are pri-
reported that an unstable protein extracted from       marily of historical interest, but because they
embryonic tissue or lymphocytes stimulated             shed light on how the whole field developed,
growth of animal tissues in culture dishes. He         they are included. For example, the discovery
called these ‘‘trephones,’’ but did not characterize   that insulin can be a growth-promoting sub-
them further. About the same time, Horning,            stance for cultured cells and that there are
working with J. M. Byrne and K. C. Richardson,         insulin-like substances in human plasma that act
observed that cellular outgrowth from pieces of a      similarly led to the discovery of the insulin-like
mouse sarcoma was stimulated by extracts from          growth factors (IGF-1 and IGF-2). The discov-
embryonic or tumor tissue and that chick embryo        ery in 1948 of a peptide factor (NGF) that
intestinal fragments exhibited enhanced cellular       stimulated nerve outgrowth in chick embryo
proliferation in the presence of embryonic ex-         limb buds was the initial finding that led inves-
tract plus tethelin, which they obtained from          tigators to look for more such factors and to
Robertson. Unfortunately, the research trail on        the serendipitous discovery of epidermal growth
these growth-stimulating substances grew cold          factor (EGF). This discovery, in turn, led indi-
soon after these reports, with the death of Ro-        rectly to the discovery of the transforming
bertson in 1930 at age 46 and the subsequent lack      growth factors (TGF-a and -b). Similarly, bio-
of encouragement of Horning for this work. It          chemical characterization of the EGF receptor
was not until 20 years later that the work of Levi-    and elucidation of its amino acid sequence led to
Montalcini, Cohen, Hamburger, and their col-           the discovery that certain cellular oncogenes
laborators established this field on a firm footing,     code for growth factor receptors or parts thereof
with the biochemical characterization of growth        (see Chapter 5). Such is the wonderful and
factors as definitive entities.                         unpredictable nature of science.
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                              161

Soon after insulin was isolated it was found to
help support growth of cells in culture.188 The
idea that insulin might be a growth-stimulatory
factor was supported by the observation that
many of the mitogenic peptides derived from
blood have an insulin-like activity. Whether in-
sulin has a physiologically important mitogenic
activity in vivo, however, is questionable, be-
cause supraphysiologic concentrations of insulin
are usually needed to stimulate cell proliferation
and the mitogenic effect is usually small com-
pared with that of total serum or other mito-
genic peptides.189 Insulin, however, frequently
acts synergistically with other growth factors,
probably because it is required for optimal up-
take and utilization of needed nutrients.
   Characterization of the cell surface receptor
for insulin has led to a broader understanding of
how insulin and insulinlike factors work. The
insulin receptor is a tetrameric disulfide-linked       Figure 4–13. Transmembrane tyrosine kinases. Struc-
complex containing two a subunits of 125,000           tural features of various receptor tyrosine kinase re-
MW and two b subunits of 90,000 MW. It has             ceptors are shown. Each receptor family is designated
been shown that the insulin receptor has in-           by a prototype ligand. Growth factors known to bind
trinsic protein kinase activity that autophos-         to receptors of a given family are listed above, and
                                                       receptors that constitute each family are listed below.
phorylates its b subunit as well as certain other      Boxes denote those growth factors or receptors whose
substrates on tyrosine residues.190–192 Phosphor-      genes were initially identified as activated oncogenes.
ylation of the insulin receptor is activated by        The c-onc designation is used to specify cellular ho-
binding the ligand and it increases the kinase         mologs of retroviral oncogenes. Open circles illustrate
activity of the receptor for other substrates. As we   immunoglobulin-likerepeats.Filledboxesindicatecon-
                                                       served tyrosine kinase domains. (From Aaronson,193
will see, this kind of ligand–receptor interaction     with permission.)
leading to receptor dimerization and activation of
an intrinsic receptor kinase activity is a common
                                                       somatomedins, and multiplication-stimulating
feature of polypeptide growth factors (Fig. 4–13).
Tyrosine kinase receptors interact with a number
                                                          A group of polypeptides present in plasma
of signal transduction pathways. This will be
                                                       that have insulin-like activity but could not be
discussed in more detail later in this chapter. The
                                                       neutralized by antibodies directed against in-
general structure of growth factor receptors and
                                                       sulin, called the nonsuppressible insulin-like
some of the ligands that activate them are shown
                                                       activities (NSILA), were separated into two
in Figure 4–13. A general scheme for how a
                                                       fractions: NSILA-P, which was precipitable by
growth factor binding to its receptor can trigger a
                                                       ethanol at low pH, and NSILA-S, which was
signal transduction pathway leading to gene ac-
                                    194                soluble under these conditions. Two peptides of
tivation is shown in Figure 4–14.
                                                       about 7000 MW with mitogenic activity were
                                                       isolated from the NSILA-S fraction.195 These
Insulin-like Growth Factors
                                                       have now been called insulin-like growth factors
The discovery of IGFs came about from a vari-          1 and 2 (IGF-1 and IGF-2).
ety of approaches but initially from the identifi-         A number of years ago it was realized that the
cation of insulin-like activities in plasma or se-     growth-promoting activity of pituitary growth
rum. At first these activities had various names        hormone was mediated through factors present
such as nonsuppressible insulin-like activities,       in the serum. These factors, termed somatome-
162                                                                                                        CANCER BIOLOGY



                                                         SH2 Shc PTB

                                            P            P                  SH3           Ras
                                                                SH2 Grb2        SOS
                               SH2                                                                                Cell membrane
                                           P             P
                                            P            P

                                 SH2                                   P   MEK                  MEK
                              PLC-        Ca2+
                                                                           MAP                  MAP   P
                                        PI1,4,5P3                                                     P
             PI4P      PI4,5P2
            PI3,4P2    PI3,4,5P3

                                                               myc     jun + fos       AP-1



                      Figure 4–14. Schematic representation of the cell signaling pathway mediated
                      by growth factor (GF) receptors with intrinsic tyrosine kinase activity. AP-1,
                      activator protein 1; DAG, diacylglycerol; MAP, mitogen-activated protein ki-
                      nase; MEK, MAP kinase kinase; P, phosphated region; PLC-g, phospholipase
                      c-g; PI3P, phosphatidylinositol(3)-phosphate; PI4P, phosphatidylinositol(4)-
                      phosphate; PI3,4P2, phosphatidylinositol(3,4,)-bisphosphate; PI4,5P2, phospha-
                      tidylinositol(4,5)-bisphosphate; PI1,4,5P3, phosphatidylinositol(1,4,5)-triphos-
                      phate; PI3,4,5P3, phosphatidylinositol(3,4,5)-triphosphate; PKC, protein kinase
                      C; PLC-g, phospholipase C-g; PTB, PTB domain; SH2, SH2 domain; SH3,
                      SH3 domain. (From Favori and DeCupis,194 reprinted with permission from the
                      American Society for Pharmacology and Experimental Therapeutics [ASPET].)

dins, which have a wide variety of stimulatory                               ethanol precipitation step. Further purification
actions, including stimulation of sulfate uptake                             produced a fraction that was 6000 to 8000 times
into cartilage and cartilage growth, insulin-                                more active than whole calf serum in stimulating
like activity on muscle and adipose tissue, and                              replication of chick embryo fibroblasts. The
proliferation-promoting activity for cultured                                purification fraction, which they called multi-
cells, have been isolated. Somatomedin C was                                 plication-stimulating activity (MSA), had a min-
found to be a basic polypeptide, of about 7000                               imal amount of insulin-like activity.
MW, that stimulates DNA synthesis in cartilage                                  It is now known that the insulin-like growth-
and cultured fibroblasts. Its activity circulates in                          promoting activities just described are all related
plasma bound to carrier proteins, which may                                  and belong to the IGF family of closely related
provide a ‘‘reservoir’’ for the polypeptide and                              mitogenic polypeptides. These polypeptides can
slow its removal from the blood.                                             be placed into two groups based on their iso-
   Dulak and Temin196 isolated a small poly-                                 electric points. The basic group has isoelectric
peptide from serum using the same initial steps                              points above pH 7.5 and includes IGF-1, so-
as for NSILA purification, but avoiding the acid-                             matomedin C, and basic somatomedin. All of
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                          163

these are now known to be identical. Moreover,        synthetic retinoid fenretinide has also been
IGF-2 and MSA represent the same activity and         shown to lower plasma levels of IGF-1 in breast
belong to the neutral-acidic group of insulin-like    cancer patients, Providing a possible rationale
activities with isoelectric points below pH 7.5.      for combination therapy with a retinoid and ta-
The circulating levels of IGF-1 depend on pi-         moxifen.203
tuitary growth hormone and IGF-1 mediates its            IGF-1 and IGF-2 both have mitogenic and
action, whereas plasma levels of IGF-2 do not         anti-apoptotic actions and they both have effects
appear to be so regulated. Because of the high        on the regulation of proliferation and differen-
levels of IGF-2 in fetal serum it is thought to be    tiation in developing tissues. IGF-1 expression
an important fetal growth factor.197 IGFs are         but not IGF-2 expression is turned on by growth
thought to be generated in the liver, but other       hormone. Expression of IGFs is also affected by
cells such as cultured fibroblasts and fetal           a variety of hormones including estrogens, ad-
tissues can also produce them. In addition to         renocorticotropic hormone (ACTH), thyrotro-
growth hormone, such factors as nutritional           pin (TSH), luteinizing hormone (LH), follicle-
status and circulating insulin levels can regulate    stimulating hormone (FSH), and human chor-
IGF production.                                       ionic gonadotropin (hCG), as well as by growth
   The IGF-1 receptor is structurally similar to      factors EGF, FGF, and PDGF. Most of these
the insulin receptor, also having a tetrameric        hormones and growth factors stimulate expres-
subunit structure with two a subunits of 125,000      sion of the receptor IGF-1R (reviewed in Ref-
MW and two b subunits of 90,000 MW and                erence 204). IGF has anabolic actions on pro-
possessing tyrosine kinase activity, whereas the      tein and carbohydrate metabolism by increasing
IGF-2 receptor is a single transmembrane chain        uptake of amino acids and glucose and by stim-
with a small intracellular domain, lacking tyro-      ulating glycogen and protein synthesis. IGF-1’s
sine kinase activity but possessing mannose           mitogenic actions include stimulation of cyclin
6-phosphate receptor activity (reviewed in Ref-       D1 expression, acceleration of G1 to S cell cycle
erence 198).                                          progression, and inhibition of apoptosis by stim-
   There is considerable cross-reactivity be-         ulating Bcl-2 expression and suppressing ex-
tween the various ligands and receptors in the        pression of Bax.
insulin and IGF family. For example, insulin can         Binding of IGFs to IGF-1R activates the re-
bind to both insulin and IGF-1 receptor, and          ceptor’s tyrosine kinase activity and this in turn
IGF-1 and IGF-2 can bind to the insulin re-           activates two signal transduction pathways: the
ceptor.199                                            ras-Raf-Mek-Erk pathway and the phospho-
   IGF-1 and IGF-2 as well as their receptors         inositol-3-kinase (PI3K) pathway. IGF-2R, by
are present in a wide variety of human cancer         contrast, has no tyrosine kinase activity, binds
types,200 thus paracrine or autocrine growth fac-     only IGF-2, and appears to act like an antagonist
tor stimulation of proliferation of tumor cells by    by facilitating IGF-2 degradation.
IGFs may contribute to their malignant pheno-            Interestingly, IGF-1 levels appear to mediate
type. IGF-1 and IGF-2 receptors are expressed         the caloric restriction effect on tumor growth in
on breast cancer cells, and IGF-1 and IGF-2           that the effect of caloric restriction on growth of
both have mitogenic effects on these cells.201        human prostate cancer xenografts in rats and
The growth-promoting effect of estrogen on            growth of bladder cancer in mice correlates with
breast cancer cells has been postulated to            a decrease in circulating IGF-1 and is reversed
be mediated by IGF-1. Interestingly, the anti-        by IGF-1 (reviewed in Reference 204). Fasting
estrogen drug tamoxifen has been reported to          has been shown to decrease circulating IGF-1 in
lower IGF-1 serum levels in breast cancer pa-         human subjects. A 50% reduction in caloric in-
tients and to inhibit IGF-1 gene expression in        take or a 30% lowering of protein intake pro-
the liver and lungs of rats bearing DMBA-             duces a decline in serum IGF-1 levels.
induced mammary tumors,202 a finding suggest-             Seven IGF-binding proteins (IGFBPs) that
ing that part of tamoxifen antitumor action in        modulate availability and function of the IGFs
breast cancer patients may be due to its ability to   have been identified.194 IGFBPs have complex
reduce production of the IGF-1. Moreover, the         actions and can either enhance or inhibit actions
164                                                                                CANCER BIOLOGY

of IGFs, depending on the specific IGFBPs             chick embryo limb bud regions. Subsequently, a
bound and the cellular context. The IGFBPs can       pronounced sensory innervation of the tumor
act by (1) transporting IGFs in the bloodstream,     tissue was observed.206 Further experiments
(2) protecting them from degradation, and (3)        showed that the nerve growth–promoting ef-
decreasing the availability of IGFs for IGF-1R,      fects of the sarcomas were most pronounced for
since IGFBPs generally have higher affinity for       neurons of the sympathetic nervous system and
IGFs than IGF-1R. Thus, for example, IGFBP-          the effector was a soluble protein.207 Nerve
3 binds IGF-1 and inhibits its mitogenic and         growth factor, which has been purified from
anti-apoptotic effects. More than 90% of the         various snake venoms208 and from mouse
IGFs in the blood are bound to IGFBPs, pre-          submaxillary gland,209,210 consists of two sub-
dominantly to IGFBP-3.                               units of about 13,000 MW, each with three in-
   IGF-1 acts synergistically with other mito-       trachain disulfide bonds. When NGF is isolated
genic growth factors in stimulating cancer cell      from submaxillary gland, it is bound in a 7S
proliferation in culture and presumably in vivo      protein complex of about 140,000 MW. It has
(reviewed in Reference 204). For example, in         several structural and functional similarities to
breast cancer cells, estrogens induce expres-        insulin.211 There is significant amino acid se-
sion of IGF-1 and IGF-1R. IGF-1 can also in-         quence homology with insulin, and most of the
crease expression of the estrogen receptor (ER)      identical amino acid residues being clustered in
in breast cancer cells. Interestingly, the ER        regions of NGF that align with the insulin A and
antagonist–agonist tamoxifen inhibits the effects    B chain segments of proinsulin but separated by
of IGF-1 in breast cells where it is antiprolifer-   the 35 residues needed for the C activation
ative and enhances IGF-1 effects in the uterine      peptide of proinsulin. There are an additional 37
endometrium where it is mitogenic and carci-         amino acid residues at the carboxyl-terminal
nogenic.                                             end of NGF that extend beyond the sequences
   Although the clinical and epidemiological         of its homology with proinsulin, but these resi-
data on the correlation of circulating IGFs and      dues are similar to the insulin B chain, sug-
IGFBPs and on cellular IGF-1R levels are             gesting a gene duplication event. Both insulin
somewhat contradictory, a number of studies          and NGF evoke similar biologic responses in
have reported an increased risk of solid tumors      their respective target tissues, including in-
in association with circulating levels of IGF-1      creased uptake of glucose and nucleosides and
and decreased risk with high circulating levels of   increased RNA, protein, and lipid synthesis.
IGFBP-3. However, clinical data do not con-          Thus, NGF appears to be a product of a gene
sistently support an association between IGF-1       that evolved from the same or a similar ancestral
levels and age, menopausal status, tumor size,       gene as proinsulin. In addition to its insulin-like
lymph node involvement, or tumor grade in            actions, NGF increases nerve fiber outgrowth
breast cancer patients (reviewed in Reference        and induces specific enzymes involved in sym-
204). Nevertheless, converging results from epi-     pathetic nervous transmission, for example, tyro-
demiological data and in vivo carcinogenesis         sine hydroxylase and dopamine b-hydroxylase,
models support the idea that high levels of cir-     in developing nerve cells.
culating IGF-1 are associated with the risk of a        But these are only some of the cold, hard
number of common tumors such as colorectal,          scientific facts. The story of how this all came
prostate, breast, and lung cancers.205               about is much more interesting.212 One of the
                                                     co-discoverers of NGF, Rita Levi-Montalcini,
                                                     was born in Turin, Italy, where she attended
Nerve Growth Factor
                                                     medical school and became board certified in
Nerve growth factor (NGF) was discovered as a        neurology and psychiatry. She became inter-
result of some experiments designed to test the      ested in the regulation of neuronal development
effects of rapidly growing tissue on nerve de-       and began studying development of the nervous
velopment in the limb bud regions of chick           system in chick embryos in Giuseppe Levi’s
embryos. In these experiments, two mouse sar-        laboratory in Turin. Unfortunately, both she and
comas, S-37 and S-180, were transplanted to          Levi were barred from academic life in 1939 as
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                            165

