Steve Nissen has (once again) made waves with an analysis of cardiovascular risk. This time the subject is Avandia (rosiglitazone), a therapy for diabetes that's the oldest PPAR-gamma drug on the market. A meta-analysis of 42 reported clinical trials of the drug led to the conclusion that rosiglitazone is associated with a statistically significant risk of cardiac events.
The similarities to the Vioxx situation are what have made headlines (and what sent GlaxoSmithKline's (NYSE:GSK) stock down about 8% on the day the paper was released). But there are some important differences. Merck's ran into the Vioxx numbers in their own clinical data - the arguing has been whether they recognized the effects earlier (or should have), but it was a specific trial of theirs that led to the statistics that sank the drug. A meta-analysis is a much different beast, since you're trying to fit a large number of different trials, run in different ways for different reasons, into the same framework. Not everyone trusts them, even when the analysis is performed by someone as competent as Nissen, who does mention the limitations of the approach in the paper:
Our study has important limitations. We pooled the results of a group of trials that were not originally intended to explore cardiovascular outcomes. Most trials did not centrally adjudicate cardiovascular outcomes, and the definitions of myocardial infarction were not available. Many of these trials were small and short-term, resulting in few adverse cardiovascular events or deaths. Accordingly, the confidence intervals for the odds ratios for myocardial infarction and death from cardiovascular causes are wide, resulting in considerable uncertainty about the magnitude of the observed hazard. Furthermore, we did not have access to original source data for any of these trials. Thus, we based the analysis on available data from publicly disclosed summaries of events. The lack of availability of source data did not allow the use of more statistically powerful time-to-event analysis. A meta-analysis is always considered less convincing than a large prospective trial designed to assess the outcome of interest.
And that's what's happening here. A number of people at large diabetes treatment centers aren't ready to buy into a cardiovascular risk for Avandia yet, because they're wary of the statistics. There's a large cardiovascular outcome trial of the drug going on now, which won't wrap up until 2009, but several people seem to want to wait for that as a more definitive answer.
If Nissen's data hold up - and statistically, I'm definitely not up to the task of evaluating his approach - then we might be looking at a Vioxx-like risk level. Out of some 14,000 patients on the drug in the various studies, there were 86 heart attacks in the treatment groups, and 72 in the controls. That comes out to be statistically significant, but (as you can see) the problem is that Type II diabetics have a high background rate of CV problems. Looking at Nissen's Table IV, it also seems clear that most of the significance he's found comes from the pooling of the smaller studies. The larger trials are nowhere near as clear-cut, which makes you wonder if this effect is real or an artifact.
I'm certainly not prepared to say one way or another, and I just hope that the ongoing trial settles the question. It's certainly not unreasonable to imagine a PPAR gamma drug having this side effect, but if this were a strong mechanism-based phenomenon the numbers would surely be stronger. If a risk is confirmed, though, we'll then be faced with a risk-benefit question: Does the glycemic control that Avandia provides lead to enough good outcomes to offset any cardiovascular risk over a large population? If you think getting the current numbers is a tough job, wait until you try to work that one out.