Lung cancer screening with CT

Document Sample
Lung cancer screening with CT Powered By Docstoc
					  Lung cancer screening with CT
            Jud W Gurney MD

               Why is screening controversial?

  Lucky's      What’s the evidence for CT?

               What should I do?

Our objectives are 3 fold. 1. review the question of why
we don’t currently screen for lung cancer. 2. review the
current controversy generated by CT screening, both
pro and con. And 3rd end with the practical question of
what should I do now in my clinical practice.

        Why is screening controversial?
           Cancer deaths

              Lung          175,000
              Colon          50,000
              Breast         45,000
              Prostate       40,000

Lung cancer is surely a worthy tumor to screen because it is the
most common cause of cancer deaths in both men and women.
Yearly lung cancer deaths exceed that of the next 3 most common
cancers combined.

        Why is screening controversial?

           Lung         None

           Colon        FOB

           Breast       Mammography

           Prostate     PSA

One reason that deaths are lower in the also rans is that
screening is accepted for those cancers but not for lung


                  Why is screening controversial?
                    Smoking trends                      Nonsmoker


                                                        Current Smokers

                                                        Former Smokers


           1960      1965   1970   1975   1980   1985   1990        1995

90% of lung cancers are due to smoking, thus the obvious target to
screen is smokers. Nearly half of the US population are either
current or former smokers. The rise in percentage of former smokers
leads to the curious statistic that newly diagnosed lung cancer is
more likely to come from a former smoker than a current smoker. A
fallacy is that former smokers eventually lose their risk of developing
lung cancer.

                             Where have you gone……

This is not true. For example, Joe DiMaggio developed lung
cancer nearly 40 years after his last cigarette. As a young
man, he smoked 3 pks/day. In a recent biography, note was
made that one player in the dugout was assigned to have a
cigarette lit for Joe when he came in from the outfield.

         Why is screening controversial?
                 Stage      5 yr survival (%)

                    Ia           70-80
                    Ib           60-70
                   IIa            35
                   IIb            25      13% 5 year survival
                  IIIa            10      13% present Stage I

                  IIIb             5
                   IV              1     Mountain Chest 1997 111: 1701-17

Lung cancer is curable and thus would be amenable to screen if a
good screening test existed. The 5 year survival for Stage I lung
cancer is as good as that seen in breast cancer. The current 5-year
survival of 13%, which hasn’t changed in decades, is consistent with
advanced disease at diagnosis. In fact, lucky 13 is the same
percentage that currently present with stage I disease

         Why don’t we screen?
            It doesn’t work!
              Philadelphia Pulmonary Neoplasm Project
              North London Trial
              Erfurt 2 County Trial
              NCI Lung Cancer Projects
                 Johns Hopkins Lung Project
                 Memorial Sloan Kettering Lung Project
                 Mayo Clinic Lung Project
              Czechoslovakia Trial

Lung cancer screening with either chest x-rays or sputum cytology
has been investigated for 30 years. None has shown benefit. The
last large randomized trials, the NCI sponsored lung cancer projects
and a Czech study, were conducted in the 1970’s. Let’s review this
data, which has concluded that CXR’s are not worthwhile and why
this conclusion is controversial.

                 Sputum Cytology
                                                        Mortality                  Stage I
                                        5,072                 90                   59 (41%)
                                       CXR + Sputum

             10,400                          Yearly

                                        4,968                 92                   58 (40%)

           * JHLP identical protocal
                                                      Flehinger BJ Am Rev Respir Dis 1984: 130: 550-60

Sputum cytology doesn’t work. For example, in the Memorial Sloan Kettering Lung
Project, 10,000 patients were randomized into 2 groups. The mortality rate, here
noted by the number of deaths from lung cancer, was equal in both. But what can we
conclude about the role of chest radiographs? Nothing! Both the control group and the
experimental group had chest radiographs. When the trials were organized, the main
objective was to find the next PAP smear. Chest radiographs were thought of as the
current “standard of care”. As it turns out, most of the cancers were detected with
chest radiographs.