a result of Mussolini’s anti-Semitic ‘‘Manifesto      NT-3, NT-4, and NT-5 are growth factors for
delle razze.’’ Determined to continue her work,       sensory neurons, but have somewhat different
she set up a primitive lab in her basement and        effects on other nerve tissue cell types.
continued to do research even as the bombs fell          A high-affinity receptor for NGF has been
during World War II. In some of her early             identified as the 140 kDa glycoprotein gene
studies she found that fewer nerve cells grew         product of the trk-A protooncogene.214,215
into an area where the chick limb bud had been        This receptor, known as gp140trk-A, has tyrosine
eliminated, suggesting that limb end cells were       kinase activity and appears to act through the
releasing some trophic factor that stimulates         Ras, Raf-1, MAP kinase signal transduction sys-
nerve cell growth. A paper published on this          tem.216 Two Trk-A homologs, Trk-B and Trk-C,
topic caught Viktor Hamburger’s eye and he            are receptors for other members of the neuro-
invited her to Washington University in St.           trophin family. Specificity for response of nerve
Louis in 1946 to continue her studies. Intrigued      tissue cells occurs because there is cell-type
by Bueker’s observation206 that mouse sarcomas        specificity for the Trk receptors. The transform-
transplanted to areas of the limb bud caused          ing version of Trk was first observed in colon
nerve cells to grow, she began to try to purify the   cancer cells as a truncated, chimeric protein
trophic factor produced by the sarcoma cells.         fused to tropomyosin and having tyrosine kinase
About that time, she began to collaborate with        activity.217 A second low-affinity receptor for
Stanley Cohen, a biochemist who was then at           NGF, called gp75,NGFR has also been identified
Washington University, on the purification.            and in some cell types it is involved in a crucial
Initially, they found that the growth stimulatory     way with gp140trk-A in producing a response, but
material contained both protein and nucleic           in other cells, gp140trk-A by itself is sufficient.213
acid, so they added snake venom, which is known          In a study of tissue from 80 untreated neu-
to contain a diesterase that breaks down nucleic      roblastomas, the absence of gp140trk-A mRNA
acids, to see if they could determine which           was associated with tumor progression.218 Thus
component contained the activity. What they           a lack of a response system for NGF may foster
found was that even their snake venom–only            unregulated proliferation of malignant nerve
controls had nerve growth activity. This led to       cell tissue.
the idea that salivary glands might be a source of
the active substance. This turned out to be the
                                                      Epidermal Growth Factor
case, and the material we now know as NGF was
ultimately purified from mouse salivary glands,        Epidermal growth factor (EGF) was discovered
which also proved to be a source for epidermal        during the course of some experiments on NGF
growth factor (EGF). In 1986, Levi-Montalcini         activity in submaxillary gland extracts.219 When
and Cohen won the Nobel Prize for their pio-          these extracts were injected into newborn ani-
neering studies on growth factors.                    mals, precocious eyelid opening and eruption of
   It is now known that there is a family of NGF-     incisor teeth were observed. These phenomena
like factors (reviewed in Reference 213), which       were caused by stimulation of epidermal growth
includes brain-derived neurotrophic factor            and by keratinization. The material responsible
(BDNF), and neurotrophins NT-3, NT-4, and             for these effects was isolated from mouse sub-
NT-5. The members of this family have about           maxillary gland and found to be a low-molecular-
60% sequence homology and their activity ap-          weight, heat-stable polypeptide,219 which has
pears to be limited to neuronal tissue. Various       been purified from mouse submaxillary gland
neurotrophins have growth-promoting activity          and human urine. A single polypeptide chain of
for various nervous tissues—for example, NGF          53 amino acids, it has three intramolecular disul-
has positive cell survival and differentiation ef-    fide bonds that are required for biologic activity.
fects on sensory, sympathetic, and cholinergic        The EGF activity isolated from submaxillary
neurons, as well as on PC12 pheochromocytoma          gland at neutral pH is bound to a carrier protein
cells; BDNF has positive effects on sensory,          to form a high-molecular-weight (74,000) com-
cholinergic, and dopaminergic neurons and on          plex that consists of two molecules of EGF (6000
retinal ganglion cells, but not PC12 cells; and       MW) and two molecules of binding protein
166                                                                                 CANCER BIOLOGY

(29,300 MW). The binding of EGF to its carrier        the three intramolecular disulfide bonds), bind
protein depends on the presence of the carboxyl-      with high affinity to EGF receptors, and produce
terminal arginine residue, and because the bind-      mitogenic responses in EGF-sensitive cells.224
ing protein has arginine esteropeptidase activity,    The role of the large, eight EGF sequence–
it has been suggested that biologically active        containing prepro-EGF isn’t clear, but in some
EGF is generated from a precursor protein by          cells it exists as a membrane protein and re-
the action of the carrier protein peptidase.220 A     tains EGF-like biological activity. Several other
similar but not identical carrier protein with ar-    membrane-associated, growth factor–containing
ginine esteropeptidase activity has been discov-      proteins have also been discovered. A number of
ered in association with NGF, and the carrier         these have multiple EGF-like repeats (see below
enzyme is also involved in formation of NGF           under TGF-a). In some cases, e.g., TGF-a,
from its precursor.221 The EGF activity isolated      cleavage of the repeats releases soluble growth
from human urine is very similar in structure and     factors. In other cases, the membrane-anchored
function to mouse EGF and has significant amino        growth factor–containing polyproteins may be
acid sequence homology with it, suggesting that       acting as ‘‘juxtacrine’’ growth factors in develop-
human and mouse EGF evolved from the same             mental processes that require cell–cell interac-
gene.222 It has also been found that the gastrin      tion. This function may not be brought about by
antisecretory hormone urogastrone in human            secreted, diffusible growth factors.
urine is, in fact, EGF; this finding suggests an          The finding that a cultured human epidermal
additional biologic function of this factor.223       carcinoma cell line, A431, has a large number of
   The mouse EGF gene has been cloned and is          specific EGF receptors has provided a model
surprisingly large compared with the gene size        system in which to study the receptor and the
required to code for the mature EGF polypep-          effects of EGF–receptor interaction. The study
tide (reviewed in Reference 224). Mature EGF          of EGF receptors has produced some precedent-
is only 53 amino acids long, yet the gene con-        setting results that relate to a variety of growth
tains sufficient information to encode about           factors.
1200 amino acids. Contained within this 1200             The EGF receptor (EGFR) is an intrinsic
amino acid sequence are eight EGF-like se-            membrane glycoprotein of about 170,000
quences, indicating that the gene codes for a         MW.225 It is monomeric, unlike the tetrameric
large prepro-EGF molecule that is then pro-           insulin and IGF-1 receptors, and contains a core
cessed to mature EGF and a family of EGF-like         polypeptide of 1186 amino acids and N-linked
polypeptides in a manner similar to the pro-          oligosaccharides, the latter of which make up
duction of ACTH from pro-opiomelanocortin.            about 25% of the molecular weight of the re-
   A wide variety of cells, including epidermal       ceptor. The receptor binds EGF with high
cells, corneal cells, fibroblasts, lens cells, glial   affinity (KD ¼ 10À9 to 10À10 M) and high specific-
cells, granulosa cells, vascular endothelial cells,   ity. Occupation of the receptor by EGF induces
and a large variety of human cancer cells (see        an autophosphorylation of the receptor as well
below), have specific cell surface receptors for       as phosphorylation of certain other cellular sub-
EGF, indicating the ubiquitous nature of the          strates (see below). The autophosphorylation site
target cells for this growth factor. Specific re-      is a tyrosine226 rather than a serine or threonine,
ceptors for EGF have even been found in liver         which was a surprising finding at the time be-
membrane fractions from certain fish. Thus,            cause the only other known tyrosine kinase was
EGF is a phylogenetically old protein.                the pp60src product of the src oncogene (see
   It is now realized that there is a family of       Chapter 5). This observation for the EGF re-
EGF-like growth factors that includes TGF-a,          ceptor came before similar findings for the insu-
amphiregulin (a growth factor first detected in        lin, IGF, and PDGF receptors and so was indeed
phorbol ester–stimulated MCF-7 cultured breast        precedent setting.
carcinoma cells), and pox virus growth factors. All      The extracellular domain of the EGFR binds
the members of the EGF family are 50–60 amino         EGF and EGF-like ligands with high affinity,
acids in length, have six half-cystines in the same   contains 10 to 11 N-linked oligosaccharides, and
register (indicating the structural importance of     has a high content of half-cystine residues that
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                                167

could form up to 25 disulfide bonds, providing          Table 4–5. Biologic Effects of Epidermal Growth
a complicated tertiary structure. The cytoplas-        Factor
mic domain contains the tyrosine kinase activity,      IN VIVO
four sites for tyrosine autophosphorylation, and       Accelerated proliferation and differentiation
several sites for serine/threonine phosphoryla-          Of skin tissue
tion that are substrates for phosphorylation by          Of corneal epithelial tissue
                                                         Of lung and trachea epithelia
non-receptor kinases such as protein kinase C,         Potentiation of methylcholanthrene carcinogenesis
suggesting cross talk between kinase systems           Inhibition of gastric acid secretion
and additional control mechanisms for recep-           Increased activity of ornithine decarboxylase and
                                                         accumulation of putrescine
tor activity. For example, PKC-mediated phos-          Formation of fatty liver
phorylation of threonine 654 attenuates EGF-           Increase of disulfide group content in skin
activated autophosphosylation. The binding of          Hepatic hypertrophy and hyperplasia
                                                       Potentiation of cleft palate
EGF to its receptor induces a dimerization that
is related to its functional activation. The Erb B     ORGAN CULTURES
oncoprotein is a truncated version of the EGF          Accelerated proliferation and differentiation
                                                         Of skin tissue
receptor and lacks most of the ligand-binding            Of corneal epithelial tissue
domain.                                                  Of mammary gland epithelial tissue
   An important biological regulator in animals,       Induction of ornithine decarboxylase and accumulation
                                                         of putrescine
EGF is capable of eliciting a wide variety of          Enhanced protein synthesis, RNA synthesis
physiologic and cellular responses (Table 4–5).        Inhibition of palate fusion
It is not clear, however, what its key biologic role   CELL CULTURES
is in vivo. In cell culture systems, EGF stimu-        Increased transport
lates cell proliferation of a wide variety of cell        Of a-aminoisobutyrate
types. A number of studies have shown that                Of deoxyglucose
                                                          Of Kþ
maximal stimulation of DNA synthesis in cul-           Activation of glycolysis
tured cells occurs at about 25% saturation of          Stimulation of macromolecular synthesis
EGF binding sites, indicating that only a frac-           Of hyalurionic acid
                                                          Of RNA
tion of available receptors need be occupied by           Of protein
EGF to trigger a mitogenic response. Moreover,            Of DNA
cells grown in the presence of EGF continue to         Enhanced cell multiplication
                                                       Alteration of membrane properties
proliferate after the cultures become confluent,        Stimulated hCG secretion
simulating the loss of density-dependent growth        Increased biogenesis of fibronectin
inhibition observed in transformed cells. This         Enhanced prostaglandin biosynthesis
                                                       Alteration of viral growth
finding suggests that continued exposure to the         Increased squame production by keratinocytes
growth-stimulating activity of EGF or EGF-like
                                                       From Carpenter and Cohen219
factors could be involved in the excessive pro-
liferation of cells and the loss of feedback
inhibition of growth seen in neoplasia. In this        in certain other human cancer cells (e.g., glio-
regard, it has been observed that certain trans-       blastomas), the gene coding for the EGF recep-
formed cell types produce their own EGF-like           tor is amplified several-fold,228 which explains
growth factor that appears to bind to EGF cell         why some cell types have high receptor content
surface receptors (see below).                         and why they may be so responsive to a prolif-
   In addition to human epidermoid carcinoma           erative signal provided by EGF-like growth fac-
cells, other human cancer cells have receptors         tors. Truncation and deletion mutants of EGFR
for EGF or EGF-like growth factors. Among              have also been observed in human cancers.227
these are ovarian, cervical, renal, lung, bladder,        Occupancy of EGF receptors by EGF trig-
and breast carcinomas, and glioblastomas (re-          gers a mitogenic response. The molecular signals
viewed in Reference 227), but this is not a uni-       involved in this triggering mechanism have been
versal finding; human leukemia and lymphoma             defined and a number of cellular events associ-
cells have low or absent EGF receptors. In hu-         ated with the EGF response have been iden-
man epidermoid A431 carcinoma cells as well as         tified. Stimulation of the mitogenic cascade
168                                                                                    CANCER BIOLOGY

