Year: 2009
Type: letter
Abstract: Sirs, Yuan et al. report that the incidence of endoscopic ulcers reported in the placebo arms of controlled studies has not changed significantly over three decades.1 They ascribe the fact that endoscopic ulcers occur at all to the mysterious placebo effect. We have a different explanation for the results that we believe better explains the data and is in accord with the early studies (i.e. studies from 1975 to 1989), which reported a 0% incidence of endoscopic ulcers in the placebo groups. In many clinical trials of ulcer disease, the outcome (e.g. time to healing, or pain relief) is variable so that the mean for any group is expected to vary. An identical placebo is therefore used to ensure that the result with the test agent actually differs from what would have occurred by chance (i.e. without the intervention). It is well recognized that there is something called a placebo effect which is an effect in addition to chance; however, it is not clear whether placebo effect is a reasonable cause of endoscopic ulcers.2 In most clinical studies using potential gastrotoxic drug, subjects undergo an initial endoscopy to confirm that they have a normal upper gastrointestinal tract. They then receive a gastrotoxic agent known or suspected to cause damage or a placebo and after a suitable interval undergo repeat endoscopy to assesses and score any damage as minor or major. Ulcers are characterized as major damage. Hundreds if not thousands of normal individuals have undergone upper endoscopy clinically and as part of clinical trials. Even in the Helicobacter pylori era, ulcers were rarely found in asymptomatic individuals.3 As ulcers were found albeit rarely, it was considered prudent to perform pretest endoscopy. Placebo was initially used to confirm that the experiment was not flawed (i.e. H. pylori was transmitted causing acute infection, the patients were not unreliable and took undisclosed NSAIDs, etc). In the early studies, there was no concern regarding whether the endoscopist was reliable in terms of their observations. Many early studies were conducted in Houston by Lanza or his disciples who learned from him how to evaluate gastrotoxic agents. Those endoscopists were also calibrated against one another to ensure that there would be agreement and that the data would be accurate and additionally, the findings were documented by photographs. This attention to detail differs from many multicentre clinical trials which use with different observers of unknown skill. For example, one Merck study reported by Laine et al. had two follow-up endoscopies (6 and 12 weeks) and 82 investigative sites with an unknown number of endoscopists. At each interval, approximately 3% of procedures reported endoscopic ulcers using the ≥3 mm definition and 2% using the ≥5 mm definition.4, 5 At the time we suggested that this was possibly because of the background inaccuracy of the procedure itself (i.e. the average practising endoscopists were probably not more than 95% accurate).5 It still seems to us that most experienced investigators would respond with disbelief if 100 normal asymptomatic individuals, not taking any gastrotoxic medications, had upper endoscopy with confirmed normal mucosa, 2 or 3% would have an ulcer when endoscoped after a short interval. The explanation that they were taking placebo and that it was responsible is not likely be considered a plausible explanation. However, we agree with their conclusion that placebo controls are needed. We disagree on why. We believe that placebo provides an estimate of the error of endoscopy as a technique to assess damage. This error could probably be reduced. One approach would be to use only expert calibrated endoscopists. Another is to utilize the advantages of this video age where whatever is seen can be captured and reviewed. There is little justification for continuing to exclude modern documentation in such trials. However, accurate identification of endoscopic ulcers may not be clinically important considering that as a surrogate marker for severe outcomes, endoscopic ulcers have never been validated (i.e. one that could be used as a replacement endpoint in trials designed to provide reliably accurate estimates of the level of tangible benefit to patients).6, 7 Nevertheless, endoscopic ulcers have long proved useful for marketing purposes and are therefore expected to be used continuously.8 Declaration of personal interests: Dr Graham is an unpaid member of the executive committee of the PRECISION trial which is supported by Pfizer and is designed to compare the cardiovascular safety of celecoxib, naproxen and ibuprofen in higher risk cardiovascular patients. Dr. Graham is also a consultant for Novartis in relation to vaccine development for treatment or prevention of H. pylori infection and a paid consultant for Otsuka Pharmaceuticals, manufacturer of the 13C-urea breath test. In the past 3 years, Dr Chan received a consulting fee from Pfizer and lecture fees from Pfizer, Eisai, Takeda and AstraZeneca. Dr Chan is also Chairman of the steering committee of the CONDOR trial that is sponsored by Pfizer. Declaration of funding interests: None.
Authors: D. Y. Graham, F. K. L. Chan
Language: English
Subfield: Gastroenterology
Field: Medicine
Domain: Health Sciences
Open Access status: bronze
Funders: Pfizer, AstraZeneca, Eisai, Otsuka Pharmaceutical
OpenAlex ID: W2002598013