1 Zeidler M, Sellar RJ, Collie DA, et al. The treatment of the samples4 we found out Pediatrics, First Department of Pediatrics,
pulvinar sign on magnetic resonance imaging Budapest, H-1083, Hungary; First Department of
the GNB3 C825T polymorphism Pediatrics and Second Department of Gynecology
in variant Creutzfeldt-Jakob disease. Lancet
2000; 355: 1412–18. according to Siffert and colleagues.2 and Obstetrics, Semmelweis University, Hungary;
2 Will RG, Ironside JW, Zeidler M, et al. A Continuous variables were compared by and Buda Children’s Hospital, PKU and Metabolic
Screening Center, Budapest
new variant of Creutzfeldt-Jakob disease in two-sided t tests. Categorical data were (e-mail: email@example.com)
the UK. Lancet 1996; 347: 921–25. assessed by 2 test.
3 Chazot G, Broussolle E, Lapras CI,
Birthweight, gestational age, and 1 Hocher B, Slowinski T, Stolze T, Pleschka A,
Blattler T, Aguzzi A, Kopp N. New variant
of Creutzfeldt-Jakob disease in a 26-year-old genotype of full-term and preterm Neumayer HH, Halle H. Association of
French man. Lancet 1996; 347: 1181. neonates are given in the table. The T maternal G protein b3 subunit 825 T allelle
with low birthweight. Lancet 2000; 355:
4 Bahn MM, Kido DK, Lin W, Pearlman AL. allele frequency of the control and 1241–42.
Brain magnetic resonance diffusion VLBW subjects did not differ. The
abnormalities in Creutzfeldt-Jakob disease. 2 Siffert W, Rosskopf D, Siffert G, et al.
Arch Neurol 1997; 54: 1411–15. relative number of TC and CC Association of a human G-protein beta3
genotypes did not differ between control subunit variant with hypertension. Nat Genet
1998; 1: 45–48.
infants, infants of appropriate weight for
3 Curhan GC, Willett WV, Rimm EB,
G protein in very low birth- gestational age, and infants small for Spiegelman D, Ascherio AL, Stampfer MJ.
weight infants gestational age. There was also no Birth weight and adult hypertension, diabetes
difference in the birthweight of babies mellitus, and obesity in US men. Circulation
with TC and CC genotype (mean [SD]: 1996; 94: 3246–50.
Sir—Berthold Hocher and colleagues
4 Szalai Czinner A, Révai K. The frequency of
(April 8, p 1241)1 report, that the birth- 1230  g and 1197  g, medium chain acyl CoA dehydrogenase
weight of otherwise healthy full-term respectively). This finding is in G985 mutation in Hungary. Eur J Pediatr
neonates is lower, when the maternal G accordance with the observation of 1996; 155: 256.
protein 3 subunit 825T allele (GNB3 Hocher and colleagues, who did not find
825T) is present. They did not find any any effect of presence of GNB3 825T on
association between the genotype of the birthweight in healthy full-term babies.
babies and their birthweight. However, The TT genotype was more frequent Information on
premature babies were excluded from in the control group (p<0·01), than in
their study. VLBW babies. Samples for neonatal thiazolidinediones
The presence of GNB3 825T is screening programme are collected after Sir—Kristina Schoonjans and Johan
associated with higher risk of the fourth postnatal day from VLBW Auwerx stated in their review (March
hypertension.2 The prevalence of babies and we could not recruit those 18, p 1008)1 of the thiazolidinediones (a
hypertension is also higher in those who who died during the first postnatal week. family of diabetes drugs) that “much of
had been born with a low birthweight.3 Therefore, we speculate that the absence the information discussed here has had
In our study, we investigated, in very low of TT genotype in VLBW infants might to come from company websites”. We
birthweight (VLBW) infants (birth- be attributed at least in part to a higher sympathise with the authors’ difficulty in
weight <1500 g), whether their reported susceptibility of VLBW infants with TT finding peer-reviewed data on these
higher adulthood prevalence of genotype towards perinatal death. To drugs. This stems from the failure of
hypertension is associated with a higher answer this question studies are needed drug companies to publish their
frequency of GNB3 825T. to analyse the survival rate of VLBW findings—it appears that only one of 11
We reviewed the medical records of infants with different GNB3 efficacy studies submitted to the US
142 VLBW singleton infants (gestational polymorphisms. Our data, the Food and Drug Administration (FDA)
age 26–36 weeks). They were born in significance of which needs to be as part of the approval packages for
1999 and treated at the Perinatal clarified, indicate that the high rosiglitazone and pioglitazone has been
Intensive Care Unit at the second prevalence of hypertension in people published.