involves activation of the Ras-Raf-MAP kinase           regulation of cellular responses to a number of
system (see section on Signal Transduction, be-         polypeptides, including EGF, insulin, transfer-
low). The cellular responses to EGF include             rin, a2-macroglobulin, immunoglobulins, and
both early and late events (reviewed in Refer-          low-density lipoprotein (LDL).232 In the case
ence 229 and 230). Within 5 minutes after EGF           of EGF binding to cultured human fibroblasts,
addition to cultures of responsive cells, increases     recovery of complete EGF binding capacity
in protein phosphorylation, ion fluxes, and mem-         requires several hours and depends on RNA and
brane ruffling occur. Internalization of bound           protein synthesis,233 indicating that resynthesis
EGF, increased amino acid and glucose trans-            of EGF receptors is required.
port, onset of RNA synthesis, and increased                A number of growth factors induce a rapid
protein synthesis occur, in that order, between         change in ion fluxes across cell membranes as an
15 and 60 minutes after initiation of EGF               early event. These events include increased
binding to its cell surface receptor. In contrast,      Naþ/Hþ exchange, increased levels of intracel-
induction of DNA synthesis requires exposure of         lular free Ca2þ, and a transient rise in intra-
cells to EGF for about 12 hours, and onset of cell      cellular pH.234–236 These events are commonly
division follows within a few hours after that.         induced changes in cells after exposure to a wide
   The relationship between the early and late          variety of mitogenic agents, and ultimately they
cellular responses is not clear. Presumably the         trigger the events leading to DNA synthesis via a
pleiotypic response to EGF, as well as to other         signal transduction cascade.
growth factors, follows some closely regulated             The role that EGF plays in vivo, particularly
temporal cascade of events. Since increased             in humans, is not totally clear. EGF has been
phosphorylation of membrane and other cellu-            isolated from human urine, as noted, and thus
lar proteins is one of the earliest events observed     presumably does have a physiologic effect in
after EGF binding, it is a likely candidate for the     adult tissues, probably as a paracrine growth
signaling event of the cascade.                         factor involved in tissue renewal and wound
   Internalization of EGF occurs in cells by            repair. EGF has been shown to be present in
means of a process called receptor-mediated             human milk, thus EGF may play a role in milk
endocytosis, which is common to a number                production and be a growth-promoting agent for
of polypeptide hormones and growth factors              the newborn.237 Because EGF is acid stable and
(Fig. 4–15).231 After a polypeptide binds to its        at least partially active after oral administration,
specific receptor, there is a clustering of ligand-      it may conceivably have such a function. In
bound receptor into patches on the cell surface.        addition, EGF has been reported to restore
These clusters are then invaginated into organ-         spermatogenesis in male mice whose subman-
elles called clathrin-coated pits (clathrin is a pro-   dibular glands (the major organ of EGF produc-
tein lining the cytoplasmic face of these vesicle-      tion in mice) have been removed, a finding sug-
like structures). The pits then pinch off from the      gesting a role for EGF in male reproductive
cell membrane and form true vesicles. These             function.238 High-molecular-weight forms of
vesicles become acidified and, under the acidic          EGF-like activities have been found in the urine
conditions, the ligand becomes dissociated from         of patients with disseminated cancer, and in a
the receptor. These vesicles are then shuttled to       number of breast cancer cases there is a corre-
lysosomes, where they fuse with the lysosomes,          lation with the presence of EGF-like material in
and the receptors are degraded in them. In some         the urine and advanced stage of the disease.239
cases (e.g., transferrin receptors), the endo-             The finding that the erb B oncogene encodes a
cytotic vesicles appear to become associated            protein that has significant sequence homology
with the Golgi complex and receptors may then           with the cytoplasmic, protein kinase–containing
be recycled to the cell surface. Because this           domain of the EGF receptor (see Chapter 5)
process in effect removes receptors from the            suggests that cancer cells may elevate their ability
cell surface, the cells become refractory to fur-       to respond to endogenously or exogenously pro-
ther stimulation by exogenous growth factor.            duced growth factors, which may be a key to their
This refractory period, called receptor down-           ability to grow autonomously with minimum
regulation, appears to be a common pathway for          regulatory control. Since this truncated receptor
Figure 4–15. Mechanisms involved in receptor-mediated endocytosis. The
pathway of receptors and ligands shown here was determined for galactose-
terminal glycoproteins but is thought to apply in the case of other ligands and
receptors as well. Ligand binds to receptors diffusely and then collects in
coated pits, which invaginate and are internalized as coated vesicles, whose
fusion gives rise to endosomes and then to a CURL (so designated because it
is a compartment of uncoupling of receptor and ligand). In the acidic CURL
environment, ligand is dissociated from receptors. Ligand accumulates in the
vesicular lumen of the CURL, and the receptors are concentrated in the
membrane of an attached tubular structure, which then becomes separated
from the CURL. The vesicular part moves deeper into the cell and fuses with
a lysosome, to which it delivers the ligand for degradation. In some cases
(e.g., transferrin receptors), the membranous tubular structure is thought to
recycle receptors to the plasma membrane. This is not, however, the case for
EGF receptors. (From Dautry-Varsat and Lodish,231 with permission.)
170                                                                                 CANCER BIOLOGY

protein lacks the ligand-binding domain, it           and others; (2) in many Erb B2–overexpressing
may not be as subject to shut-off by the down-        breast cancers, Erb B3 contains high levels of
regulation mechanisms normally triggered by           phosphotyrosine (indicating a high level of
growth factor binding and thus may be consti-         receptor activation); (3) Erb B2 inactivation de-
tutively active as a protein kinase.                  creases Erb B3 phosphorylation; and (4) inac-
   EGF family members play an important role          tivation of Erb B3 inhibits proliferation of breast
in normal development. As will be described in        cancer cells as efficiently as blockade of Erb B2.
more detail below, the EGF family of growth           These results indicate that Erb B2/Erb B3 di-
factors activate one or more of four receptors:       mers act as an oncogenic ‘‘team’’ to stimulate
Erb B1 (EGF receptor or EGFR), Erb B2 (also           cancer cell proliferation.
called HER-2/neu), Erb B3, and Erb B4.                   Mutant forms of the epidermal growth factor
Huotari et al.240 showed that Erb B ligands           receptor also occur. EGFR (Erb B1) is ampli-
EGF, TGF-a, heparin-binding EGF, beta-                fied in human glioblastomas. In addition, the
cellulin, and neuregulin-4 are detected in the        EGFR gene is mutated in about 50% of these
developing pancreas of mouse embryos by day           cancers and is associated with a poorer prog-
13 of gestation. Some differential effects of the     nosis (reviewed in Reference 242). The most
EGF family ligands were seen in organ cultures        common mutation is deletion of exons 2
from mouse embryos, some of which were ob-            through 7, producing a truncated EGFR, called
tained from EGFR (À/À) gene knockout ani-             EGFRvIII, lacking most of the extracellular
mals. Overall, their results ‘‘suggest that ligands   domain. EGFRvIII doesn’t bind EGF but has a
of the Erb B1 and Erb B4 receptors regulate           ligand-independent tyrosine kinase activity that
the lineage determination of islet cells during       is constitutively turned on. It also appears to use
pancreatic development.’’ These data as well as       a different set of downstream signaling pathways
several other lines of evidence (reviewed in          compared to unmutated EGFR (reviewed in
Reference 240) indicate a role for EGF family         Reference 242). In mouse models of glioblas-
ligands receptors in pancreatic islet cell differ-    toma, EGFRvIII carcinogenic activity requires
entation.                                             mutations at the INK4a/ARF tumor repressor
   Binding of EGF family ligands to Erb B re-         locus. Recall that INK4a is a repressor of cyclin-
ceptors’ extracellular domain triggers formation      dependent kinase 4 (Cdk4) and that ARF reg-
of homo- and heterodimers, which activates            ulates p53 (see Cell Cycle, above). These
the intrinsic-tyrosine kinases of these receptors.    defects are also likely to be involved in hu-
Erb B2 appears to be a co-receptor and the            man glioblastoma because the EGFRvIII and
preferred partner for other ligand bound Erb B        INK4a/ARF mutations are frequently seen in
receptors (reviewed in Reference 241). The            human glioblastomas.
importance of Erb B2 heterodimers in devel-              EGFRvIII is also detected in cancers of the
opment is demonstrated in Erb B2 knockout             breast, ovary, and lung, and in medulloblasto-
mice in which loss of Erb B2 produces defects in      mas, but not in normal adult tissues. Tang
neuronal and cardiac development.                     et al.243 have shown that overexpression of
   Erb B2 (Her-2/neu) overexpression (often           EGFRvIII transforms a nontumorigenic hema-
by gene amplification) occurs in 25%–30% of            topoietic cell line into one with a highly tu-
breast cancers and correlates with lower sur-         morigenic phenotype in athymic mice and that
vival. Inhibition of Erb B2 in cell culture blocks    this transforming ability of EGFRvIII is dose
breast cancer cell proliferation. These and sim-      dependent. In addition, they showed that ex-
ilar observations have led to the development of      pression of a low level of EGFRvIII in the hu-
a successful treatment for Her-2/neu over-            man breast cancer cell line MCF-7 significantly
expressing breast cancers (see below). However,       enhances the tumorigenicity of these cells in
Erb B2 (Her-2/neu) action in stimulating cell         athymic mice, suggesting that the mutant EGFR
proliferation depends on its interaction with Erb     gene could play a pivotal role in breast cancer
B3. Several lines of evidence point to this:241 (1)   progression to a more aggressive tumor.
co-overexpression of Erb B2 and B3 is seen in            The now well-known Herceptin treatment for
many human cancers, including breast, bladder,        breast cancer is the first and so far the best ex-
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                           171

ample of targeting the EGFR family in human            whole serum as a growth promoter for some cell
cancers.244 This is a great story in bench-to-         lines (e.g., 3T3) but not for others (e.g., human
bedside translational research. As noted above,        diploid fibroblasts). In the latter case, serum is
25%–30% of breast cancers here amplified                still required for maximal cell proliferation. Al-
Her-2/neu (Erb B2), and this is seen in the            though FGF is mitogenic for several cell types,
breast cancer cells themselves. Preclinical mod-       it does not affect cells transformed with polyoma
els showed that HER-2 amplification plays a             or SV40 virus,189 presumably because these
direct role in transformation of cultured cells        transformed cells have a lower growth factor
and in mammary carcinogenesis in mice and              requirement.
that the Her-2 pathway promotes hormone-                  Originally, two FGF-like polypeptides were
independent growth of human breast cancer              purified to homogeneity and characterized:247
cells in culture (reviewed in Reference 244). It       the FGF from bovine pituitary, called basic
was also observed that women with breast can-          FGF, and an acidic FGF from bovine brain.
cers that overexpressed Her-2 had a more ag-           Basic FGF has also been isolated from a variety
gressive form of the disease and poorer overall        of other tissues, including adrenal gland, corpus
survival. A humanized monoclonal antibody to           luteum, retina, and kidney. There is significant
Her-2 was shown to inhibit proliferation of Her-       amino acid sequence homology between the
2 overexpressing human breast cancer cells in          two polypeptides, and they are potent mitogens
culture and in mouse xenografts, and to be             for diverse cell types including capillary endo-
synergistic with chemotherapeutic drugs in tu-         thelial, vascular smooth muscle, adrenocortical,
mor inhibition in preclinical models. These re-        bone, and ovarian granulosa cells.
sults led to phase I and phase II clinical trials in      It is now clear that the FGFs constitute a
which this antibody, called trastuzumab and la-        large family of growth and differentiation factors
ter Herceptin, was safe and effective in women         for cells of mesenchymal and neuroectodermal
with Her-2-positive metastatic disease who had         origin. It includes the two original members
relapsed after chemotherapy. A phase III trial         acidic FGF (also called FGF1), basic FGF
showed that ‘‘trastuzumab increased the clinical       (FGF2), as well as at least seven other current
benefit of first-line chemotherapy in metastatic         members (reviewed in References 248 and 249).
breast cancer that overexpresses Her-2.’’244           FGF3 (also known as Int-2) is a gene product
   This success has led to other attempts to block     first observed in mouse mammary tumor virus
one or more members of the EGFR family of              (MMTV)-transformed cells and is a 239–amino
receptors. These include monoclonal antibodies         acid protein with 44% homology to FGF2.
such as cetuximab (Erbitux) and small-molecule         FGFs -4, -5, and -6 were found by screening
EGFR inhibitor such as gefitinib (Iressa). Clinical     tumors for oncogenes that could transform NIH
trials have shown some responses to cetuximab in       3T3 cells. FGF4 was also isolated from Kaposi’s
colon cancer and to gefitinib in non–small cell         sarcoma and is called K-FGF. FGF7 (KGF) is a
lung carcinomas. Other monoclonal antibodies           potent mitogen for keratinocytes and other epi-
and small molecules that target EGFR are in            thelial cells and differs from the other FGFs in
development (reviewed in Reference 245).               that it does not stimulate the proliferation of
                                                       fibroblasts or endothelial cells. It is expressed in
                                                       a restricted number of adult tissues, including
Fibroblast Growth Factor
                                                       kidney, colon, and ileum, but not in brain or
Using initiation of DNA synthesis in mouse 3T3         lung as the others are. FGF8 is an androgen-
fibroblasts to monitor purification, Gospodar-           induced growth factor cloned from mouse
owicz246 isolated from bovine pituitary a mito-        mammary carcinoma cells. FGF9, also called
genic polypeptide, which he called fibroblast           glia-activating factor (GAF), was purified from a
growth factor (FGF). Distinct from the IGFs,           human glioma cell line.250
FGF is a potent mitogen for 3T3 cells and other           The FGF family members have 30% to 70%
cultured fibroblasts; however, by itself it is only     amino acid sequence identity, with the greatest
about 30% as potent as whole serum. A combi-           identity in a ‘‘core’’ region represented by most
nation of FGF and glucocorticoids will replace         of the sequence of FGF2, which is considered
172                                                                                 CANCER BIOLOGY

the prototype FGF. All of the FGFs have the           murine and human bek product (also called K-
ability to bind heparan sulfate, which is impor-      Sam) and chicken cek3; FGFR3 is murine flg-2
tant to their biological function (see below).        and chicken cek2; FGFR4 and FGFR5 have
Except for FGF1 and FGF2, all the FGFs have           no other names. As noted above, some FGFRs
an N-terminal signal sequence and are secreted        arise by alternate mRNA splicing. For example,
from cells. Yet FGF1 and FGF2 are abundant in         FGFR2 and the receptor for KGF (KFGR) are
the ECM, indicating that they must be exported        both derived from the same gene (bek) by al-
from cells. Binding to the ECM appears to             ternate splicing of one exon.
provide a reservoir for FGFs, which can then             The FGFRs have several features in common:
be mobilized in response to requirements for          (1) an extracellular ligand-binding domain with
wound healing, angiogenesis, etc.                     three disulfide bonded loops and an immuno-
   FGFs play an important role in embryonic           globulin-like structure; (2) a single transmem-
development (reviewed in Reference 249). For          brane domain, followed by a relatively long (80
example, FGF1 and FGF2 are produced in                amino acids) juxtamembrane domain; (3) two
early mesoderm and appear to be involved in           tyrosine kinase domains separated by a 14–amino
mesoderm induction. Members of the FGF                acid insert; and (4) a carboxyl-terminal tail of
family appear to regulate differentiation of a        about 50 amino acids, containing tyrosines that
variety of cell types during development and          may be autophosphorylated upon ligand binding.
they are expressed in a temporally and spatially         There is significant sequence homology
regulated way. FGF4 and FGF5 are expressed            among these receptors and overlap in their
in embryonic muscle cell precursors in the            binding specificity for various FGFs. For in-
mouse. FGF1, -2, and -5 are expressed in ner-         stance, one isoform of FGFR1 binds FGF1 and
vous system tissue. FGF4 is the first FGF de-          FGF2 with slightly different affinities. An iso-
tected in early mouse development, as early           form of FGFR2 binds FGF1, -2, and -4 but not
as the four-cell stage, and continues at least        FGF5 or -7, whereas an FGFR2 splice variant
through the early formation of tissue layers, i.e.,   binds FGF1 and -7, but not FGF2. FGFR3 can
mesoderm, endoderm, and ectoderm, and is              bind FGF1, -2, -4, and -5. FGFR4 binds FGF1
later expressed in a tissue-specific manner. FGF3      with high affinity and FGF4 and FGF5 with 10-
and FGF5 are expressed prior to gastrulation          fold lower affinity. This functional redundancy
and appear to be restricted to parietal and vis-      raises questions about what provides the speci-
ceral endoderm, respectively. FGF3, -4, and -5        ficity for response to the FGFs and why differ-
mRNAs are detectable during mesoderm for-             ent receptors are needed to bind the same li-
mation, but in distinct spatial orientations,         gand. It seems likely that the tissue distribution
which suggests a specific role for each FGF            of the FGFRs and the local production of spe-
in development of specific tissues. FGF1 and           cific arrays of FGFs provide some selectivity of
FGF2 are expressed as early as day 10 1/2 in          response in various developing tissues. For ex-
mouse embryos and are detectable in a variety of      ample, FGFR1 is highly expressed in develop-
tissues by day 13 1/2. FGF6 is observed during        ing brain, skin, and growth plates of bones,
middle and late gestation, but peak levels are        whereas FGFR2 is highly expressed in choroid
observed on day 15.                                   plexus, skin, lung, kidney, and brain temporal
   Five different FGF receptors (FGFRs) have          lobe. FGFR3 is abundant in the intestine, lung,
been cloned and more may be found (reviewed           kidney, and bone growth plates. FGFR4 is
in References 248 and 249). Isoforms of FGFRs         highly evident in adrenal gland, lung, kidney,
are generated by alternate mRNA splicing.             and liver. Even though FGFR1 and FGFR2
FGFRs belong to the tyrosine kinase receptor          have similar ligand-binding specificities, FGFR1
family and some of them were originally iden-         is seen predominantly in mesenchyme of limb
tified by cloning of tyrosine kinases, e.g., the       buds and somites, whereas FGFR2 is in highest
hormone flg, chicken cek1, and mouse bek gene          abundance in epithelial cells of skin and de-
products. The receptors are now numbered              veloping internal organs. These different sites
FGFR 1 to 5. FGFR1 is the mammalian flg gene           suggest alternative and coordinate roles in these
product (same as chicken cek1); FGFR2 is the          two tissue layers as organs develop. FGFR4 is
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                         173