          Chest Radiograph
                                         Mortality                Stage I/II
                           4,618             133*                 99 (48%)
                          CXR + Sputum

         9,211               Quarterly

                           4,593             119*                 51 (32%)
                          CXR ~50%
        *through 1996
                                     Marcus PM J Natl Cancer Inst 2000: 92: 1308-16

Are chest x-rays worthwhile? The NCI sponsored lung cancer screening programs
at 3 institutions. We’ve just seen the protocol from Memorial Sloan Kettering, which
was identical to the Johns Hopkins Lung Project. Both the experimental and
control group had chest x-ray’s. However, in the MCLP, the control group was
advised to have chest x-rays (which was followed by nearly 50%). It’s from this
contaminated study that the role of chest radiography could be investigated. Even
though the number of lung cancers and the proportion of Stage I or Stage II lung
cancers were increased in those that were screened, there was no decrease in
mortality (in fact, mortality was worse).
                   Chest Radiograph
                                                       Mortality*             Stage I/II
                                      3,172                 247                19 (53%)

                  6,346             0 1   2 3 4    5    6

                                      3,174                 216                 4 (21%)
                                                            *15 yrs

                                                  Kubik AK Cancer 2000; 89: 2363-8

The second randomized trial investigating chest x-rays was performed in Czechoslovakia.
This study had an unusual design intended to eliminate lead time bias, (a subject we will talk
about later) in that the control group also had chest x-rays for the last 3 years of screening.
Similar to the Mayo study, the number of lung cancers and the proportion with early stage
disease was increased. However, mortality did not decrease, in fact it increased in those that
were screened.

        Why is screening controversial?
           CXR screening
             Increase Stage I
             Increase resectability
             Increase 5 year survival

So what did we learn from the last randomized studies. First,
the NCI projects were poorly designed with investigators trying
to answer questions without appropriate data. Titillating was
the increase in percentage with stage I cancers, increase rate
of resectability, and increase in the 5 year survival.

        Why is screening controversial?
          CXR screening
            Increase Stage I
            Increase resectability
            Increase 5 year survival
            No Mortality advantage?

However, there was No mortality advantage and it was
concluded that CXR’s were not worthwhile in screening for
lung caner.

         NCI Lung Cancer Projects
            Flawed design
            Flawed execution
            Flawed analysis

                                Strauss GM Chest 1997; 112: 216S-28S

Before we examine why mortality is important in screening studies,
I’ve already pointed out that these studies are controversial because
of their design. In fact, in the last 20 years, more papers have been
written critiquing these studies than those dealing with the data.
However, the chest x-ray, which the NCI believed was the standard
of care when they designed the studies, is now considered futile in
screening for lung cancer.

        Sobering statistic
           NCI lung projects

             CXR sensitivity for Stage I lung cancer only

                                  Flehinger J Natl Cancer Inst 1988: 80: 337-44

Whether or not chest radiographs are worthwhile is not the
point of this lecture, however, it’s important to point out that
unless we learn how to improve our ability to interpret films,
the low sensitivity of CXR’s for early disease precludes using
this modality to screen for lung cancer.

         Survival vs Mortality
                                       Mortality           5 year survival
                         4,618             133*                 35%
                        CXR + Sputum


                         4,593             119*                 19%

                                 Marcus PM J Natl Cancer Inst 2000: 92: 1308-16

Let’s return to Mortality in screening studies. Survival, which is what
we deal with in individual patients, and makes sense as a meaningful
statistic is misleading. Let’s again look at the data from the MCLP.
Note here the 5 year survival Is almost doubled yet mortality was
unchanged. The difference between survival and mortality will
become of major importance when we discuss current CT screening.

           Lead Time
               False + pathology
               Clinically unimportant tumors

The reason that survival is misleading is the statistical concept
of BIAS. The most important aspects of bias are lead time and
overdiagnosis bias.

