Are Cardiac Risks of Rosiglitazone Legitimate?

Article excerpt

The controversy surrounding rosiglitazone (Avandia, GlaxoSmithKline) and cardiac risk prompted the addition of a special symposium devoted to this topic at the 67th Annual Scientific Sessions of the American Diabetes Association (ADA) held recently in Chicago.

Standing behind a meta-analysis showing excess cardiac risk associated with rosiglitazone was Steven E. Nissen, M.D., lead investigator of the meta-analysis (New England Journal of Mediane 2007;356:2457-2471). The 43% relative excess hazard of myocardial infarction (MI) found with rosiglitazone compared with controls in his analysis of 42 randomized trials, most of them unpublished, is in line with Glaxo's own meta-analysis provided to the Food & Drug Administration in 2006, and with an independent analysis conducted by the FDA, he said.

Subsequently, an observational study performed by Glaxo found no such excess cardiac risk with rosiglitazone, but Nissen pointed out several flaws in this study. It used an insurance claims database with an incomplete assessment of covariates, it followed patients for only one year, and it included no comparison with the other marketed thiazolidinedione, pioglitazone (Actos, Takeda). "In my view, this study was very weak," said Nissen, chairman of cardiovascular medicine at the Cleveland Clinic.

In assessing Nissen's meta-analysis, Philip Home, DM, accused him of "data snooping on quite a big scale." In reaction to the meta-analysis, Home released interim data from a prospective trial designed to address cardiovascular safety with rosiglitazone. The study, known as RECORD (Rosiglitazone Evaluated for Cardiac Outcomes & Regulation of Glycemia in Diabetes), found no increased risk of rosiglitazone on a composite endpoint of hospitalizations and death from cardiovascular causes, compared with metformin and a sulfonylurea, at an unplanned analysis at 3.75 years (N Engl J Med 2007;June 5).

Based on these interim data, rosiglitazone should maintain a role in the glucose-lowering armamentarium, and "RECORD should continue to its planned end," which is a total of five years, said Home, professor of diabetes medicine, University of Newcastle upon Tyne, U.K.

"Major errors in design have rendered this study futile," said Nissen in enumerating the weaknesses of RECORD. The openlabel design and the chosen endpoint, which did not include MI, are two such weaknesses. Another is that the annual event rate in RECORD was only 3. …