found in muscle cell precursors and may have a        melanocytes transforms them, although expres-
prominent role in muscle development.                 sion of bFGF by itself doesn’t make them tu-
   A feature shared by all FGFs is their high         morigenic in vivo.252 The ability of some FGFs to
affinity for binding to heparan sulfate and he-        induce blood vessel growth has been observed in
paran sulfate proteoglycans (HSPGs). A number         tumors (tumor angiogenesis).
of studies show an important role for heparan-           An FGF (K-FGF or FGF4) isolated from
sulfate-FGF complexes in binding of FGFs to           Kaposi’s sarcoma induces vascularization of this
their receptors. HSPG-like molecules are in-          tumor. K-FGF’s action is inhibited by a heparin
volved in presentation of FGF to the recep-           analog called pentosan polysulfate, which blocks
tor and in stabilization of high affinity FGF-         angiogenesis in a Kaposi’s sarcoma growing in
receptor complexes, which provides an ‘‘anchor’’      nude mice and induces tumor regression.253
for the interaction with cells. Binding of FGF        Basic FGF has been shown to be an autocrine
ligands to HSPGs may increase the half-life           growth factor for human and rat glioma cells,
of FGFs by limiting proteolytic degradation.          melanoma cells, and endometrial carcinoma
HSPGs also appear to act as a reservoir for           cells (reviewed in Reference 254). Injection of
FGFs by providing long-term storage sites.            an antibody to bFGF into tumor-bearing nude
HSPGs can be divided into cell surface forms          mice inhibited tumor growth, in part because the
(syndecans and glypicans) and secreted extra-         antibody blocked the potent angiogenic effects
cellular matrix forms (e.g., perlecan). Specific       of bFGF.254 Basic FGF is also found in areas of
HSPGs appear to act as co-receptors for FGF2          human squamous cell head and neck tumors
in stimulating tumor angiogenesis. For example,       with a high thymidine labeling index and signif-
glypican-1 is up-regulated in gliomas and this        icant endothelial cell proliferation, indicating a
enhances FGF2-induced angiogenesis in these           correlation between tumor vascularization, tu-
tumors.251                                            mor cell proliferation, and bFGF production.255
   FGFs have three possible signal transduction       Through immunostaining for protein and in situ
mechanisms: (1) ligand-initiated activation of re-    hybridization for mRNA, both acidic and basic
ceptor tyrosine kinase activity that leads to auto-   FGFs have been detected in about 60% of sur-
phosphorylation and phosphorylation of other          gical samples of human pancreatic cancer and
key cellular proteins such as Raf-1; (2) activation   their presence correlates with advanced tumor
of phospholipase C-g1 (PLC-g1) and the phos-          stage and poor prognosis.256 Presence of bFGF
phoinositol hydrolysis (PI) pathway leading to        in the cytoplasm of human renal cell carcinomas
mobilization of intracellular calcium and acti-       also correlates with shorter survival time.257
vation of protein kinase C (see Signal Trans-            High concentrations of FGF2 have been de-
duction Mechanisms below); and (3) internal           tected in the urine and serum of cancer patients,
localization of FGF in nuclei that could lead to      and high serum levels are associated with a poor
DNA binding and direct activation of gene             prognosis in small cell lung cancer.258 Since
transcription. There is evidence for all three        secreted FGF2 is a mitogen for endothelial cells
mechanisms. Interestingly, there are high-            and a potent inducer of angiogenesis in vivo, the
molecular-weight intracellular forms of FGF1          poor prognosis observed in patients with high
and FGF2 that don’t appear to get secreted and        circulating levels of FGF2 may reflect active
have nuclear localization sequences. Forms of         angiogenesis and increased tumor growth and
FGF1 and FGF2 have also been found in cell            aggressiveness. Elevated FGF2 serum levels
nuclei.                                               have also been observed in patients with head
   Various FGFs are produced by malignant tu-         and neck cancer, colorectal carcinoma, non-
mors and constitutive expression of some FGFs         Hodgkin’s lymphoma, and chronic lymphocytic
induces a transformed phenotype in cultured           leukemia.258
cells. There are several examples of this. Basic
FGF (bFGF, or FGF2) is not produced by normal
                                                      Platelet-Derived Growth Factor
melanocytes but is made constitutively by human
metastatic melanoma cells grown in culture, and       The discovery of platelet-derived growth factor
transfection of the bFGF gene into normal mouse       (PDGF) arose from an observation by Balk,259
174                                                                              CANCER BIOLOGY

who found that normal chicken embryo fibro-          be divided into early and late events. Early
blasts proliferated better in culture medium        events (after 1 to 10 minutes) include tyrosine-
supplemented with animal serum than in me-          specific phosphorylations, stimulation of phos-
dium containing platelet-poor plasma. He also       phatidylinositol turnover, and reorganization of
found that Rous sarcoma virus–transformed           actin filaments.264 Late events (after 30 to 180
cells grew equally well in both, and he specu-      minutes) include increased transcription of spe-
lated that a ‘‘wound hormone’’ was released into    cific genes, stimulation of IGF binding, and in-
serum during the process of clot formation.         creased amino acid transport. A clear difference
During the clotting of blood to form serum, a       between PDGF and EGF or IGF is that the
number of things happen, including the con-         stimulatory signal for cell division provided by
version of fibrinogen to fibrin and the clumping      PDGF is ‘‘locked in’’ after the cells are exposed
of platelets followed by platelet factor release.   to PDGF for only about 30 minutes, and PDGF
Thus, serum will contain some platelet-derived      can then be removed from the growth medium,
factors not present in plasma collected in the      whereas EGF and IGF must be present con-
presence of anticoagulants and then centrifuged     tinually to stimulate cell division. Moreover,
to remove the cellular elements and platelets.      PDGF by itself does not induce a mitogenic
Other investigators have confirmed the obser-        response, but requires other plasma factors,
vation that platelet-poor plasma is deficient in     among which are IGF and EGF. For example,
growth-promoting activity for cultured cells and    when growth-arrested 3T3 fibroblasts are trea-
have shown that platelet extracts could restore     ted with PDGF for a short period of time and
optimal growth to cells cultured in the presence    then transferred to platelet-poor plasma, the
of platelet-poor plasma. The growth-promoting       cells are observed to enter S phase 12 hours
factor from human platelets has been purified        after the addition of plasma. This 12-hour lag
and characterized.260–262                           before the onset of DNA synthesis occurs re-
   Four PDGF polypeptide chains have been           gardless of whether the plasma is added at the
identified, and these constitute the five dimeric     same time as PDGF or up to 13 hours after this.
isoforms of PDGF: PDGF-AA, -AB, -BB, -CC,           Thus, PDGF and plasma factors appear to
and -DD.263 These isoforms act by binding           control different events in the cell cycle. Scher
three types of PDGF receptors: aa, ab, and bb       et al.265 have postulated that PDGF induces
(Fig. 4–16). These receptors are tyrosine kinases   cells to become ‘‘competent’’ to enter the S
and have mitogenic effects on cancer cells, an-     phase of the cell cycle and that plasma factors
giogenic effects in both normal tissues (e.g.,      allow only competent cells to undergo ‘‘pro-
wound healing), and cancer tissues, and para-       gression’’ through the G1 phase to enter S phase.
crine effects in stromal fibroblasts and perivas-    Thus, PDGF is thought to prime cells to re-
cular cells. As shown in Figure 4–16, all the       spond to other growth factors in plasma. As
PDGF isoforms, except PDGF-DD, induce               noted earlier, the PDGF-induced competent
PDGF a-receptor dimerization and activa-            state must be stable for at least 13 hours after
tion. PDGF-BB and PDGF-DD activate recep-           PDGF is removed, since the addition of plasma
tors by dimerization. PDGF isoforms -AB, -BB,       at any time up to 13 hours will permit progres-
-CC, and -DD do the same for ab receptors.          sion into S phase.
The PDGF ligand binding-mediated receptor              Among the most potent factors in plasma that
subunit dimerization induces receptor auto-         stimulate progression are the insulin-like growth
phosphorylation and triggers a downstream cas-      factors. The evidence for this comes from ex-
cade of signal transduction involving Ras, PI-3     periments in which plasma from hypophysec-
kinase, phospholipase C-g, and protein kinase C     tomized rats was shown to be 20-fold less potent
as intermediaries.                                  in permitting PDGF-treated competent 3T3
   PDGF receptor activation results in activa-      cells to enter S phase than normal rat plasma.
tion of a number of genes involved in cell pro-     The addition of a low concentration (10À9M
liferation, including myc, fos, c-jun, and jun B.   is 0.000000001 molar) of IGF-1 to cultures
As with EGF, the cellular actions of PDGF can       of PDGF-treated cells grown in plasma from
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                               175

                                                 PDGF ligands
                                           AA CC       AB     BB    DD

                                                PDGF receptors

                   Autocrine stimulation       Paracrine stimulation of          Stimulation of
                      of cancer cells          stromal fibroblasts and           angiogenesis
                                                  perivascular cells

               Figure 4–16. The platelet-derived growth factor (PDGF) system is involved
               in multiple tumor-associated processes. Upper: PDGF ligand binding spec-
               ificity to PDGF receptors. Lower: PDGF receptors are expressed by many
               different cell types within tumors, and signaling from PDGF receptors can
               thus promote tumor progression in various ways. Tumor cells, purple; endo-
               thelial cells, red; fibroblasts, smooth muscle cells, and pericytes, green; extra-
               cellular matrix, dark solid lines. (From Pietras et al.,263 reprinted with permission
               from Elsevier.)

hypophysectomized rats allowed the cells to                plete the cell cycle; if they are delayed some-
enter S phase. Pure IGF-1 without plasma,                  where past this restriction point, they frequently
however, did not do this, indicating that other            die. The competence-initiating factors may in-
materials in plasma are also needed for pro-               duce the synthesis of a critical initiator protein,
gression to occur. By using this assay system, the         whereas progression-inducing agents may pro-
competence- and progression-stimulating ac-                mote cell division by stimulating the enzymes
tivities of a variety of growth factors have been          necessary for DNA synthesis. However, there
tested. Factors that have potent competence                appears to be more than one point at which
activity include PDGF, FGF, Ca2þ, and                      PDGF-treated cells can be arrested before they
‘‘wounding’’ (i.e., scraping a clear area through a        enter into S phase,265 which suggests a cascade
sheet of confluent cells). Progression-inducing             of events, each of which might have different
agents are IGF-1, IGF-2, insulin, and EGF. The             regulatory signals. The fact that a specific com-
mechanisms controlled by these two classes of              petence-inducing factor appears to be required
growth factors are not clear, but they probably            for the proliferation of fibroblast-like cells sug-
relate to the restriction point at which cells             gests that other tissues may have similar re-
make a commitment to enter S phase. Once this              quirements for competence activities. The less
commitment is made, cells enter S and com-                 specific progression-stimulating activities, such
176                                                                               CANCER BIOLOGY

as those of the IGFs, may be general growth-         was discovered that the PDGF B chain has a
promoting agents needed for the growth and           virtually identical amino acid sequence to the
development of many tissues in the body.             product of the sis oncogene first isolated from a
   Of interest is the fact that transformation of    sarcoma virus carried by the Woolly monkey
fibroblasts with SV40 virus circumvents the           (see Chapter 5). This cellular oncogene was
need for both competence and progression ac-         apparently picked up during evolution by the
tivities. The SV40-transformed 3T3 cells grow to     simian sarcoma virus and is part of the trans-
high density in either serum or platelet-poor        forming activity of the virus. The v-sis oncogene
plasma and require less serum than nontrans-         can activate a and b PDGF receptors and ini-
formed cells. Thus the requirement for PDGF          tiate cellular transformation when it binds in-
appears to be lost or greatly diminished dur-        tracellularly to PDGF receptors.272
ing the process of transformation. In addition,         The PDGF family of ligands and their re-
transformed human and mouse cell lines that          ceptors play an important role in normal de-
produce tumors in nude mice grow to high             velopment.263 For example, activation of the
density in platelet-poor plasma, whereas cell        PDGF-b receptor stimulates pericyte recruit-
lines that are not tumorigenic in nude mice grow     ment for blood capillary formation, develop-
poorly in platelet-poor plasma.266 These data        ment of vascular smooth muscle, and kidney
imply that transformed cells can produce a           development. Activation of PDGF-a receptors
PDGF-like growth factor that occupies PDGF           by PDGF-AA is needed for formation of lung
receptors and thus does not require exogenous        alveoli, hair follicle development, villus forma-
PDGF. It has now been shown that a wide              tion in the gut, and oligodendrocyte produc-
variety of murine and human cell lines trans-        tion in the developing brain. In adults, PDGF
formed with oncogenic viruses or chemical            actions foster wound healing by stimulation of
carcinogens produce a PDGF-like substance            fibroblast and smooth muscle cell proliferation.
that competes for binding to PDGF receptors          PDGF-b receptors also regulate interstitial fluid
and that antibody to PDGF can inhibit this           pressure by affecting fluid transport from the
PDGF-like activity.267 Moreover, a number of         vasculature into the extracellular compartment
human cancer cell lines have been shown to           of connective tissue.
secrete PDGF-like factors; these include cells          PDGF-mediated activities can also be targets
derived from osteosarcomas, glioblastomas, and       for chemotherapeutic attack. PDGF antagonists
fibrosarcomas. Some human tumor cells co-             include antibodies, DNA aptamers, and small-
express both PDGF-like factors and PDGF re-          molecule PDGFR-associated tyrosine kinase
ceptors, setting the stage for self-stimulation      inhibitors. Clinically, the most effect anti-PDGF
of cell proliferation, given exposure to the right   therapy has been with Gleevec, which inhibits
progression factors.268                              receptor tyrosine kinase activities associated
   When human melanoma cells are transfected         with the Bcr/Abl translocation in chronic mye-
with PDGF-BB cDNA, they produce in nude              loid leukemia, the Kit receptor in gastrointesti-
mice actively growing nests of tumor cells with a    nal stromal tumors (GIST), and the PDGF re-
distinct stroma and abundant blood vessels,          ceptor isoforms in glioblastoma.
suggesting a role for PDGF as an inducer of a
vascularized connective tissue stroma on which
                                                     Transforming Growth Factors
tumor cells can thrive.269 Amplification and/
or overexpression of PDGF receptors has been         The discovery of transforming growth factors
observed in human glioblastomas,270 and acti-        (TGFs) came about as the result of experiments
vation of PDGF-B gene expression in concert          showing that mouse 3T3 cells transformed with
with the PDGF-b receptor gene correlates with        murine or feline sarcoma viruses rapidly lost
the conversion of human hydatidiform mole into       their ability to bind EGF, whereas cells infected
choriocarcinoma.271                                  with nontransforming RNA viruses maintained
   The first clear link was forged between            normal levels of cell surface EGF receptors.273
growth factors and oncogene products when it         These initial results suggested that the sarcoma
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                          177