                                                   5 year survival Not Screened 0
                                                   5 year survival Screened 40%
             Lead Time



                        Onset              Diagnosis                            Death

         Not Screened

                          0      1     2       3         4       5       6          7 years

                                           Lead Time

First let’s look at lead time. Assume we have 2 groups of smokers with
identical tumors. In the screened group, 2 patients have cancer and a
diagnosis is made while still asymptomatic. However, the tumor has
already metastasized and will eventually cause death. In the second
group the diagnosis is made when symptoms occur. The disease is
advanced. All die at the same time. The 5 year survival in the screened
group is 40% and yet the mortality is the same. The apparent
prolongation of life in the screened group is due to lead-time bias.
        Why is screening controversial?
           Common sense lung cancer “facts”
             “Early” disease curable
             “Small” = early

One of the tenets of lung cancer screening or any cancer
screening is to discover early tumors, prior to the development
of metastases. By early we mean small, how small does small
have to be?

                                                          Diameter (cm)

            Growth Curve                                                       20

                               Doublings                                       2
       1    5    9   13   17        25     29   33   37     41    45      49

Let’s stop and consider the biology of tumor growth and what we mean by “early”. Assume a
single 10-micron cell that divides and grows at a constant rate. The diameter of this tumor will
follow this exponential curve. The growth curve can be divided into 2 phases. In the first
portion of the curve, the tumor is small. Small changes in diameter equate with doublings.
The second phase is what we deal with in clinical practice. The tumor is large at diagnosis
and seemingly enlarges quickly, which reflects the steep portion of the growth curve. the
average size of missed tumors of 1.5 cm is already a far advanced tumor. What we don’t
know is when do metastases typically occur. If they consistently occur before the tumor is 1.5
cm, then radiographic screening strategies may be futile. Sagawa found that 21% of
peripheral lung cancers measuring < 20 mm in diameter are associated with lymph node
metastases (Cancer 74:2239-44, 1994)
                                                                       Diameter (cm)

         Growth Curve                                                                       20

                 Tumor DT typically 100 days
                 Assume it takes 50 doublings before death
                 Typical tumor life span is 13 years
                                      Doublings                                             2
     1   5   9       13      17               25      29     33   37     41    45      49

The normal DT of tumors is approximately 100 days with a
range of 40-400. Assuming that it takes 50 doublings before a
tumor causes death, for our average tumor the life span of a
typical tumor is just over 13 years and a 1 cm tumor is already
nearly 10 years old.

         Growth rate small lung cancers
            Mean DT 452 ± 381 days (52 - 1733 days)
            27 tumors > 450 days DT
            7 Doublings 3 mm to 15 mm
            452 x 7/ 365 = 8.6 years

                             Hasegawa Br J Radiol 73: 1252-1259, 2000

However, screen detected cancers grow very slowly. The mean doubling
time of these 61 tumors found in a mobile screening study was 452 days. It
takes 7 doublings for a tumor to grow from 3 mm in diameter to 15 mm in
diameter. For the average tumor, it would take 8 1/2 years before a tumor
reaches this size. Note that almost 50% of the tumors in the study had a
slow growth rate. For the tumor with a growth rate of 1733 days it would
take 33 years for a tumor to grow to this size. Tumors that grow this slowly
may never become clinically important.
        Why is screening controversial?
          Small = early?
          510 Stage IA (T < 3.0 cm)
          No relationship between size and survival

                            Patz Jr. Chest 2000; 117: 1568-71

Does small equal early? In a review of 510 patients with Stage
IA tumors, there was no relationship between size and
survival. This data was published to support the argument that
small nodule detection with screening CT may not improve
lung cancer mortality.

