Analyses of Cancer Data from Three Ezetimibe Trials

To the Editor:

The article by Peto et al. (Sept. 25 issue),1 which reports cancer incidence and mortality in three clinical trials of ezetimibe, raises disturbing scientific and ethical questions. Premature unblinding of ongoing trials is not a reliable approach to the evaluation of drug safety. In their discussion of one of the trials, the Improved Reduction of Outcomes: Vytorin Efficacy International Trial (IMPROVE-IT), the authors report that the mean exposure to ezetimibe was only 12 months, which is insufficient time for any hazard to emerge. Analysis of short-term trials dilutes any evidence of an excess risk. The most relevant statistical analysis is the upper 95% confidence interval for the observed relative risk. For cancer mortality in the two confirmatory trials, the Study of Heart and Renal Protection (SHARP) and IMPROVE-IT, the relative risk was 1.35 (approximate 95% confidence interval, 0.98 to 1.84), which was not reported by Peto et al. Thus, these two trials can only rule out a risk of death from cancer of 84% or more. The conclusion that there is no “credible evidence” for a cancer risk associated with ezetimibe is simply not supported by the data. The use of incomplete studies to rule out a safety hazard represents a dangerous precedent. If such a standard were widely applied, risky therapies would be routinely misclassified as safe, with potentially catastrophic public health consequences.

Steven E. Nissen, M.D.
Cleveland Clinic, Cleveland, OH 44195
nissens@ccf.org

Dr. Nissen reports receiving research support through the Cleveland Clinic Coordinating Center for Clinical Research from Pfizer, AstraZeneca, Novartis, Sankyo, Takeda, Sanofi-Aventis, and Eli Lilly and consulting fees from many pharmaceutical companies, with the stipulation that all fees will be donated directly to charity so that he receives neither income nor a tax deduction. No other potential conflict of interest relevant to this letter was reported.

Author/Editor Response

It is not in the interest of public health to label potentially useful drugs as unsafe if there is no credible evidence that they are — or, of course, to overlook reliable evidence of hazard if it does emerge. To help avoid both these serious errors, unexpected findings, such as those regarding the risk of cancer in the Simvastatin and Ezetimibe in Aortic Stenosis (SEAS) trial, that generate but do not prove the hypothesis of a new hazard should be tested independently in a separate, substantial data set.

For ezetimibe, the only large, randomized data set is that from interim results of the SHARP and IMPROVE-IT trials. It already involves four times as many cancers as SEAS did but does not suggest any increase in cancer incidence, either overall (313 in the ezetimibe group vs. 326 in the control group) or more than 3 years after randomization (20 in the ezetimibe group vs. 24 in the control group, all in the SHARP trial; with 19 vs. 17 after year 3 in the SEAS trial). In the SHARP and IMPROVE-IT trials, a nonsignificant difference in the number of cancers that had already caused deaths (97 in the ezetimibe group vs. 72 in the control group) was counterbalanced by a nonsignificant difference in the number of cancers (216 vs. 254) that had not yet caused death. (Fig. 4 of our article gives the confidence interval that Nissen mistakenly thought we had omitted.)

Provided an appropriate distinction is made between hypothesis-generating and hypothesis-testing findings (as in our article), the trial results provide no credible evidence of an adverse effect of ezetimibe. Continuation of the SHARP and IMPROVE-IT trials will provide further evidence about the effects of a statin plus ezetimibe (which reduces the level of low-density lipoprotein cholesterol more than monotherapy can), not only on safety but also on the major vascular outcomes this treatment may prevent.

Concerns about possible conflicts of interest were raised by the Congressional Committee on Oversight and Investigations; the response from the Clinical Trial Service Unit to the U.S. Congress is available on the unit's Web site.1

Rory Collins, M.B., B.S.
Richard Peto, F.R.S.
Oxford University, Oxford OX3 7LF, United Kingdom

DOWNLOAD PDF FULL

SHARECASH (U.S recommended)
ZIDDU (worldwide)

Mirror 1
Mirror 2