New analyses yield insufficient data on safety of rosiglitazone

American College of Cardiology 60th Annual Scientific Sessions

NEW ORLEANS — Current evidence remains insufficient to incriminate or exonerate rosiglitazone, on the basis of recent analyses, according to Sanjay Kaul, MD, and George A. Diamond, MD.

The safety of rosiglitazone was questioned in 2007 after a meta-analysis by Steven E. Nissen, MD, and Kathy Wolski, MPH, of the Cleveland Clinic, was published in The New England Journal of Medicine. The original meta-analysis, which included 42 trials of 27,847 patients, found a 43% increased risk for myocardial infarction and a 64% increased risk for cardiovascular death among rosiglitazone users. Nissen and Wolski recently updated their meta-analysis, which included 56 trials of more than 35,000 patients, and found that rosiglitazone was associated with a 28% greater risk for MI but was not associated with increased risk for CV death.

Because the updated analysis excluded 15 trials on MI and 29 trials on CV death in which no events were observed (so-called "zero-event" trials), Kaul and Diamond, both from Cedars-Sinai Medical Center, sought to determine whether these exclusions impacte the updated results on the CV safety of rosiglitazone (Avandia, GlaxoSmithKline). They compared the original results reported in the updated meta-analyses of the 56 trials with the ones that used different pooling methods and corrected models that accounted for "zero-event" trials.

According to the results, ORs ranged from 1.17 to 1.28 for MI and from 0.94 to 1.03 for CV death.

“Corrected models resulted in smaller ORs and narrower confidence intervals than did uncorrected models,” Kaul and Diamond wrote in their study abstract. “Although corrected risks remain elevated, none are statistically significant, except for the ‘treat as one trial’ method employed by the authors that ignores within-trial randomization and is prone to bias.”

Further, ORs were nonsignificant when Kaul and Diamond excluded results of the hypothesis-generating DREAM trial or the seven trials in which rosiglitazone is not indicated or is contraindicated.

“Given the fragility of the data … additional data will be required to adjudicate these inconclusive results. Such findings serve as insufficient bias for definitive regulatory decisions and clinical recommendations,” Kaul and Diamond said. - by Katie Kalvaitis

Disclosures: Drs. Diamond and Kaul report no relevant financial disclosures. Dr. Soni is an employee of Amarin.

For more information:

Kaul S. 1079-302. Presented at: ACC 60th Annual Scientific Sessions; April 2-5, 2011; New Orleans.

This is a very interesting analysis that raises questions about the initial analysis that was presented and published. I haven’t looked at all of the data, but there is some validity in what the researchers are saying, in that all the studies are not consistent with conclusions of previous analyses.

- Paresh Soni, MD, PhD
Internist,
Clinical Research, Amarin

Follow CardiologyToday.com on Twitter.