virus genome produced something that altered         inducer of anchorage-independent cell growth,
EGF receptors. Later, however, it was found that     as is EGF itself. TGF-b is also mitogenic for
murine sarcoma virus–transformed mouse fibro-         NRK cells and, to some extent, for mouse
blasts produced a polypeptide growth factor that     and human fibroblasts, but does not induce
competed for binding with EGF on cell sur-           anchorage-independent growth if added alone
faces.274 This factor was called sarcoma growth      to test cultures of NRK cells. However, it acts
factor (SGF), a 6000 to 10,000 MW, heat-stable,      synergistically with either TGF-a or EGF to in-
trypsin-sensitive polypeptide that stimulated pro-   duce the transformed phenotype and anchorage-
liferation of transformed and untransformed fi-       independent growth.
broblasts. It competed with EGF for binding to          A wide variety of RNA and DNA tumor viruses
EGF receptors, but it had a different molecular      stimulate production of TGF-like substances
weight and was immunologically distinct from         in cells transformed by them. Oncogenic RNA
EGF. It also had the interesting property of         viruses such as Harvey, Kirsten, and Moloney
being able to promote anchorage-independent          murine sarcoma viruses as well as Abelson mu-
growth in cultures of normal fibroblasts and thus     rine leukemia virus and DNA tumor viruses such
to confer on normal cells properties associated      as SV40 and polyoma have this property.276 Not
with the transformed phenotype. This pheno-          all the TGFs produced by cells transformed by
menon was reversible, so that after the SGF was      these viruses are identical, and this finding as well
removed from the growth medium, the cells re-        as others led to the concept that there are fami-
gained normal growth properties. Thus, SGF           lies of TGF-a and TGF-b produced by different
appeared to be a growth factor produced specif-      types of tumor cells. There is a close relation-
ically by transformed cells and capable of stim-     ship between the release of TGF-a and cellular
ulating their proliferation. This was the first ob-   transformation by murine sarcoma viruses, as
servation suggesting that neoplastic cells are       shown by the use of temperature-sensitive mu-
capable of autostimulation by producing their        tants of the transforming viruses.276,277 When
own growth factors. In this way, they could pre-     cells are grown at the temperature allowing cell
sumably escape the negative feedback systems of      transformation, TGF-a is produced, but when
the normal host that control the production          they are grown at the nonpermissive tempera-
and release of endogenous hormones and growth        ture, the factor is not produced.
   Although the EGF-competing activity of SGF
was contained in a 6000 to 10,000 MW fraction,
the cellular transforming activity appeared to       As discussed previously, TGF-a belongs to a
require, in addition, a fraction of higher mo-       family of growth factors that includes EGF,
lecular weight (20,000 to 25,000). The trans-        amphiregulin, and vaccinia virus growth fac-
forming activity of SGF isolated from murine         tor. TGF-a is produced as a 160–amino acid
or feline sarcoma virus–transformed cells was        proTGF-a form that is cleaved to produce a 50
subsequently shown to be separable into two          amino acid–soluble form of TGF-a (reviewed in
fractions: one of about 6000 MW, which com-          Reference 278). Although TGF-a was first
petes with EGF for binding to EGF receptors          found in culture fluids of oncogenically trans-
and induces only small colonies of normal rat        formed cells and is expressed by a wide variety
kidney (NRK) cells in soft agar, and one of about    of human cancer cells, its expression is not
25,000 MW, which does not compete for EGF            limited to neoplastic cells. During rodent em-
binding but is required for production of large      bryogenesis it is expressed in maternal decidua
colonies of NRK cells in soft agar.275 The for-      and in developing kidney, pharynx, and otic
mer, EGF receptor-binding form has been              vesicle. TGF-a mRNA and/or protein is also
termed transforming growth factor-a (TGF-a)          found in adult pituitary, brain, keratinocytes,
and the latter is called TGF-b. Like EGF, TGF-       ovarian theca cells, and macrophages, implying
a is a potent mitogen and appears to act through     a role in the economy of normal adult tissues
the same receptor, but by itself it is only a weak   as well. TGF-a and EGF have a similar ability
178                                                                                   CANCER BIOLOGY

to promote proliferation and differentiation of        TGF-b-related factors, including the bone mor-
mammalian mesenchymal and epithelial cells.            phogenetic proteins (BMPs), growth differenti-
This ability is not unexpected since they activate     ation factors (GDFs), activins, inhibins, Mul-    ¨
the same cell surface receptors.                       lerian inhibiting substance (MIS), Nodal, and
   In addition to the soluble form of TGF-a,           Lefty 1 and 2 (Fig. 4–17).289
there is a membrane-anchored form (the puta-              These TGF-b family members are translated
tive ‘‘juxatcrine’’ form). TGF-a shares this feature   as prepropeptide precursors with N-terminal
with several other membrane-anchored proteins          signal peptides that put them on the secretory
bearing EGF-like repeats on their extracellu-          pathway. Typically, they have a number of intra-
lar surface. Some of these membrane-bound              molecular disulfide bonds that facilitate a con-
glycoprotein forms are cleaved to yield soluble        formational structure and intermolecular dis-
EGF-like growth factors, while others, such as         ulfide bonds that form covalent dimers. The
the Drosophila Notch, Delta, and Crumbs gene           TGF-bs are secreted as latent forms and are
products as well as the C. elegans lin-12 and          activated by proteolysis.
glp-1, are not cleaved to release soluble factors.        The TGF-b superfamily of ligands binds to
Although the role of these cell surface EGF-like       and activates a family of transmembrane serine/
repeats isn’t clear, it seems probable that they       threonine receptor kinases. These receptors are
interact with receptors on the surface of adjacent     designated type I or type II, based on their
cells to sustain cell–cell adhesion and cell–cell      structural and functional characteristics. The
regulation of proliferation and differentiation.279    receptors have received a variety of names, but
   While TGF-a plays a role in normal devel-           the one that seems to stick is activin-receptor-
opment, wound healing, ECM production, an-             like kinase (ALK). There are seven ALK recep-
giogenesis, and cellular adhesion, it is clear that    tors (ALK-1 to -7) and a few others, including
its production at the wrong time or wrong place        BMP-R2 and TGF-bR2.289 In mammals, five
or its overproduction can favor neoplastic trans-      type II and seven type I receptors have been
formation and/or progression. Transfection of          identified. Type I and type II receptors exist as
the TGF-a gene into cultured cells can be              homodimers and binding of TGF-b type ligands
transforming. Overexpression or inappropriate          facilitates formation of a type I–II tetramer,
production of TGF-a has been observed in hu-           which then undergoes autophosphorylation and
man lung adenocarcinoma,280 squamous cell              triggers a signal transduction pathway involving
carcinoma,281 breast carcinoma,282 endometrial         intermediates called Smads (Fig. 4–18).290 The
adenocarcinoma,283 and hepatocellular carci-           phosphorylated, activatedSmadsform complexes
noma.284 TGF-a has been detected in the urine          that translocate to the nucleus and via interac-
of patients with hepatocellular carcinoma285 and       tion with cell type–specific co-activators or co-
in diffusion fluids of patients with a variety of       repressors turn genes on or off, depending on the
cancers, including ovarian, breast, and lung           cellular context. The herculean effects of this fam-
cancers, often as a bad prognostic sign.286 In         ily of growth factors is possible because of the
addition, liver carcinomas have been shown to          multiple combinatorial interactions of ligand-
develop in transgenic mice that constitutively         bound receptor complexes and interaction with
overexpress TGF-a.284,287 This latter observa-         other signal transduction pathways (Fig. 4–19).
tion supports the notion that deregulated ex-          Thus, a given TGF-b family ligand can induce
pression of TGF-a is a problem. It is expressed        different signaling pathways according to the
in developing liver, repressed in adult liver, and     composition of the receptor complexes. With all
re-expressed in regenerating liver,288 thus its        the effects that TGF-b family members have in
expression may be coupled to cell proliferation        development of various organ systems in the
and differentiation in a carefully regulated way.      embryo, including the heart, skeleton, and cra-
                                                       niofacial structures; left–right symmetry orien-
                                                       tation; nervous system development; and effects
                                                       in the adult organism, including reproductive
There are three TGF-b polypeptides, but these          function, wound healing, angiogenesis, extracel-
are members of a much larger superfamily of            lular matrix production, and modulation of the
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                      179

                                                                 TGF- 2
                                                                 TGF- 3
                                                                 TGF- 1
                                                                 BMP-8/OP-2 (human)
                                                                 BMP-8a (mouse)
                                                                 BMP-8b (mouse)
                                                                 Activin C
                                                                 Activin E
                                                                 Activin A
                                                                 Activin B

              Figure 4–17. Transforming growth factor b (TGF-b) superfamily. Amino acid
              sequences of the carboxyl-terminal polypeptides of the mouse TGF-b super-
              family members (and human BMP-8) were aligned using the PILEUP program
              (Genetics Computer Group, Madison, WI). Mouse and human sequences are
              available for all sequences except BMP-8. In the mouse, there are two BMP-8
              sequences (BMP-8a and BMP-8b), but only one in humans, because of a du-
              plication of the ancestral gene. (From Chang et al.,289 with permission.)

immune system, this is indeed a herculean list of    fibroblasts. However, as time went on and more
tasks.289,290                                        experiments were done, it became clear that
   In carcinogenesis, TGF-b plays both a good-       TGF-b had an ubiquitous tissue distribution
cop and a bad-cop role.291 Originally, TGF-b         and a key role in normal development that was
received its name as a ‘‘transforming growth         clearly at odds with its designation as a fac-
factor’’ because it assisted in inducing malig-      tor responsible for inducing malignant transfor-
nant transformation in cultured, nonmalignant        mation. In fact, TGF-b was shown to be a
180                                                                                    CANCER BIOLOGY

               Figure 4–18. General mechanism of TGF-b receptor and Smad activation.
               At the cell surface, the ligand binds a complex of transmembrane receptor
               serine /threonine kinases (types I and II) and induces transphosphorylation
               of the GS segments (orange) in the type I receptor by the type II receptor
               kinases (blue). The consequently activated type I receptors phosphory-
               late selected Smads at C-terminal serines, and these receptor-activated
               Smads (R-Smads) then form a complex with a common Smad4. Activated
               Smad complexes translocate into the nucleus, where they regulate transcrip-
               tion of target genes, through physical interaction and functional cooperation
               with DNA-binding transcription factors (X) and CBP or p300 coactivators.
               Activation of R-Smads by type I receptor kinases is inhibited by Smad6 or
               Smad7. R-Smads and Smad4 shuttle between nucleus and cytoplasm. The E3
               ubiquitin ligases Smurf 1 and Smurf 2 mediate ubiquitination and conse-
               quent degradation of R-Smads, yet can also interact with Smads6 and 7 and
               thereby ubiquitinate the type I receptors (not shown). (From Derynck and
               Zhang,290 reprinted by permission from Macmillan Publishers Ltd.)

proliferation-suppressing factor in epithelial and     in Reference 291). However, in mice expressing
lymphoid cells, so it was thought to be a tumor        activated TGF-b receptor, an increased per-
suppressor for these cell types. But ‘‘aha,’’ the      centage of metastatic foci was observed. Some of
wheel turned again when it was realized that           these latter effects may be due to TGF-b’s effects
TGF-b actually facilitated invasiveness and me-        on extracellular matrix deposition and turnover,
tastasis of later, progressing tumors. The data        allowing cancer cells an escape route from
indicate that activation of TGF-b signaling ini-       tumors. In addition, TGF-b stimulates angio-
tially delays the appearance of mammary tumors         genesis in vivo, and production and secretion of
in a Her2/neu-overexpressing mouse (reviewed           TGF-b by cancer cells suppress the activity of
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                          181

               Figure 4–19. R-Smad activation is regulated by receptor-interacting proteins
               and Smad6 and Smad7. SARA, Hgs/Hrs, Dab3, Dok-1, TRAP-1 (TGF-b
               receptor–associated protein), Axin, and ARIP (activin receptor–interacting
               protein) (green) interact with type I or type II receptors and R-Smads. SARA
               or HRS and Dab2 stabilize the Smad2–Smad3 interactions with TGF-b type I
               receptors and function in internalization with the endocytic machinery in
               endosomes. Other proteins, such as the RasGAP-binding protein Dok-1, the
               PDZ-domain protein ARIP1, and axin, also probably control subcellular lo-
               calization of receptors and link Smad2 / Smad3 to the receptors. TRAP-1, a
               homologue of the yeast sorting protein Vam6p, interacts with TGF-b or ac-
               tivin type I receptors first, and then with Smad4 upon receptor activation,
               possibly facilitating Smad4 interaction with activated Smad2 or Smad3.
               Smad6 and /or Smad7 expression can be induced by several signaling path-
               ways, including TGF-b/BMP signaling through Smads, and attenuates R-Smad
               activation. STRAP interacts with type I and type II receptors and with Smad7,
               thus stabilizing the interaction of Smad7 with the receptor complex. (From
               Derynck and Zhang,290 reprinted by permission from Macmillan Publish-
               ers Ltd.)

infiltrating immune cells and hence allow tumors        TGF-b superfamily of ligands has on the patho-
to escape immune surveillance.291 High levels of       genesis of cancer makes this superfamily of fac-
circulating TGF-b are found in patients with           tors and their signal transduction mechanisms a
invasive prostate cancer or colorectal cancer, and     potentially large number of targets for cancer
high urinary levels of TGF-b have been reported        therapeutics.292
in patients with hepatocellular carcinoma.291 The
mechanism by which TGF-b can switch from a
                                                       Hematopoietic Growth Factors
tumor suppressor function in early malignant
transformation to a tumor progression-inducing         The hematopoietic growth factors (Table 4–6)293
factor in more advanced disease is illustrated in      include erythropoietin, which stimulates red blood
Figure 4–20. The large panoply of effects that the     cell formation, the granulocyte, macrophage, and
182                                                                                     CANCER BIOLOGY

           NORMAL                                                            INVASIVE
          EPITHELIUM                Changes in genetic and               METASTATIC CANCER
                                      epigenetic context

            Suppressor                                TGF- expression/          Pro-oncogenic
             activities                                      activation            activities
             dominate                                                             dominate
        tumor cell autonomous                                          tumor cell autonomous
              growth inhibition                                               EMT
              apoptosis                                                       invasion/motility
              genomic stability                                               survival
                                                                       effects on tumor stroma

              Figure 4–20. TGF-b switches from tumor suppressor in the premalignant
              stages of tumorigenesis to pro-oncogene at later stages of disease leading to
              metastasis. Progression to metastatic disease is generally accompanied by
              decreased or altered TGF-b responsiveness and increased expression or ac-
              tivation of the TGF-b ligand. These perturbations, along with other changes
              in genetic or epigenetic context of the tumor cell and its stromal environment,
              combine to alter the spectrum of biological responses to TGF-b. (From Ro-
              berts and Wakefield,291 with permission.)