        Why is screening controversial?
          Small = early?
          510 Stage IA (T < 3.0 cm)
          No relationship between size and survival
          5 year survival 85%
          Only 5% had nodules < 1.0 cm

                            Patz Jr. Chest 2000; 117: 1568-71

However, not emphasized in the report is the good news. As long as
the tumor was less then 3.0 cm in size, the survival curve flattened
out at 85% survival. In addition, only 5% of the Stage IA tumors
were less than 1 cm in size. In these 26 patients there were 4
deaths, overall extremely small numbers with which to make broad
sweeping conclusions about size and mortality.

           False + pathology
           Clinically unimportant tumors

The other component of bias which distorts survival statistics
is overdiagnosis bias. Overdiagnosis bias arises either from a
false positive diagnosis or the diagnosis of clinically
unimportant tumors (also known as length time bias). Either
leads to an increase in survival in the screened population.

         Overdiagnosis bias
                                                       5 year survival Not Screened 0
                                                       5 year survival Screened 40%



                    Onset              Diagnosis                          Death

     Not Screened

                      0      1     2       3       4       5       6       7 years

Let’s look at overdiagnosis bias. The screened and nonscreened
group are the same. Both groups have a “clinically unimportant
tumor”. The tumor is discovered in the screened group. The patient
survives. The clinically unimportant tumor is never discovered. The
screened group has better survival (20% vs 0) due to the discovery
of the clinically unimportant tumor. The mortality of 20% is the same
in both groups.

           False + pathology?
               Multicentric adenocarcinoma
                  158 resections
                  22% had more than 1 tumor
                  100x reported rate
                  Reason: “Bouin” fixative

                                           Zwirewich Radiology 1990 176: 185-190

Is it really possible to have false positive pathology? Perhaps that’s what happened in
this study. In patients who underwent resection for lung cancer, 22% (a rate almost 100x
that reported previously) had multiple tumors in the lobe or lung resected. The high
detection of tumors was ascribed to the use of a special fixative that made the tumors
more conspicuous. However, this is at variance with clinical practice where second
tumors (at CT for example) are relatively rare. Think about your own practice. In patents
with unresectable disease are you seeing second and third tumors in the same lobe?
One wonders whether some of these tumors were really false positive diagnoses.

             Clinically unimportant tumors?

                 Yale autopsy study
                   1/6 of cases were “surprise”
                   Prevalence “surprise” in men 1%
                   Probably underestimate

                                       Chan et al. Chest 1989;96(2):291-6

The second component, clinically unimportant tumors seem more
certain. Autopsy studies provide irrefutable evidence that many
individuals die with rather than from lung cancer. In a 30-year review of
all adult autopsies at the Yale New Haven Hospital , about one in six
lung cancers observed at autopsy had not been recognized premortem.
In the 10 most recent years of the review, about 1% of men had had
previously unsuspected lung cancer, most cases of which were
resectable and presumably asymptomatic.

       Clinically unimportant tumors?
                 CT screening (rural mobile scanner)

                  n     Lung Ca   adenoca   Size    Ia   GG     Women        Men

    Nonsmokers   4251     45        22      12.4    22    38       36         9

    Smokers      3596     39        4       18.2     7    16        0        39

                                            Feng Acad Radiol 10:1013, 2003

Lung cancer is rare in nonsmokers, thus it is interesting that these investigators
studied a population, the majority of which were nonsmokers The 1% detection rate
of lung cancer was identical in both groups. Most of the lung cancers in
nonsmokers occurred in women and were smaller Stage I slow growing
adenocarcinomas with a ground glass appearance characteristic of
bronchioloalveolar cell carcinomas. What’s lacking in this study is the definition of
nonsmokers. Whether these individuals were previous smokers or were exposed to
second hand smoke or other known carcinogen is unknown. Nevertheless, the high
cancer detection rate in non-smokers supports the idea that many lung cancers will
not become symptomatic during a patient’s lifetime.                                     28
        Clinically unimportant tumors
          Doesn’t make sense
             Consequences of not treating
               45 patients had Stage I lung cancer unresected
               Only 2 (4.4%) survived 5 years

                                       Flehinger Chest 1992: 101:1013-18

In counterpoint however, although there may be or clinically
unimportant tumors, in the NCI sponsored lung projects, there
were 45 patients with Stage I disease that declined treatment.
Only 2 survived 5 years. This would suggest that there is no
such thing as an unimportant lung cancer.