Department of Obstetrics, Semmelweis who were premature at birth is probably However, drug company websites are
University, Semmelweis, Hungary. We not associated with the higher frequency not the only, and certainly not the best,
contacted the Neonatal Metabolic of G protein 3 subunit (825T) allele. place to find unpublished data on drugs.
Screening Programme and asked for Consulting readily available US
This study was supported by OTKA Grants
dried blood-spot samples of the chosen T031850, F032024. We thank Andrea Horvárth governmental sources on the internet
babies. The samples of 97 babies were for the collection of samples and Ágnes Czárán can lead to different conclusions
available: 26 were born small for and Szilvia Kruspér for technical assistance.
regarding drug safety and efficacy. Using
gestational age (per centile weight <5%), *Barna Vásárhelyi, István Kocsis, FDA data sources, we have just
while 71 were of appropriate weight for Ágnes Schuler, András Nobilis, completed our own assessment of the
gestational age. Dried blood spots of 101 Tivadar Tulassay thiazolinediones. This research resulted
term neonates (birthweight >2500 g) *Joint Research Programme of the Hungarian in a petition to the FDA to relabel all
served as controls. After the pre- Academy of Sciences and the First Department of
three drugs to state that they are not only
less efficacious than already-approved
drugs, but appear to have class toxicities,
Control (n=101) VLBW
particularly cardiac failure (www.citizen.
SGA (n=26) AGA (n=71) org.hrg/PUBLICATIONS/1514.htm;
Mean (SD) birthweight (g) 3421 (376)* 1240 (254) 1195 (244) accessed June 23, 2000), a serious
Mean (SD) gestational age (weeks) 40 (2·4)* 33 (2·7) 29 (1·7) finding not previously adequately
acknowledged in the product labelling or
TT 16* 0 0 on the company websites.
TC 42 21 46 These conclusions could never have
CC 43 5 25 been reached had we depended upon
Frequency of GNB3 825T 37% 40% 32% the company websites alone. Unlike the
*p<0·01 compared with the VLBW infants; VLBW=very low birth weight; SGA=small for gestational age; AGA=appropriate company websites, which provide only
for gestational age. the product labelling and superficial
Birthweight, gestational age, and genotype of VLBW infants and control infants information about the drugs, the FDA
254 THE LANCET • Vol 356 • July 15, 2000
For personal use only. Not to be reproduced without permission of The Lancet.
website includes detailed summaries and Recognising lesions on number of adenomas found other
assessments of the actual data submitted than to say that polyps were found in
in support of the New Drug Application. colonoscopy 26·9% of cases. In comparison,
Reviews of pre-clinical studies and Sir—We agree with B J Rembacken and adenomas were found in 32·1% of
clinical trials submitted to the FDA colleagues (April 8, p 1211)1 that examinations in Leeds. A possible
for troglitazone, rosiglitazone, and training in the recognition of flat, reason for this discrepancy is that dye
pioglitazone were available to elevated, and depressed lesions at spraying techniques were not used in
Schoonjans and Auwerx through a colonoscopy is important in the Glasgow.
Freedom of Information Act (FOIA) detection of early colorectal cancer. In contrast to the series from
request (www.fda.gov/foi/foia2.htm; However, we suspect that these lesions Glasgow, which excluded lesions with
accessed June 23, 2000). For some have been observed previously by UK severe dysplasia or intramucosal
drugs approved after 1997, reviews can colonoscopists, but not reported as carcinoma, we included these lesions in
be downloaded directly from the such. In a study of over 2000 our analysis. We referred to these lesions
agency’s web site (www.fda.gov/ colonoscopies done between 1990 and as early cancers, because failure to
cder/approval/index.htm; accessed June 1998 we identified nine flat or depressed detect and treat such lesions is likely to
28, 2000). Use of the FOIA is not cancers, three of which were less than lead to advanced colorectal carcinoma.
limited to US citizens. 10 mm in diameter.2 This compares This was clearly stated in our results
In the past, FDA clinical, statistical, with four flat or depressed cancers in the section.
and pharmacology reviews were only series of 1000 colonoscopies done by In their article, Smith and colleagues
made publicly available after a drug was Rembacken and colleagues. Given the classified lesions as pendunculated,
approved, severely limiting scientific indications for colonoscopy in the study villous, and flat compared with
debate over the appropriateness of drug by Rembacken and colleagues we were polypoid, depressed, and flat in our
approval. As a result of a lawsuit we surprised that only 3% of patients had a study. It is therefore interesting to learn
brought against the agency, the materials colorectal cancer diagnosed. As well as that a similar proportion of early
distributed to FDA Advisory Committee training endoscopists in recognition of colorectal cancers (T1 stage) where flat
members before approval, including lesions, it is also critical that clinicians or depressed in Glasgow (nine of 18)
summaries of the FDA reviews, are now are trained to target this important compared with Leeds (four of six).