granulocyte-macrophage colony-stimulating fac-         clinically and commercially successful products
tors (G-CSF, M-CSF and GM-CSF), the inter-             of the new age of genetic engineering.
leukins, of which there are at least 20,294 which         The colony-stimulating factors are a subset of
act on various stem cell populations in hemato-        regulatory polypeptides of the ‘‘cytokine’’ family
poiesis, and various factors such as stem cell         that are involved in the proliferation and differ-
factor (SCF) and leukemia inhibitory factor            entiation of granulocytes and monocyte or mac-
(reviewed in Reference 293).                           rophages. The term CSF has stuck, and subsets
   The first hematopoietic GF to be discovered          of CSFs, based on their ability to stimulate par-
was erythropoietin (EPO) in 1906. By the mid-          ticular pathways of hematopoietic cell differen-
1960s, semi-solid culture techniques that could        tiation, have been identified (Fig. 4–21).295
support the growth of blood cell colonies from            The CSFs are glycoproteins with 15,000 to
normal bone marrow became available, and it            21,000 Da polypeptide chains and variable
soon became clear that soluble substances re-          amounts of carbohydrate. They consist of a single
leased into the conditioned culture medium of          polypeptide chain, except M-CSF, which is a
such cell types were necessary to support the          homodimer. They are produced by multiple cell
growth of these colonies. In the 1970s several of      types, including fibroblasts, placenta, endothelial
these ‘‘colony-stimulating factors’’ began to be       cells, lymphocytes, and bone marrow stromal cells.
purified and characterized, and by 1983, EPO,           Blood levels of CSFs are normally low, but their
GM-CSF, G-CSF, M-CSF, and interleukins                 production can be rapidly elevated in response to
(IL) -1, -2, and -3 had all been purified. Between      infection.
1984 and 1986 human cDNAs for EPO, GM-                    In vitro, some cell type specificity can be
CSF, and G-CSF became available, allowing              demonstrated: GM-GSF and IL-3 stimulate
their development for clinical use. These are          formation of granulocyte and macrophage colo-
considered by many observers to be the first            nies; G-CSF favors granulocyte colony formation;
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                                              183

Table 4–6. Some of the Hemopoietic Regulators                            implying that broad signaling cascades are initi-
                                                                         ated by receptor occupancy. As one might expect
Regulator (Abbreviation)                   Hemopoietic Cells             from these observations, there are numerous
Erythropoietin (Epo)                       E, Meg
                                                                         potential interactions among CSFs. For instance,
Granulocyte-macrophage                     G, M, Eo, Meg, E              combinations of two CSFs can produce additive
  colony stimulating                                                     or synergistic responses. Because CSF receptor
  factor (GM-CSF)
Granulocyte colony                         G, M
                                                                         levels are low (a few hundred per cell),296 occu-
  stimulating factor                                                     pancy of more than one type of CSF receptor may
  (G-CSF)                                                                be required for an optimal proliferative response.
Macrophage colony                          M, G
  stimulating factor
                                                                         Alternatively, progenitor cells may have multiple
  (M-CSF)                                                                receptor types so that they are able to respond
Multipotential colony                      G, M, Eo, Meg,                either to their normal, most appropriate ligand,
  stimulating factor                         Mast, E, Stem
                                                                         and secondarily to another less optimal ligand
Interleukin 1 (IL-1)                       T, Stem                       that may be turned on by a different stress and /or
Interleukin 2 (IL-2)                       T, B                          a different CSF-producing cell type, so that the
Interleukin 4 (IL-4)                       B, T, G, M, Mast
Interleukin 5 (IL-5)                       Eo, B
                                                                         host can respond to any of a number of emer-
Interleukin 6 (IL-6)                       B, G, Stem, Meg               gencies. In general, more mature, committed or
Interleukin 7 (IL-7)                       B, T                          single lineage cells can respond to stimulation by
Interleukin 9 (IL-9)                       T, Meg, Mast
Interleukin 10 (IL-10)                     T
                                                                         single growth factors, whereas less mature stem
Interleukin 11 (IL-11)                     Meg, B                        cells often require combined signalling from
Interleukin 12 (IL-12)                     NK                            multiple factors.293 It should be noted that CSFs
Megakaryocyte colony                       Meg
  stimulating factor
                                                                         were the first growth factors to show clearly that
  (Meg-CSF)                                                              growth factors can both stimulate cell prolifera-
Stem cell factor (SCF)                     Stem, G, E, Meg, Mast         tion of developing stem cells and induce a dif-
Leukemia inhibitory                        Meg
  factor (LIF)
                                                                         ferentiation pathway in cell-lineage progenitor
Oncostatin M (OSM)                         ?                             cells. Some hematopoietic growth factors have a
Macrophage inflammatory                     Stem                          fairly limited range of target cells and others have
  protein a (MIP-1a)
                                                                         a wide spectrum of target cells. For example, IL-3
Abbreviations: B, B lymphocytes; G, granulocytes; E, erythroid cells;    and IL-6 act on hematopoietic precursor cell
Eo, eosinophils; M, macrophages; Mast, mast cells; Meg; megakar-
yocytes; NK, natural killer cells; Stem, stem cells; T, T lymphocytes.   types, and leukemia inhibitory factor (LIF) acts
From Metcalf 293                                                         on megakaryocytes, osteoblasts, neuronal tissue,
                                                                         hepatocytes, and adipocytes.293
                                                                            Other ligand–receptor interactions have also
and M-CSF fosters macrophage colony growth.                              been observed. Binding of GM-CSF to its re-
In addition to stimulating progenitor cell prolif-                       ceptor down-regulates expression of G-CSF re-
eration and cellular commitment to a particular                          ceptors. GM-CSF and IL-3 compete for binding
differentiation pathway, the CSFs are also nec-                          to the same receptor. Macrophages are induced
essary to maintain functional activity of mature                         by IL-3 and M-CSF to produce G-CSF and also
cells, for example, chemotaxis, phagocytosis, and                        by GM-CSF to produce M-CSF. Obviously,
production and release of cytotoxic factors.296                          there is a great deal of cross talk between cells in
   Interestingly, receptors for multiple CSFs are                        the hematopoietic system, and it must require
present on many hematopoietic progenitor cell                            some finely tuned regulation, the mechanisms
types. There is a redundancy in the signaling pro-                       for which are only vaguely understood.
cess for hematopoietic cell proliferation, as if                            In spite of all the redundancies in the system, it
Nature has built in multiple mechanisms to pro-                          is clear that in populations of bipotential pro-
tect the host from invading organisms and other                          genitor cells, G-CSF fosters development of the
stresses. For example, granulocyte-macrophage                            cells in the granulocyte lineage and M-CSF fos-
progenitor cells and their maturing progeny ex-                          ters development of cells in the monocyte–
press membrane receptors for GM-, G-, M-, and                            macrophage lineage. This is borne out in vivo in
multi-CSF. In addition, CSF-occupied receptors                           that injection of G-CSF into mice induces a
can initiate multiple functions in responding cells,                     greater increase in peripheral blood granulocytes
184                                                                               CANCER BIOLOGY

               Figure 4–21. Hemopoiesis and cytokines. (From Miyajima,295 with permis-

than other blood cells, whereas GM-CSF induces       bone marrow. This increase in the peripheral
a rise in both macrophages and granulocytes          blood provides the capability of harvesting stem
(reviewed in Reference 296). IL-3 administra-        cells from the peripheral blood rather than the
tion elicits a rise in granulocytes, macrophages,    marrow, resulting in the recovery of many more
eosinophils, and megakaryocytes, as might be         stem cells and decreased trauma to the patient.
expected from its broad target cell specificity.      A number of cytokines are currently being used
   Clinically, the CSFs have been used in AIDS,      to attain maximal mobilization of stem cells into
aplastic anemia, congenital or cyclic neutropenia,   the peripheral blood. Data from clinical trials
and in cancer. The latter use has been to restore    indicate the use of peripheral stem cells is in
bone marrow function after chemotherapy, often       many cases as effective as use of bone marrow
accompanied by bone marrow transplantation.          cells in providing marrow reconsitution after
Both G-CSF and GM-CSF have been shown                chemotherapy.
to replenish peripheral blood neutrophils after         The receptors for hematopoietic GFs have
high-dose chemotherapy followed by autologous        several common features (reviewed in Reference
bone marrow transplantation.297,298 Positive ben-    293 and 300). They have highly related a chains
efits include decreasing the frequency of infec-      with low-affinity binding sites, which when di-
tions and shortening the stay in the hospital.       merized with b chains that provide some GF
   An intriguing sidelight of CSF therapy was        specificity produce a high-affinity receptor. Once
the observation of a dramatic rise in the number     receptor dimerization occurs, association with
of progenitor cells in the peripheral blood.299      one of a family of cytoplasmic tyrosine kinases
Usually, these cells are largely restricted to the   such as Tyk2 or JAK2 induces tyrosine phos-
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                         185

phorylation on cellular substrates involved in the    the signal transduction mediated by GM-CSF
signal transduction cascade. Thus, although the       binding to its receptor.
CSF family of receptors are not themselves ty-
rosine kinases, once activated they become re-
ceptor tyrosine kinases in disguise. A model for
                                                      Hepatocyte Growth Factor
these interactions is shown in Figure 4–22, where
                                                      and Scatter Factor
the a binding component of the cytokine binds to
the a chain of the receptor, providing a stable       Hepatocyte growth factor (HGF) and scatter
‘‘matrix’’ for sequestration of one b chain of the    factor (SF) were originally thought to be distinct
receptor, followed by dimerization of the recep-      cytokines that stimulated proliferation of cul-
tor via a second b-chain binding site on the          tured hepatocytes and that promoted motility
growth factor. This leads to activation of a cyto-    of epithelial cells, respectively. HGF was first
plasmic tyrosine kinase of the JAK family. Since      identified in the serum of partially hepatecto-
different cytokine receptors are capable of gen-      mized rats as a potent mitogen for cultured rat
erating qualitatively different signals in the same   hepatocytes and later also found in human
cells, it is likely that the receptors have some      plasma and serum, rat liver, and rat platelets
signalling specificity through their b subunits        (reviewed in Reference 302). HGF has been
for either recruiting different sets of JAKs or       cloned and sequenced.303 Scatter factor (SF)
attracting different substrates, or both.             was originally found as a secretory product of
   Extracellular matrix components are known          fibroblasts that dissociates epithelial cell colo-
to up-regulate GM-CSF signaling in neutrophils        nies into individual cells and stimulates migra-
and monocytes. There is evidence to suggest           tion of epithelial cells (reviewed in Reference
that this occurs via the ability of GM-CSF to         304). Purified SP also promotes invasiveness of
release binding of the laminin receptors (LR)         cultured human carcinoma cells into collagen
from the GM-CSF receptor, thus allowing               matrices, suggesting a role of SF in metastasis.
binding of the neutrophils and monocytes to the       Once both SF and HGF were cloned, it became
ECM.301 This ECM–cell attachment facilitates          clear that they were the same molecule.

               Figure 4–22. Stepwise formation of a generic cytokine receptor complex.
               Hatched boxes in b components represent conserved box 1 and box 2 se-
               quences. (From Stahl and Yancopoulos,300 with permission.)
186                                                                                 CANCER BIOLOGY

   HGF/SF is a disulfide-linked heterodimer of        the normal stem cells stop proliferating at some
55–65 kDa and 32–36 kDa subunits and is ex-          point and stop making these factors, tumor cells
pressed in several tissues of mesodermal origin,     may continue to make them until they undergo
including vascular smooth muscle cells, and has      enough genetic drift to become growth factor
angiogenic properties302 as well as the ability to   independent and capable of autonomous growth.
induce morphological changes and induce an-          The missing signal, or missing signal-receptor,
chorage-independent growth in HGF/SF trans-          that is necessary to convert the proliferating
fected epithelial cells.304                          cancer cell type into a differentiating cell type is
   The receptor for HGF/SF has been found to         not known for most human cancers; clearly this
be identical to the c-met proto-oncogene, which      is an important area for future research in cancer
is another member of the receptor tyrosine ki-       biology. Candidates for negative growth regu-
nase family.305 Activation of this receptor, which   lators have been found in a number of cell
is found on a variety of cells including keratino-   types, including lymphocytes, granulocytes, liver,
cytes, melanocytes, endothelial cells, and other     mammary gland, epidermis, and fibroblasts.
epithelial cells, triggers autophosphorylation and   One of these growth regulatory substances,
produces stimulation of the phosphatidyl inositol    Oncostatin M, is 28–36 kDa polypeptide cyto-
hydrolysis pathway and activation of Ras by shift-   kine that is produced by activated T lymphocytes
ing the equilibrium toward the active GTP-bound      and phorbol ester–treated monocytes and in-
state.306                                            hibits the ability of melanoma cells and other
                                                     cancer cell lines to grow in vitro. Paradoxically,
                                                     it is also a potent mitogen for AIDS-derived
Miscellaneous Growth Factors
                                                     Kaposi’s sarcoma cells in culture.308 These
More modulators of cell proliferation and dif-       data indicate the complexity of growth factor–
ferentiation are being identified as they are         stimulated events and demonstrate the cellular
looked for in normal and tumor tissues and cell      ‘‘context’’ dependency of their actions, much as
lines derived from different tissues (reviewed in    was discussed above for TGF-b.
Reference 307). Some of these were called bone          A variety of growth regulatory factors for mam-
cell–derived growth factors (BDGF), uterine-         mary gland epithelial cells have been reported
derived growth factor (UDGF), mammary                (reviewed in Reference 309). Some of these fac-
gland–derived growth factor (MDGF), mela-            tors have been detected in milk, some in condi-
nocyte growth factor, lung cancer, ovarian can-      tioned medium of cultured mammary cells, some
cer, and Wilms’ tumor derived growth factors, as     in mammary tissue extracts, and some even in
well as a family of estrogen-inducible growth        neoplastic cells. Some appear to be produced
factors called estromedins, found in a variety of    constitutively and others are induced by anti-es-
tissues including uterus, kidney, and pituitary      trogens. One regulatory factor, called mammasta-
gland. Whether all of these growth factors are in    tin, is produced by normal mammary cells310 and
fact distinct chemical entities or are in fact       can be detected in serum of women at the onset of
members of already identified growth factor           menstruation and in rat mammary gland in late
families is yet to be determined.                    pregnancy. This finding suggests mammastatin or
   The ubiquity of growth and differentiation        similar factors as candidates for differentiation-
factors leads one to predict that they will be       inducing agents that may provide the protective
found in all tissues in the body. These factors      effect of early pregnancy for breast cancer.
most likely act by paracrine or autocrine mech-
anisms to induce cell renewal or tissue repair of
damage, but under some circumstances they            SIGNAL TRANSDUCTION
may be released from the tissue and act on other     MECHANISMS
organs through an endocrine mechanism like
other known hormones. When cells undergo             Some of the signal transduction pathways in-
malignant transformation, they may continue to       volved in cancer either as oncogenic or tumor
produce these factors, much as their normal pro-     suppressor functions are shown in Figure 4–23
liferating stem cell counterparts do. Although       (see color insert).311 This is a simplified scheme
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                            187