        Why is screening controversial?
          Nonrandomized trials
            No control of bias
               Lead time

Well as you an see, results from nonrandomized results are
subject to error because of bias and the results are
controversial. What may seem favorable may not be so. We
must keep this in mind as we proceed to current CT studies.

         What’s the rationale for CT?
            CT more sensitive than CXR
            CT standard for detecting metastases
            CT? screen for lung cancer

The rationale for screening CT is straightforward. We know that CT
is inherently more sensitive than chest radiographs with at least a 10
fold increase in contrast sensitivity. CT has long been the standard
for detecting pulmonary metastases. Because the sensitivity of chest
radiographs for stage I lung cancer is poor, doesn’t it seem logical to
try CT as a screening tool?

           Anti-lung cancer association
                          Stage I    Mean Diam (mm)
                 Pre CT    53%             30 ± 14

                Post CT    93%             16 ± 7

                                     Kaneko, Radiology 1996

The first bit of evidence came from Japan. When CT
screening was added to a preexisting screening protocol
which used chest radiographs, the proportion with Stage I
disease increased dramatically, fulfilling the requirement that
CT detect a high proportion of early stage disease.

             ELCAP Prevalence
                                         Malignant              Stage I
                         233                  27                23 (85%)
                                               7 identified on chest x-ray
        >60                                    14 mm mean diameter
        >10 pk-yrs

                                        Henschke, Lancet 1999

In the US, this spurred Dr. Henschke to undertake the Early Lung
Cancer Action Project. Not surprisingly, CT was more sensitive in
detecting cancer finding 4 times as many cancers as found at chest
radiography. Note that these studies were nonrandomized, without a
control group. Similar to what we saw earlier, the increase in Stage I
disease may not equate with a decrease in mortality.

              ELCAP Incidence
                                           Malignant              Stage I
                           30                   7*                5 (71%)
                                                12 mm mean diameter
         Repeat                                      * 2 interval endobronchial

                                          Henschke, Cancer 2001

After the prevalence screen, the incidence cases continue to show favorable
stage distribution. And now we have data concerning interval tumors, that is
those tumors which become known between the screening interval. There
were no interval parenchymal tumors though there were interval
endobronchial tumors. How does this compare to the NCI studies? There,
approximately 50% of the tumors were interval cases. Obviously it is difficult
to do screening if a large majority of the cases arise between scheduled
screens. Is this enough evidence to suggest that we screen?

          13% (4 of 30) nodules missed
          71% (5 of 7) Stage I resectable

                                         Henschke, Cancer 92, 2001

However, this study also raises serious questions - 13% of the
lung cancers were evident on previous studies and the miss
rate for all nodules was 50% and they specifically used the
term resectable. Were they resected?

             13% (4 of 30) nodules missed
             71% (5 of 7) Stage I resectable
             What they also failed to mention
                63% (5 of 8) cancers missed
                Overdiagnosis: 35% (3 of 8) of cancers
                  1 probable cancer died before diagnosis
                             122 day doubling time
                  2 Stage I not resected due to poor health

                                                              Henschke, Cancer 92, 2001

On closer examination the possibility of overdiagnosis is raised. One patient
had a ground glass nodule that grew in one year from 4 to 8 mm but died
prior to diagnosis. With a doubling time of 122 days this was probably a
cancer, but since it wasn’t proven it was not included in the results. In
addition, 2 Stage I lung cancers were not resected because of poor health.
One could argue that if you can’t treat for cure that this would represent a
clinically unimportant tumor. Thus up to 1/3rd of the lung cancers could
potentially be overdiagnosis cases. And remember the term resectable? Of
the 6 nonsmall lung cancers, 1/3rd did not have definitive surgical resection.
           Mayo Prevalence
                                     Malignant               Stage I
                      782                23                   14 (56%)
                     Nodules (51%)        2 sputum cytology alone
                                          17 mm mean diameter

