A rash of headlines appeared this weekend as a Senate report on Avandia was leaked to the New York Times. It contained the usual accusations and concerns of certain FDA staffers such as Dr Graham who spoke 1-2 years ago at prior Avandia Advisory Boards and whose comments were largely based upon the meta-analysis of Nissen and Wolski using post hoc data derived from short term studies with Avandia. Those studies were largely conducted to obtain glycemic control data. Since that time considerable republication of similar meta-analyses, using the same or similar data have appeared. These have spread the concerns regarding Avandia further. Of course since that original meta-analysis was published in 2007, a number of subsequent studies have appeared to alleviate or to directly address the concerns raised by Drs Nissen and Wolski. In 2007, two prior large scale randomized prospective clinical trials had also contained data regarding CHD events with Avandia, namely DREAM and ADOPT. Both studies failed to show increased risk with Avandia, contrary to the conclusions attributed to them in the current New York Times article by Gardiner Harris. But these were post hoc analyses and therefore not definitive. Of course the meta-analysis was as well. Subsequent studies were also analyzed for potential adverse consequences of Avandia, including ACCORD and VADT. Neither trial showed adverse effects and VADT may have suggested benefit instead. A direct test of cardiovascular safety – the RECORD was completed and published in Lancet in June 2009 (373:2125-35) by Philip Home and colleagues. This five year study of nearly 4500 patients with diabetes randomized to treatments containing or not containing Avandia showed no increased risk of death or hospitalization for cardiovascular disease in patients taking Avandia as compared to those not taking Avandia. Since this was the prespecified primary endpoint, it seems proven that the suggestions provided by the meta-analysis have no validity. This conclusion is support by the recently published Perspective by Charles Hennekens and David DeMets in JAMA (302:2361-2, 2009). Here the authors decry overreliance upon meta-analyses of small scale short term studies in favor of data provided by large scale randomized long term trials. Dr Hennekens, a former Professor of Preventive Medicine at Harvard, and a colleague on the NCEP, concludes that the findings of RECORD lay to rest the concerns raised by the meta-analysis.
None of these newer studies are addressed or even considered in the article now appearing in the New York Times and the concerns it raises appear to have been adequately addressed and laid to rest by these newer studies. In short, this is old news, quite old indeed.