posted on the internet the day before resource to the right population, in However, the recognition that early
the Advisory Committee meeting respect to age and indication. Mean age cancers may appear flat or depressed has
(www.fda.gov/ohrms/dockets/ac/00mtbc of patients undergoing colonoscopy in important implications beyond semantic
.htm; accessed June 23, 2000). The the study by Rembacken and colleagues definitions of flat, villous, and sessile.
transcripts of these FDA Advisory was only 59 years . Until now, most early carcinomas were
Committee meetings are another thought to be polypoid. From the
G A Smith, S MacKenzie, P J O’Dwyer
valuable source of information realisation that non-polypoid neoplasia
Department of Surgery, Western Infirmary,
(www.fda.gov/foi/electrr.htm; accessed Glasgow G11 6NT, UK account for a large proportion of early
June 23, 2000). colorectal cancers follow two important
The failure of Schoonjans and 1 Rembacken BJ, Fujii T, Cairns A, et al. Flat conclusions. First, endoscopists should
Auwerx to consult these free sources, and depressed colonic neoplasms: a be trained in the techniques necessary
and their reliance instead on self-serving prospective study of 1000 colonoscopies in for their detection and management.
the UK. Lancet 2000; 355: 1211–14.
data from the company, seriously Second, gastroenterologists and hospital
2 Smith GA, Oien KA, O’Dwyer PJ.
undermines the credibility of their Frequency of early colorectal cancer in managers may reconsider the role of
review. patients undergoing colonoscopy. Br J Surg radiology in investigating the large
1999; 86: 1328–31. bowel as contrast studies and a
Public Citizen’s Health Research Group is a
non-profit, consumer research and advocacy substantial proportion of cancers in their
early and curable stages may be missed
*Sidney M Wolfe, Peter Lurie, Sir—In planning our study, we decided by virtual colonoscopy.
Larry D Sasich, Elizabeth Barbehenn not to preselect patients to allow We disagree with the implication that
Public Citizen’s Health Research Group, comparisons between different units colonoscopy is a limited resource that
1600 20th Street NW, Washington, both within and outside the UK. It is should be targeted on elderly patients.
DC 20009-1001, USA Colonoscopy is superior to radiological
not surprising that in the unusual series
done in Glasgow, UK, by G A Smith methods in detecting early cancer and in
1 Schoonjans K, Auwerx J. Thiazolidinediones:
an update. Lancet 2000; 355: 1008–10. and colleagues2 there was a slightly the diagnosis of such mucosal diseases
higher proportion of colorectal cancers as angiodysplasia, inflammatory bowel
Author’s reply disease, and infective colitis.
detected (4·3% in Glasgow vs 3·1% in
Sir—We strongly disagree with the Leeds). As the indications for Colonoscopy remains the standard for
statement that our review relies on self- colonoscopy were similar, the most investigating the lower gastrointestinal
serving data from companies. Our likely reason for this difference is that tract. Furthermore, patients undergoing
review was based on all available sources patients in the study by Smith and colonoscopy only require one
of information, including the sources colleagues were on average 16 years investigation for the examination of the
quoted by Sidney Wolfe and colleagues. older (median age 77 years in Glasgow colon rather than two (sigmoidoscopy
The sources quoted by Wolfe and vs 61 years in Leeds). In fact, no and barium enema). We believe that
colleagues did not provide us with any patients below age 60 years were greater resources should be channelled
additional information that would have investigated with colonoscopy in the into colonoscopic services.
made us change any of the conclusions series from Glasgow.
*B J Rembacken, T Fujii, S Yoshida,
reached in our review. Paradoxically, although the patients
D M Chalmers, A T R Axon
in Leeds were almost two decades
younger, more adenomas were found in *Centre for Digestive Diseases, The General
Institut de Genetique et Biologie Moleculaire et Infirmary at Leeds, Leeds LS16 8LT, UK; and
Cellulaire (IGBMC), 67404 Illkirch, France this population. In their article, Smith National Cancer Centre, Tokyo, Japan
(e-mail: firstname.lastname@example.org) and colleagues do not elaborate on the (e-mail: BJR@firstnet.co.uk)
THE LANCET • Vol 356 • July 15, 2000 255
For personal use only. Not to be reproduced without permission of The Lancet.