Nissen vs Peto: Smackdown, New England-Style
In this year’s brand-spanking-new issue of the NEJM, cardiologist and consistent Vytorin critic Steven Nissen officially charges that September’s ezetimibe-cancer analysis of Sir Richard Peto et al, in which the authors concluded that the cancer risk with ezetimibe is not increased, “raises disturbing scientific and ethical questions.” The Peto analysis was performed in response to a finding in the SEAS trial that ezetimibe, one half of the combo Vytorin pill, may increase the risk of cancer. However, Peto et al found that unblinded adverse-event data from the much larger, ongoing SHARP and IMPROVE-IT studies did not confirm the finding.
Among his criticisms of the Peto analysis, Nissen cites the following in his correspondence:
- The premature unblinding of the ongoing trials, which is “not a reliable approach to the evaluation of drug safety.”
- The insufficient exposure to ezetimibe in IMPROVE-IT for assessing drug hazards.
- The failure of Peto et al to report the relative risk of cancer mortality in the 2 studies.
- The limited statistical power of the 2 studies to assess the risk of cancer death with ezetimibe.
Elsewhere Nissen has argued that data from all 3 studies should have been combined to assess cancer risk with the drug.
In their reply, Peto and coauthor Rory Collins stress the importance of separating data in a hypothesis-generating trial—in this case, the SEAS trial—from data in hypothesis-testing studies. They implicitly rationalize the unblinding of cancer data from SHARP and IMPROVE-IT by noting that these are the only large, randomized data sets from which relevant cancer data may be obtained to confirm any increased risk with ezetimibe. They also cite the magnitude of these studies in relation to the SEAS trial; together SHARP and IMPROVE-IT generated 4 times as many cancers as the SEAS trial, but their data did “not suggest any increase in cancer incidence.”
Peto and Collins also correct Nissen in his mistaken charge that the authors did not report the relative risk of cancer mortality (and the associated 95% confidence interval) in SHARP and IMPROVE-IT. (The data were right there in the Fig. 4 legend.) Although there were more deaths due to cancer with ezetimibe in the analysis, the difference was not statistically significant (P = .07).
Last, although Nissen did not allege an industry-related conflict of interest by Peto et al (and Nissen’s hardly in a position to do so), Peto and Collins, both from the UK, cite their not-so-dispassionate response to the US Congressional Committee on Oversight and Investigations, which raised the issue.
IMPROVE-IT = Improved Reduction of Outcomes: Vytorin Efficacy International Trial; SEAS = Simvastatin and Ezetimibe in Aortic Stenosis; SHARP = Study of Heart and Renal Protection.
Public-domain image of Hamilton-Burr duel, which admittedly took place in New Jersey (not New England), from Wikipedia.