                             Swensen Am J Respir Crit Care Med 2002; 165: 508-13

Similar to the ELCAP design, the Mayo clinic study comprised
smokers over the age of 50 with minimum of 20 pk-year
history of smoking. Two facts are striking, the stage
distribution is not as favorable and the large number of
patients with indeterminate nodules. However, even though
the stage distribution was not as favorable 22 of the 23
patients underwent resection for cure.
           Mayo Incidence
                                             New nodules
                                 375                    191 (13%)
                               Nodules Missed
       1,464                                            no interval cancers

               1000 patients with nodules or 66% original population

                                       Swensen Am J Respir Crit Care Med 2002; 165: 508-13

97% returned for their yearly incidence scan. In 375 patients,
nodules were visible in retrospect on their prevalence scan. 27
patients were informed that a nodule was missed on their
previous normal screening CT. At year 1, the proportion of the
population with nodules increased 10%. In total, nearly 60% of
the screened population now have nodules.

           Mayo CT Screening
              29 Resections
                7 for benign disease
                  5 grew

                                Swensen Am J Respir Crit Care Med 2002; 165: 508-13

The consequences of false positives includes resections for
benign disease. 5 of the 7 benign nodules grew. One of the
tenets of screening is not to harm those with no disease.
Obviously, if a large number of patients undergo resection for
benign disease then screening will not be favorable.

               29 Resections
                 7 for benign disease
               Incidental significant findings (14%)
                 2 carcinoids
                 4 renal cell cancers
                 3 breast cancers
                 2 lymphomas
                 2 gastric tumors
                 1 pheochromocytoma

                                 Swensen Am J Respir Crit Care Med 2002; 165: 508-13

More Controversy! In the Mayo Clinic protocol scanning went from
the mid neck to mid abdomen, purportedly to make sure that the
technologist included the entire lung. Thus they performed a nearly
total body scan. Note, the frequency of significant incidental findings
including numerous other asymptomatic tumors. Are these all
unimportant tumors?

        False Positives

                       Japan   New York   Mayo
                       10%       25%      50%

           Technique   10 mm    10 mm     5 mm
                       2:1      2:1       reconstruct 3.5 mm

The Mayo Clinic study has by far the largest number of false
positive nodules. Partly, this is due to the high prevalence of
Histoplasmosis in the Midwest. Part may also be due to the
screening technique, which is more sensitive than previously
used protocols for nodules.

           Indeterminate nodule = 3 CT’s
           $75 Billion

While most patients with nodules can be managed non-
operatively, significant costs accrue from the multiple follow-
ups that are required to exclude growth. Investigators at the
Mayo Clinic have estimated the cost of screening smokers in
the US based on their experience and it’s a whopper.

                                      Mayo Clinic Lung Cancer Screening Study Recommendations

                                                                    Clear         Protocol
     If Nodule Indeterminate

     Choose an option (mouse click)                                    Thin-section CT (1-3 mm collimation)
       New                                                             @ 3 months
             Single or multiple nodules < 3 mm
             Single or multiple nodules, one at least 4 - 7 mm
             Single nodule 8 - 20 mm                                Benign            Indeterminate       Growth
             Single nodule > 21 mm                                                                        Biopsy or 3-month followup

             < 7 mm nodule stable 1 year                                         Thin-section CT @ 3 months
             8 - 20 mm nodule stable 1 year

                                                                 Growth-                                  Stable-
                                                                 Biopsy or 3-month followup               Thin-section CT q 3-12 months

                                                                             Stable-                          Growth-                       Decrease
                                                                             Thin-section CT q 6 months       Biiopsy or 3-month followup   Benign

The protocol for screening is complicated. Here is the
algorithm for the nodule at least 4 mm in diameter.