and many other interconnecting pathway loops           the receptor for colony stimulating factor CSF-
implicated as playing a role in cancer are still       1. The met and tck protooncogene products turn
being discovered. In this section, some detail will    out to be receptors for hepatocyte growth factor
be given for many of the key signaling pathways        and nerve growth factor, respectively.
that play a role in the oncogenic process.                Some of the key substrates for receptor–
   The first signal transduction pathway to be well     tyrosine kinase coupled activity include (1) phos-
defined is the cyclic AMP-dependent protein             pholipase C (PLC-g) which in turn activates
kinase system (reviewed in Reference 312). This        phosphatidyl inositol hydrolysis, releasing the
system was the first of the so-called guanine nu-       second messengers diacylglycerol (DAG) and
cleotide binding protein-coupled receptors, or         inositol trisphosphate (IP3) that activate protein
GPCRs (reviewed in Reference 313), to be inves-        kinase C (PKC) and mobilize intracellular calcium
tigated. Study of the GPCR mechanism showed            release (a number of tumor promoters also acti-
that hydrolysis of protein-bound GTP could act as      vate PKC); (2) the GTPase-activating protein GAP
a signaling switch, and this led to the discovery of   that modulates Ras proto-oncogene protein func-
the receptor as a seven-transmembrane domain           tion; (3) Src-like tyrosine kinases; (4) phosphatidyl
protein. The first to be so identified was rhodopsin,    inositol kinase (PI3K) that associates with and
but upwards of a thousand GPCRs have now been          may modulate the transforming activity of poly-
identified, making this the largest known receptor      oma middle T antigen and the v-src and v-abl gene
family. In addition, the largest number of marketed    products; and (5) the raf proto-oncogene product
pharmaceutical agents interact with GPCRs.             that is itself a serine/threonine protein kinase.
   Transmembrane signaling by phosphorylation             Thus, activation of protein kinases is a key
and dephosphorylation mechanisms was de-               mechanism in regulating signals for cell prolif-
fined during the 1980s and 1990s. Many of these         eration. The substrates of these kinases include
turned out to be protein tyrosine kinase–              transcription regulatory factors such as those
coupled receptors (PTKRs) that phosphorylate           linked to mitogenic signaling pathways, e.g.,
tyrosine on substrates as opposed to serine or         proteins encoded by the jun, fos, myc, myb, rel,
threonine kinases such as the cAMP-activated           and ets proto-oncogenes.
kinase PKA and the TGF-b receptors. The                   The central role of tyrosine phosphorylation
PTKRs are all transmembrane proteins with a            in cell proliferative signaling mechanisms also
cytoplasmic domain that has intrinsic kinase           provides a target for chemotherapy. One should
catalytic activity activated by ligand binding. A      not forget, however, the catalysts for the other
partial listing of the protein kinases identified in    half of this reaction, the phosphatases. Although
various organisms is shown in Table 4–7.               it has been known for a long time that pro-
   The tyrosine kinase–coupled receptors men-          tein phosphatases play a regulatory role in cer-
tioned above are one potential target for carci-       tain cellular metabolic functions, e.g., in the
nogenic alteration. Activation of these receptors      activation–inactivation steps for glycogen syn-
can lead to phosphorylation of a number of key         thase and phosphorylase, it was more recently
substrates. Many growth factor receptors me-           demonstrated that phosphatases play a role in
diate their cellular effects by intrinsic tyrosine     the activity of various receptors and in the
kinase activity, which in turn may phosphorylate       function of certain cell cycle–regulating genes
other substrates involved in mitogenesis. As           (reviewed in References 315 and 316). For
noted in Chapter 5, a number of transforming           example, expression of a truncated, abnormal
oncogene products have growth factor or growth         protein tyrosine phosphatase in baby hamster
factor receptor–like activities that work via a        kidney (BHK) cells produces multinucleated
tyrosine kinase–activating mechanism. For ex-          cells, possibly by dephosphorylating the cyclin-
ample, the v-src gene product is itself a cell         dependent kinase p34cdc2.316 Activation of
membrane–associated tyrosine kinase. The v-sis         p34cdc2 requires dephosphorylation of a tyrosine
oncogene product is virtually homologous to the        residue, and this activation drives the cell from
B-chain of platelet-derived growth factor. The         G2 into M phase. The truncated phosphatase
v-erb product is a truncated form of the EGF           apparently interferes with the normal synchrony
receptor. The fms gene product is analogous to         between nuclear formation and cell division.
Figure 4–23. The molecular circuitry of cancer. Although countless differ-
ences between normal cells and cancer cells have been documented, much
progress in identifying and connecting the fundamental pathways responsi-
ble for programming malignant cell growth has been made. Most cancer-
associated mutations disrupt essential homeostatic mechanisms that regulate
cell proliferation and survival. In many cases, particular mutations have been
linked to specific biological phenotypes shown by cancer cells (yellow boxes).
The cellular machinery responsible for controlling mammalian cell physiology
is largely shared between human and mouse cells (black lines). Comparisons
of human and mouse experimental cancer models identify several pathways
that seem to have more prominent roles in human-cell transformation (red
lines), as well as other molecular pathways that serve in dominant positions in
mouse cancer models (blue lines). Perturbation of these five pathways (RB,
p53, telomere maintenance, HRAS, and ST-PP2A) allows transformation of
human cells. However, significant interactions among these pathways and
other molecules that are implicated in the development of particular types of
human cancer exist and remain to be characterized in detail. For the pur-
poses of clarity, this representation is simplified and is illustrative rather than
comprehensive. For example, clear evidence links RAS signaling to induction
of cyclin D1 expression and the role of PP2A in most human cancers remains
to be elucidated. An alternative method of displaying the molecular circuitry
that programs the cancer phenotype can be found at
nrc/journal /v2/n5/wienberg_poster/. ALT, alternative lengthening of telo-
meres; CDK, cyclin-dependent kinases; Cyc, cyclin; E6, human papillomavi-
rus E6 oncoprotein; E7, human papillomavirus E7 oncoprotein; LT, SV40
large T antigen; ST, SV40 small T antigen; TERT, telomerase reverse tran-
scriptase; WAF1, also known as p21 (encoded by CDKN1A). (From Hahn and
Weinberg,311 reprinted by permission from Macmillan Publishers Ltd.)
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                                    189

   Protein tyrosine phosphatases (PTPases) are a                aberrant phosphorylation state of tyrosine in cer-
diverse family of enzymes that exist in cell mem-               tain key proteins such as c-Src or c-Raf that can
branes. Some of them are associated with recep-                 lead to cellular transformation could theoretically
tors that have tyrosine kinase activity. Phospha-               come about through deregulation of a protein ki-
tases are also in other intracellular locations. The            nase or underexpression of a protein phosphatase.

            Table 4–7. Mammalian, Drosophila, and Yeast Protein Kinases

            Protein-Serine/ Threonine Kinases                         Protein-Tyrosine Kinases
            CYCLIC NUCLEOTIDE REGULATED                               SRC GENE FAMILY
            cAMP-dependent protein kinases (Ca, Cb)                   pp60c-arc (fibroblast, neuronal forms)
            cGMP-dependent protein kinase                             pp62c-yes, pp56ick
                                                                      fgr, hck, fyn, lyn proteins
            Phosphorylase kinase (distinct liver and                  ABL GENE FAMILY
              muscle forms?)
            Myosin light chain kinases (skeletal,                     p150c-abl (type I and type II
              smooth muscle)                                            N-terminus)
            Type II-calmodulin dependent protein kinase               arg protein
              (brain a, b, b0 subunits; liver a, a0 subunits;
              muscle b, b0 subunits)                                  FPS GENE FAMILY
            Calmodulin-dependent protein kinases I and III
            DIACYLGLYCEROL REGULATED                                  NCP94
            Protein kinases Cs (a, b and b0 , g, d [RP14])            c-fps-related proteins (TKR11
                                                                         and TKR16)
            Casein kinases I and II                                   GROWTH FACTOR RECEPTORS
            Nuclear protein kinases N1 and N2                         EGF receptor family
            Protease-activated kinases I and II                          EGF receptor (c-erbB protein)
            Glycogen synthase kinases 3 and 4                            neu protein (erbB2 protein)
            Heme-regulated protein kinase                             Insulin receptor family
            Double-stranded RNA regulated protein kinase                 Insulin receptor
            Double-stranded DNA regulated protein kinase                 IGF-1 receptor
            S6 kinase                                                    c-ros, met, trk proteins
            b-adrenergic receptor kinase                              PDGF receptor family
            Rhodopsin kinase                                             PDGF receptor
            Histone H1 kinase                                            CSF-1 receptor (c-fms protein)
            Hydroxymethyglutaryl-CoA reductase kinase                    c-kit protein
            Pyruvate dehydrogenase kinase                             c-sea, ret proteins
            Branched charin ketoacid dehydrogenase kinase
            Polypeptide-dependent protein kinase
            Polyamine-stimulated protein kinase                       OTHERS
            c-mos, c-raf, A-raf, pks, pim-1 proteins                  p75 (liver)
            CDC-R (PSK-J3), CDC2Hs, PSK-H1, PSK-C3                    p120 (brain)


            Protein-Serine/ Threonine Kinases                         Protein-Tyrosine Kinases
            CYCLIC NUCLEOTIDE REGULATED                               Dsrc64B protein
            cAMP-dependent protein kinase-related                     Dsrc28C protein
              (C0, C1, C2)                                            Dash protein
            cGMP-dependent protein kinase-related                     fps-related protein
              (G0 [2 genes], G1)                                      EGF receptor (types I, II, III
            DIACYLGLYCEROL REGULATED                                  Insulin receptor
            Protein kinase C                                          sevenless protein

            Casein kinase II
            raf protein
190                                                                                                             CANCER BIOLOGY

            Table 4–7. Mammalian, Drosophila, and Yeast Protein Kinases (countinued)

            cAMP-dependent protein kinase-related
            TPK1, TPK2, TPK3, SRA3 (S. cerevisiae)
            CDC28 (S. cerevisiae (&cdc2þ in S. pombe)
            CDC7 (S. cerevisiae)
            KIN28 (S. cerevisiae)
            wee1þ (S. pombe)
            nim1þ (S. pombe)
            STE7, STE11 (S. cerevisiae)
            KIN1, KIN2 (S. cerevisiae)
            SNF1 (S. cerevisiae)
            ran1þ (S. pombe)
            Protein kinases are listed under protein-serine/threonine kinase and protein-tyrosine kinase in subfamily
            groups. Protein kinases included in this table have either been characterized as distinct by complete or
            partial protein purification, or have been identified as unique based on nucleotide sequencing. The reader
            should be aware, however, that many of the protein kinases whose existence is deduced from sequences of
            cDNA clones have not yet been proven to be protein kinases. Conversely, until the complete amino acid
            sequences are available for all of the protein kinases that have been identified on the basis of their enzymatic
            activity, one cannot be certain that they are distinct proteins. The primary references have been omitted to
            save space, but are available from the author,
            (From Hunter,314 with permission.)

For example, cells treated with vanidate, a PTPase                   ligands, such as growth factors, that bind to the
inhibitor, had increased protein phosphotyro-                        extracellular domains of these receptors.
sine levels and a transformed phenotype.317                             An extension of the concept of ligand-
   An interesting fallout from comparative geno-                     activated transmembrane signaling was estab-
mic sequencing is the way in which kinase and                        lished by studies of the mechanism of activation
phosphatase pathways have been conserved over                        of the Ras-Raf-Map kinase pathway. In this case,
evolution and how large a percentage of genes                        the signaling molecule Ras is anchored to the
in the genomes of various organisms are devoted                      cell membrane via a linker, and its activation
to protein phosphorylation–dephosphorylation                         leads to GTP-GDP exchange and interaction
regulation (reviewed in Reference 318). For ex-                      with a Grb2/SH2/SH3 complex, which leads to
ample, in the yeast S. cerevisiae, there are 114                     activation of downstream effectors such as Raf,
protein kinase genes (none of them protein                           MAPK, and PI (see below). Interestingly, Ras
tyrosine kinases) out of 6217 genes (1.8%). In                       was the first signaling protein identified as being
C. elegans there are 400 protein kinase genes (92                    conserved through evolution from yeast to hu-
are PTKs) out of 19,099 genes (total of 2.1%). In                    mans, with homologs of many of the pathway
humans, there are predicted to be >1100 protein                      components being functionally interchangeable
kinase genes (at least 150 of which are PTKs) out                    among organisms.318
of a genome of about 30,000 genes (3.7%). In                            A large family of nuclear receptors has also
addition, in C. elegans there about 200 phospha-                     been identified. These include the receptors for
tase genes and there are hints that there may be                     hormones such as estrogen, progesterone, and
one phosphatase for each protein kinase.318 Fine                     corticosteroids, as well as retinoic acid and other
tuning indeed! It is also interesting that PTKs do                   DNA-binding receptors. In this case, the ligand-
not exist in yeast but do in C. elegans, one of the                  activated receptor itself is the ‘‘second messen-
simplest multicellular organisms, a finding sug-                      ger.’’ The unliganded receptors act as repres-
gesting that PTKs evolved along with the need for                    sors because of their interaction with histone
intercellular communication. This makes sense                        deacetylases. Ligand binding activates receptor-
because a majority of PTKs are transmembrane                         mediated gene transcription by releasing histone
receptors that respond to secreted extracellular                     deacetylases and recruiting histone acetylases.
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                            191

   The phosphoinositol (PI)-mediated pathway           transcription of cAMP-responsive genes. This sort
was discovered in the 1960s. These studies led         of process is typical of many transcription factors,
to the identification of a number of phospho-           many of which are regulated by phosphorylation
lipid-derived second messengers such as DAG            that enhances nuclear transport, DNA binding,
and its role in regulating PKC, activation of          or transactivation events. As will be seen under
which impacts a number of downstream effec-            the description of individual signal transduction
tors (see below). The idea that phospholipid           pathways below, many receptors (i.e., transmem-
metabolites constitute another class of second         brane, membrane associated, cytoplasmic, and
messengers produced in cells by a wide variety         nuclear) regulate gene transcription via down-
of hormones and growth factors came originally         stream effectors.
from the observation of Hokin and Hokin319 of
stimulated incorporation of 32P into phospho-
                                                            Protein–Protein Interaction Domains
lipid in hormonally activated tissues. These in-
vestigators demonstrated increased phospho-            Protein–protein interaction domains determine
lipid turnover in pancreatic tissue exposed to         the way that proteins talk to each other in
acetylcholine. Later it became clear that catab-       forming functional complexes such as recep-
olism of inositol lipids is stimulated in many         tor multimeres, transcription factor–coactivator
different tissues by many different external sig-      complexes, and DNA replication ‘‘machinery.’’
nals. The link of IP3 generation to Ca2þ mobi-         In addition, protein conformational changes can
lization is now widely recognized and explains         be induced by phosphorylation or other chemical
one pathway leading from receptor occupation           modifications such as acetylation or methylation,
to Ca2þ mobilization, PKC activation, and sub-         by ligand binding, or by interaction with extra-
sequent cellular responses.320                         cellular matrix components to form functionally
   All the second-messenger signaling mediators        active (or in some cases inactive) proteins. Also,
described above involve small molecules such as        protein interaction domains are involved in as-
nucleotides (e.g., cAMP, cGMP) or phospho-             sociation of proteins with phospholipids, small
lipids (e.g., IP3). The discovery that a gas, nitric   molecules such as drugs, and nucleic acids.
oxide (NO), could also act as a second messenger       Interaction domains are also involved in target-
was surprising. Originally described as endo-          ing proteins to specific subcellular locations, in
thelial cell–derived ‘‘relaxing factor,’’ NO is now    providing recognition sites for post-translational
known to activate cytoplasmic guanylyl cyclase         modifications or second messengers, and in
to elevate cGMP (reviewed in Reference 321),           substrate binding to an enzyme. For example,
which in turn activates cGMP-dependent pro-            phosphorylated tyrosines on PTK receptors bind
tein kinase and other cGMP-mediated events.            phosphotyrosine recognition domains on Src
                                                       homology peptides of the PTKR/Grb2/SOS
                                                       complex involved in signal transduction from
Some Key Signal Transduction
                                                       EGF receptors.
                                                          The folding of proteins into appropriate
                                                       confirmations is key to achieving a functional
     Transcriptional Regulation
                                                       structure.322 Isolated interaction domains can
     by Signal Transduction
                                                       be quite small and fold independently, with
After binding ligand, many transmembrane re-           their amino- and carboxyl-termini juxtaposed in
ceptors activate downstream effector cascades          space in a way that leaves their ligand-binding
that ultimately lead to a nuclear signal that turns    domain available.323 These domains recognize
genes on or off. The classic example of this is the    exposed interaction sites on their protein part-
protein kinase A (PKA)–cAMP-binding proteins           ners. The growth factor signal transduction
(CREB) system in which cAMP activates PKA to           pathway is a good example of this (Fig. 4–14). In
release its catalytic subunit, which translocates to   most cases, a protein interaction domain rec-
the nucleus and phosphorylates the CREB tran-          ognizes a consensus recognition sequence and a
scription factor. Phosphorylated CREB binds to         specific structural conformation with flanking
co-activator factors, and this complex activates       sequences that provide additional contacts and
192                                                                                     CANCER BIOLOGY

