In the trial by Hollander et al,1 the excess rates of cardiovascular (CV) events with rosiglitazone use (2.4% in the 2-mg/d group and 1.4% in the 4-mg/d group vs 0.9% in the placebo group) are of concern because of several reasons. First, these events occurred with rosiglitazone doses as low as 2 mg/d and after short-time exposure to the drug (within 24 weeks). Second, they were in agreement with the overall results of a meta-analysis2 of 42 trials showing that rosiglitazone use was associated with a significant risk of myocardial infarction (odds ratio, 1.43; 95% confidence interval, 1.03-1.98 [P = .03]). The meta-analysis2 included the trial by Hollander et al1 before its publication and peer review, when it was reported in a summary fashion on the Web site of the drug manufacturer. Although the interim data of the randomized RECORD (Rosiglitazone Evaluated for Cardiac Outcomes and Regulation of Glycemia in Diabetes) trial did not suggest a significant increase in the incidence of myocardial infarction or deaths from CV causes with rosiglitazone use, the frequency of congestive heart failure more than doubled in the rosiglitazone group compared with the control group (hazard ratio, 2.24; 95% confidence interval, 1.27-3.97
[P = .006]).3 Third, the authors' description that the excess rates of CV events are only “slight” and may be attributed to the high frequency of CV risk factors at baseline among patients randomized to rosiglitazone is by no means reassuring. Indeed, the impact of this “slight” increase in CV events can be substantial because millions of patients are using rosiglitazone worldwide. Moreover, the early occurrence of CV events argues against the unfavorable CV risk profile in the rosiglitazone groups at the study entry being a major contributing factor. Fourth, it is well known that rosiglitazone in maximum doses (8 mg/d) adversely affects plasma lipids by increasing levels of low-density lipoproteins and triglycerides by up to 23% and 15%, respectively.4 It would be interesting to know the changes in plasma lipids in the study by Hollander et al1 with low rosiglitazone doses.