          Scanner       Multislice    Single slice
          Collimation   5 mm          7 mm
          Recons        3.5 mm        3.5 mm
          Table         30mm/sec      2:1 pitch
          Tube          40 mA         80 mA

           Dose 0.65 mRem (5.8 mRem conventional)

Technique with modern scanners. The general trend is to
decrease the slice thickness. Of importance is the decrease in
mA. This markedly reduces dose to 1/10th that of conventional

              False negatives

                    Mayo Clinic       Anti-Lung Cancer Association
                   375 (25%) missed     22 cancers/ 1443members
                        231 < 4mm       7 missed DT (146-589 days)
                         6 > 8 mm       mean 8 mm

                                            Kakinuma Radiology 1999: 212: 61-66

CT is not perfect. False negatives are also a problem. In the Mayo Clinic, 25%
of the patients have nodules that were missed. While most of these are small,
several of the nodules were greater than 8 mm in size. In Japan, 1/3 of the
incident lung cancers were visible in retrospect on earlier exams. The latter
data just adds to the controversy. All 7 tumors were growing, most at the
expected rate of lung cancers, suggesting that they would have eventually
evolved into clinically significant lung cancers and were not clinically
unimportant tumors.
       Missed Lung Cancer

This is an example of a missed cancer. Not the large cavitary
carcinoma but the other one, can you find it?

                 9 mos later

Nine months after a pneumonectomy, the second tumor
becomes evident.

        Growth Pitfall

         Oct 99                      Jan 2000

                          Nov 2000              Mar 2001

Screening isn’t easy, let’s review this patient. At a yearly incidence screen there was a
new nodule in the left lower lobe. According to protocol, this 4 mm nodule was studied at
3 months with thin sections. The nodule is indeterminate with no calcification. While the
patient should have returned in 3 months the patient was seen 11 months later. At this
point the nodule has grown. However, instead of surgery, the clinicians did a PET study
which was negative. Feeling comfortable with that test they waited. 4 months later a
repeat study shows that the nodule has decreased in size. Now what would you do?
Because it had grown and had irregular borders I pressed for resection: Diagnosis: BAC
diagnosed 17 months after detection

          CT screening summary
       Study              n      Indeterminate   Lung cancer   Stage I   AdenoCa
         Henschke         1000            233             27        23        22
               Sone       5483            279             23        23        19
         Diederich        817             409             11         6         6
               Nawa       7956           2099             36        31        35
               Sobue      1611            186             14        10        10
          Swensen         1520            782             21        14        15
               Tiitola    602             111              5         1         2
                         18989           4099            137      108        109

Numerous 1 arm studies, many of them patterned after ELCAP have
now been performed, examining almost 19,000 patients. A high
proportion of these cancers are stage I cancers. Out of proportion to
that seen in clinical practice, 80% of these cancers are
adenocarcinomas. The number with indeterminate nodule, many of
which will be false positives varies but is as high as 50% in the
studies from Swenson and Diederich.

                   Randomized study

                                         CT Blood sputum

                                                      10 years
                                                      200 million dollars
                   > 55
                   > 30 pk yrs

                                                               Aberle, Chief Investigator

The only sure method with which to show a mortality reduction is a randomized trial. This is the ACRIN
protocol for lung cancer screening being sponsored again by the NCI. Patients older than 55 with a 30 pk-
year history of smoking will be randomized into either having chest x-ray screening or CT screening. Blood
and sputum will also be collected. The $200 million dollar price tag makes this the most expensive
screening study to date. One criticism right off the bat is the use of chest x-rays in the control group, versus
doing nothing which is the current clinical practice.
It will be more than 10 years before these results are known. Will the study be successful? The control
group is contaminated by having chest radiographs. Also I should point out the experience of
mammography. “Definitive” randomized trials are usually not definitive but controversial. None have been
done without some criticism of their methods and protocol.
In addition, another factor will complicate this trial. In nearly
every metropolitan area there are entrepreneurs who offer CT
screening at an affordable price. Keeping the control group
from becoming contaminated will be even more difficult.