some binding selectivity.323 Protein interaction        signal transduction and provides a mechanism to
domains involved in signaling are present in            prevent willy-nilly activation of signaling path-
hundreds of copies in the human proteome, and           ways. A good example is the translocation of
they are used over and over again in different          transcription factors initiated by activation of cell
functional ways. For example, SH3 domains are           surface receptors (Fig. 4–24). If these receptors
used in different proteins to regulate signal           were not separated from the site of their action or
transduction, protein trafficking, cytoskeletal          the enzymes regulating their phosphorylation
organization, cell polarization, and organelle          state and stores of downstream effectors such as
biosynthesis (reviewed in Reference 323). This          calcium, these transcription factors could be
re-use of interaction domains allows for re-            constitutively activated all the time, creating
shuffling of these domains for different func-           havoc in cell function. Take, for example, the
tional purposes and this, over evolution, may           activation of NFAT by the PLC-g pathway,
have been one way to facilitate new cellular            shown in Figure 4–24 (see color insert). NFAT
functions. This reshuffling can also have dire           nuclear localization is regulated by an activated
consequences because in cancer or hereditary            receptor that stimulates PI turnover in the
disorders these interaction domains can be              plasma membrane and that increases IP3 pro-
mutated, causing loss of important protein–             duction. IP3 mediates the release of calcium ions
protein interactions or creation of aberrant pro-       from internal stores (e.g., in the endoplasmic
tein complexes. It should be noted that drugs           reticulum). CA2þ ions stimulate the phosphatase
can be designed to target protein interaction           calcineurin that dephosphorylates key sites on
domains and either facilitate or inhibit protein–       NFAT, exposing a nuclear import signal (reviewed
protein interactions.                                   in Reference 325).
                                                           Timing is also important in signal transduc-
                                                        tion. For example, the rates of effector activa-
      Spatial and Temporal Regulation
                                                        tion in the various steps of a multistep cascade
Another important aspect of signal transduction         and the speed of movement of effector second
is the spatial orientation of receptors and their       messengers to their site of action regulate how
downstream effectors as well as the timing              cells respond to signals. Also, these rates may
of cellular responses following ligand–receptor         vary by cell type and provide some specificity of
interaction. For example, protein kinases and           response as well as a well-regulated group of
phosphatases as well as their substrates are fre-       functions. The sequence and timing of this
quently localized in different compartments in          could be key to a normal versus an aberrant
the cell. This allows for some specificity of interac-   cellular response. A good example of this is acti-
tion. Also, the orientation of signaling compo-         vation of protein kinase C by CA2þ, DAG, or
nents may be organized differently in different         tumor promoters. The timing and duration
cell types, providing some cell type selectivity. As    of this activation differs depending on the type
noted above, some receptors are transmem-               and half-life of the agonist (e.g., CA2þ vs. DAG
brane, some are cytoplasmic, and some are nu-           vs. fatty acids; see Fig. 4–25).326 For example,
clear and their downstream effectors may be             it is now apparent that activation of PKC may
similarly distributed. Although some second mes-        occur by a number of alternative routes and that
sengers such as cAMP are freely diffusible moi-         some activations of PKC may be transient, lead-
eties, many second-messenger systems rely on            ing to early or rapid cellular responses, while
protein–protein interactions at the cell mem-           other activities may be more sustained, leading
brane or in the cytoplasm and nucleus and have          to late cellular responses such as cell prolifera-
limited free diffusion. An example of this is the       tion and differentiation events.
ligand–receptor interaction of glucocorticoids             Phospholipases C, A2, and D all appear to be
with their receptor in the cytoplasm and the as-        involved in PKC activation, with PLC being more
sembly, disassembly, and nuclear transport of           involved in the early response pathway and PLA2
the corticosteroid receptor complex via heat            and PLD in the late response pathway. The
shock proteins to the cell nucleus.324 Protein          reaction products of phosphatidyl choline hy-
localization is fundamental to the regulation of        drolysis by PLA2, i.e., the cis-unsaturated fatty
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                          193

              Figure 4–24. Getting to the nucleus. Signal transduction pathways initiated
              by activated receptors control translocation of transcription factors from
              the cytoplasm to the nucleus. Movement of the transcription factors NFAT,
              NF-kB, SMADs, and STATs is associated with the phosphorylation or de-
              phosphorylation of signaling proteins by kinase and phosphatase enzymes.
              (From Cantley,325 with permission from the American Association for the
              Advancement of Science.)

acids and lysophosphatidylcholine, can both          in a way to select for the appropriate response
enhance PKC activation. The tumor-promoting          only?
phorbol esters mimic DAG for activation of PKC          The specificity of signaling via different path-
and, being more stable than DAG, favor the           ways or in different cells exposed to the same
sustained, late-response pathway involved in cell    agonists (growth factors, hormones, drugs, etc.)
proliferation and differentiation.                   can be at least partly explained by spatial and
                                                     temporal separation of receptors and down-
                                                     stream effectors as noted above, but it is also
    Signaling Networks and Cross Talk
                                                     because different cell types can have different
The old concept that metabolic and signal-           repertoires of downstream effectors, different
ing pathways were sequential linear cascades         sets of transcription factors, and/or different sets
has been shown to be incorrect in recent years.      or levels of co-stimulatory or co-repressor factors
It is now known that both of these types of          modulating gene expression.
pathway have multiple linkages and form net-            Redundancy of signaling pathways is another
works that were unpredicted when these path-         challenge. Signal transduction pathway cross-
ways were initially discovered and delineated.       talk can occur between pathways activated by a
This will be described in more detail in Chapter     single receptor or among pathways activated by
5 under Systems Biology.                             different receptors. Cross talk among signaling
   The linkage and networking of signal transduc-    pathways can result in up- or down-regulation of
tion pathways create two conceptual questions:       one of them triggering coordinate responses in
(1) How can signaling specificity be maintained?      another one. Thus, inhibition of one component
(2) How can the redundancy implicit in over-         of a signal transduction pathway may be com-
lapping or interconnected pathways be regulated      pensated for in the cell by up-regulation of
194                                                                                  CANCER BIOLOGY

                                                       to try to target the downstream events where
                                                       transduction pathways converge in their ability to
                                                       stimulate gene activation events.
                                                          An example of the cross talk among ligand–
                                                       receptor triggered events is the binding of the
                                                       growth factor beta platelet-derived growth factor
                                                       (bPDGF) to its receptor bPDGFR (Fig. 4–26).
                                                       This induces dimerization of the receptor, which
                                                       in turn triggers signal transduction pathways.
                                                       The bPDGF receptor becomes autophosphor-
                                                       ylated on multiple tyrosines by activation of
                                                       its receptor tyrosine kinase, which fosters bind-
                                                       ing to specific Src homology 2 domain (SH2)-
                                                       containing proteins that are part of the Grb2-
                                                       Sos-Ras-Raf-Mek-Erk pathway. In addition,
                                                       there is cross talk with the phosphatidyl inositol
                                                       kinase (PI3K) pathway. PI3K can also stimulate
                                                       Rac GTPase, which can activate JAK/STAT sig-
                                                       naling events. Activation of the SH2 domain
                                                       protein PLC-g1 can also potentially stimulate
                                                       PKC signaling pathways. Thus, cytoplasmic
                                                       signaling proteins form networks of interactions
                                                       rather than simple, linear pathways.327 These di-
                                                       verse signaling pathways, in turn, induce broadly
                                                       overlapping sets of genes.328
                                                          GTP-binding protein (G-protein) signaling
                                                       events are another ubiquitous pathway for gene
                                                       activation, some of which are mediated by cyclic
                                                       AMP that has protean effects on cellular pro-
                                                       cesses (see below). Mutations in components
                                                       of G protein–coupled pathways have been ob-
                                                       served, some of which are involved in a number
                                                       of human diseases, including cancer.

Figure 4–25. Schematic representation of agonist-      Overview of Some Signal Transduction
induced membrane phospholipid degradation for sus-     Pathways Important in Cancer
tained PKC activation (A). Time course of generation
of various signaling molecules (B). DG, diacylglyc-
                                                            G Protein–Linked Receptors
erol; FFAs, free cis unsaturated fatty acids; IP3,
inositol 1,4,5-trisphosphate; LysoPC, lysophosphati-   As noted above, guanine nucleotide binding
dylcholine; PC, phosphatidylcholine; PIP2, phosphat-
                                                       protein–coupled receptors are a diverse set of
idylinositol 4,5-bisphosphate. (From Nishizuka,326
with permission.)                                      ligand-activated receptors that regulate adenyl-
                                                       ate cyclase, ion channels, certain protein ki-
                                                       nases, and other signal transduction mechanisms.
another pathway. This has important therapeu-          They all share a common general structure
tic implications, because a drug that blocks an        (Fig. 4–3) with an external ligand-binding
early or upstream component of a given pathway         domain, membrane-spanning domains, and an
may be circumvented by activation of another           intracellular domain that interacts with various
parallel pathway. This phenomenon is seen,             G-protein complexes and contains sites for phos-
for example, in the development of resistance          phorylation. The ligands that interact with such
to some chemotherapeutic agents. A goal, then, is      receptors include a- and b-adrenergic agonists
THE BIOCHEMISTRY AND CELL BIOLOGY OF CANCER                                                        195

                    PDGFR                                               PDGFR


                   716                            PIP 3                 p-Tyr
                       Tyr-p     Grb2    Sos
                       Tyr-p     GAP                                    p-Tyr 740, 751
                   763 Tyr-p             Ras      Rac                   p-Tyr
                   771 Tyr-p
                                         Raf              PKB           p-Tyr
                                 Grb2                           PLC-                     STAT
                  1009                                                  p-Tyr
                         Tyr-p                    GSK3
                                 Shp2    Erk                    PKC
                         Tyr-p                                          p-Tyr 1021
                                         TCF       Jun    FKHRL1 NFAT                    STAT-p
              Transcription factors:                       AFX
                                         SRF                                             p-STAT

              Figure 4–26. A signaling network from the activated bPDGFR. The acti-
              vated receptor is a dimer. Each receptor chain becomes phosphorylated on
              multiple sites, some of which are depicted here, and binds specific SH2-
              containing proteins. The receptor itself has redundant interactions, for ex-
              ample, with Grb2. There are specific pathways leading from the receptor to
              the nucleus. There are also numerous potential cross-connections between
              distinct pathways, some of which are shown. See text for more detail. The
              figure is illustrative and by no means comprehensive; some binding partners
              and potential pathways, including Src family kinases, are not depicted. (From
              Pawson and Saxton,327 reprinted with permission from Elsevier.)

and antagonists, angiotensin, serotonin, bom-            The Ga subunits possess intrinsic GTPase
besin, bradykinin, acetylcholine (muscarinic          activity. Modification of Ga by cholera toxin
type), vasopressin, and vasoactive intestinal         activates Ga proteins by inhibiting their GTPase
polypeptide (VIP).                                    activity, and binding of pertussis toxin blocks
   The first four G proteins discovered were           receptor-mediated activation of G proteins.
designated Gs, Gt, Gi, and Go. A large number of      Thus, these two toxins, which ADP ribosylate
G proteins have subsequently been identified by        different sites on Ga subunits, are often used as
cDNA cloning (reviewed in Reference 329). G           tools to investigate the role of G proteins in
proteins are heterotrimers composed of an a           various physiological systems. G proteins are
(39–46 kDa), b (37 kDa), and g (8 kDa) subunit        anchored in the plasma membrane of cells by
(reviewed in Reference 330). The b and g sub-         lipid modifications of the subunits; g subunits
units form a tightly associated complex (Gbg)         are prenylated and some g subunits are myr-
that functions as a unit and forms trimers with       istolated.
an a (Ga) subunit. The a subunit has a high-             The Gs and Gg families regulate the adenylyl
affinity binding site for GTP or GDP. The GDP-         cyclase and phospholipase C-beta (PLC-b)
bound form of a binds tightly to the bg complex       pathways, respectively. The Gi and Go families’
and is inactive. When GTP is bound it displaces       activities are more general and less well defined.
GDP, dissociates a from bg, and induces the           The best understood is the Gi family regulation
regulatory function of a. While it has been           of the transducing pathway for light detection in
thought that the activated Ga subunit alone           the eye.329 G12 and G13 pathways share down-
regulates ion channels, adenylyl cyclase, phos-       stream effectors, yet show some selectivity for
pholipase Cb, and other enzymes, it is now ap-        certain ligands, e.g., LPA and thrombin. The
parent that the Gbg dimer also plays a role in        four broad G-protein families transduce signals
modulating the activity of these effector sys-        from a large number of diverse activating ligands
tems. Currently, there are 20 Ga, 6 Gb, and 11        and modulate a number of cellular functions
Gg subunits that have been identified.329              such as homeostasis, embryonic development,
196                                                                                          CANCER BIOLOGY

metabolic regulation, gonadal development, and             ribosylate the Ga subunit. The Gq pathway is the
memory (Fig. 4–27). The Gs pathway was the                 one activated by calcium-mobilizing hormones
original cell signaling pathway defined and is the          and activates PLC-b to produce IP3 and DAG.
one that regulates protein phosphorylation. A                 The formation of cAMP activates protein ki-
number of interconnecting linkages to other                nase A (PKA), which can have a profound effect
signaling pathways have been identified.                    on cellular metabolism (Table 4–1). While many
   The Gi-mediated pathway was originally iden-            of the cellular actions of cAMP are due to acti-
tified by its inhibiting activity on adenylyl cyclase;      vation of PKA, other actions are attributed to a
however, a number of hormones and neuro-                   direct action of gene transcription via binding to
transmitters including epinephrine, acetylcho-             cAMP-binding proteins that act as transcription
line, dopamine, and serotonin use the Gi and Go            factors (see above).
pathway. Signal flow through this pathway is                   Because cAMP appears to play a key role in
blocked by pertussin toxin via its ability to ADP-         cell proliferation and differentiation, a number

                Figure 4–27. A schematic representation of how signaling through G-protein
                pathways can regulate systemic functions. Many extracellular agents, such as
                hormones (for example, glucagons, luteinizing hormone, and epinephrine),
                neurotransmitters (acetylcholine, dopamine, and serotonin), chemokines (IL-
                8), and local mediators (LPA), signal to the four main G-protein families to
                regulate such cellular machinery as metabolic enzymes, ion channels, and
                transcriptional regulators. Modulation of the activities of the cellular machines
                in turn gives rise to altered cellular functions, such as changes in glucose me-
                tabolism in liver and muscle or altered activities of pacemaker cells in the heart.
                These cellular activities contribute to the regulation of large-scale systems such
                as organismal homeostasis and learning and memory. Thus, G-protein path-
                ways can propagate regulatory information through layers of increasing orga-
                nizational complexity. At all levels, the examples shown here represent only a