Background: Rosiglitazone (Avandia) is one of two thiazolidinedione agents available for managing glucose levels in patients with type 2 diabetes (troglitazone [Rezulin] was recalled in 2000 because of hepatotoxicity). It is well known that cardiovascular morbidity and mortality are serious issues in patients with diabetes; however, earlier studies may not have had the power to determine whether rosiglitazone affects cardiovascular outcomes in these patients. Nissen and Wolski compared the cardiovascular effects of rosiglitazone with those of other antidiabetic agents.

The Study: The meta-analysis included 42 studies that lasted at least 24 weeks and that compared rosiglitazone with placebo or another antidiabetic agent. Five of these studies were available to the U.S. Food and Drug Administration (FDA) advisory group that recommended approval of rosiglitazone in 1999. Thirty-five studies were identified from the manufacturer's clinical trial registry; 26 of these studies are unpublished. Two studies (Diabetes REduction Assessment with ramipril and rosiglitazone Medication [DREAM] trial and A Diabetes Outcome Prevention Trial [ADOPT]) were recently published. If individual patient data were unavailable, data summaries were reviewed for occurrences of myocardial infarction and death from cardiovascular disease, and odds ratios were calculated for these outcomes.

Results: The meta-analysis included 15,560 patients using rosiglitazone and 12,283 patients using another antidiabetic agent or placebo. The mean age of patients in both groups was 57 years, and the mean baseline A1C level was 8.2 percent. In the rosiglitazone group, there were 86 myocardial infarctions and 39 deaths from cardiovascular causes, compared with 72 myocardial infarctions and 22 deaths in the control groups. The summary odds ratio for myocardial infarction in the rosiglitazone group was 1.43 (P = .03) compared with the control groups. The odds ratio for death from cardiovascular causes in the rosiglitazone group was 1.64 (P = .06) compared with the control groups, but this was not statistically significant. When possible, odds ratios were calculated comparing rosiglitazone with metformin (Glucophage), sulfonylureas, and insulin. In these subgroup comparisons, the odds ratios for myocardial infarction and all-cause death were greater with rosiglitazone, although they did not reach statistical significance.

Conclusion: The authors conclude that rosiglitazone is associated with a significant increase in the risk of myocardial infarction compared with other glucose-lowering medications or placebo in patients with type 2 diabetes. The authors have called on the manufacturer of the drug and on the FDA to make source data available for independent review.

editor's note: This report prompted widespread concerns about the safety of rosiglitazone last summer; however, it has significant limitations. The small number of myocardial infarctions (158 among 27,843 patients) makes it difficult to detect a true difference in outcomes. Also, because it is a meta-analysis, its findings may be less reliable than those from a well-designed, prospective clinical study.

In contrast, the ongoing industry-supported RECORD (Rosiglitazone Evaluated for Cardiac Outcomes and Regulation of Glycaemia in Diabetes) trial was specifically designed to measure cardiovascular outcomes in patients who use rosiglitazone.¹ Because of recent concerns, the RECORD trial included an unplanned interim analysis of the 3.75 years of data that had been collected. This analysis showed that rosiglitazone users had increased rates of heart failure, but not myocardial infarction or mortality. However, these findings should be viewed cautiously because the study was underpowered (higher attrition rates and much lower adverse event rates than projected when determining the trial's sample size).²

A more recent meta-analysis³ and a retrospective cohort study⁴ showed higher relative risks of myocardial infarction and heart failure in patients receiving rosiglitazone (see accompanying table^1,3–5 ). The retrospective study also reported a greater risk of mortality (relative risk = 1.29) and is of particular interest because it only evaluated patients with diabetes who were older than 65 years, in contrast to most other current studies examining rosiglitazone.⁴

note:All values were published as relative risks (the probability of an event occurring in one group compared with the control group), except for the RECORD trial, which used hazard ratios (similar to relative risk, but takes into account differences in duration of follow-up for individual patients).

RECORD = Rosiglitazone Evaluated for Cardiac Outcomes and Regulation of Glycaemia in Diabetes.

The FDA has also issued new warnings for rosiglitazone use based on its own analyses. The FDA recommends that rosiglitazone not be used in conjunction with nitrates (if used for heart disease) or insulin because of a greater risk of myocardial ischemia. This is in addition to previous recommendations that thiazolidinediones should not be used in patients with New York Heart Association class III or IV heart failure and should be used cautiously in patients with less severe cases of heart failure.⁶

So where does this leave us? There is substantial circumstantial evidence that rosiglitazone is associated with higher risks of heart failure and myocardial infarction. Despite the awkwardness of persuading a patient to use a drug that may provide similar benefits but pose greater risks than other proven agents, the legitimate concerns raised by these studies make this a serious issue. Until there is conclusive evidence about the safety of rosiglitazone, many physicians and their patients may be more comfortable using alternative treatments.—k.t.m.