        Congressional Testimony
          We are most concerned…
          …most expensive screening trial
          …unlikely to provide an answer
          …has the same design flaws
          …we have been ignored

                  Statement Claudia Henschke, Aapril 16, 2002 Before the House Committee on Ways and Means

Before the NCI trial even began there was controversy.
Debate was conducted in a most unusual scientific forum, Dr
Henschke testified before Congress strongly criticizing the
proposed NCI trial.

         What should I do?
                  CT is valuable and will save lives
                  Offer unrestricted CT screening
                  CT no proven benefit
                  Don’t offer screening

Should you screen? As you can see, there is no right answer. Given
the controversy and the data both pro and con what should you and I
do. We’ve seen evidence that screening is worthwhile. Many patients
will give testimonials that their lives have been saved with screening.
On the other hand, it’s not clear that the favorable data does not
suffer from bias which plagues non-randomized studies. In the mean
time, patients will be asking for it and are willing to pay for it. What
should we do?
         What should I do?
                 Early is better
                 Offer unrestricted CT screening
                 No proven benefit
                 Don’t offer screening
           Let the patient decide
                 What are the benefits vs. risks?
                                             Earnest Radiology 226: 633-34, 2003

Most of the debate between the proponents and opponents of
screening have ignored the patient. The focus has been on the
physician, “yes we should, no we shouldn’t”. This is very
paternalistic, “we know best, we will decide”. However, with the
Internet, patients are much too knowledgeable and involved in their
care to be ignored. With evidence both Pro and Con, our job is to
take the information which we presently have and use the concept of
informed consent to educate the patients about the risks and benefits
of screening.                                                                      54
        Informed Consent
               Will screening save my life?
               Will screening cause me harm?
               What happens if I don’t get screened?
                                     Earnest Radiology 226: 633-34, 2003

Informed consent is a long used technique to counsel patients
about risky procedures or therapies of unproven benefit.
These are the elements and the essential questions that the
patients will have about lung cancer screening.

                        Benefits from Screening

                          Reassurance that they don’t have cancer
                             CT not perfect and will miss disease
                          Chance for cure

                                                          Earnest Radiology 226: 633-34, 2003

What are the Benefits. Patients get fair reassurance that they don’t have cancer with the
obvious qualification that CT is not perfect. Remember that in the Japanese study, 1/3 of the
incidence cancers were visible in retrospect. Whether this changes mortality is not known. It’s
important to realize that these were not INTERVAL cancers. In the NCI projects, 50% of the
tumors were interval cancers. Patients will not be happy if they develop lung cancer between
screening. In the limited ELCAP data, interval cancers were restricted to endobronchial
disease. The numbers are small (2 of 9) but this would equate to 22% of all incidence tumors
discovered as interval cases.

            Harms from Screening
                 Anxiety about lung cancer
                     Highly likely that a nodule will be found
                     Unnecessary diagnostic procedures
                        including surgery
                 Mortality benefit unknown
                     Survival benefit may not be real
                     Radiation risk
                                           Earnest Radiology 226: 633-34, 2003

There are harms, First is anxiety as more than 50% will have a
nodule which will require multiple follow-ups and potentially
surgery. In addition, we don’t know if the tumor will be caught
early enough to make a difference in longevity. Lastly, there is a
small but definite radiation risk.


           Value of lung cancer screening is unknown
           We can’t avoid the question
           If you screen, use informed consent to
              advise the risks and benefits as currently

Should you screen is a question which I cannot answer for you, but
we cannot ignore the question. There is evidence that screening may
be successful. There is evidence that screening is futile. However, it
will be many years, if ever, before there is a definitive answer to this


Shared By: