FDA discusses the future of Avandia

Clearly this is an example of how science should not be done. We all know that even well-intentioned metaanalysis cannot substitute for true randomized controlled trials to make an informed decision. The unfortunate timing the NEJM in 2007 to publish the Avandia metaanalysis by Nissen & Wolski tainted the atmosphere. The RECORD trial was already ongoing under regulatory agreement which was made between Glaxo and the European Regulatory Agency; the study to see how rosiglitazone stacked in terms of CV outcomes vs. older drugs as part of Avandia’s approval. Unfortunately, that untimely publication of the Nissen metaanalysis made it impossible to conduct a controlled trial. Design issues aside, many subjects discontinued their participation RECORD due to the bad publicity. Consequently, it became politically impossible to conduct TIDE, a trial that could have answered whether CV outcomes differed between Actos and Avandia. People came to conclusions based on no data from controlled, randomized, prospective studies with CV events as their primary endpoint.  In the end, there were people both within the FDA and outside (such as the U.S. Senate Committee on Finance and Dr. Nissen) who politicized the issue and were determined to kill Avandia off. So, now we are faced with the spectacle of a faction within the FDA to have a more “civilized” re-look at the RECORD analysis to see whether the initial reaction was justified. In my opinion, the jury is still out. I haven’t seen any conclusive evidence that Avandia is harmful as far as long-term CV outcomes; it’s all speculation derived form short term studies not designed to assess CV outcome. Another issue often got lost in the noise, if people read the Nissen paper, the absolute risk of a CV event was minimal. Everyone got caught in the headlines: 43% increase in risk of myocardial infarction if one was placed on Avandia; however, it was missed that the absolute incidence was 0.6%, which means that 99.4% of people did not have an event. It might be too late in the day to try to resurrect Avandia, but I think having a somber unemotional look at RECORD analysis by independent parties is timely.

Unfortunately, well before the FDA meeting, both the warring parties within the FDA and outside individuals (such as Dr. Nissen in Forbes and other lay mass media) were already trying to derail even this attempt at adjudication of RECORD events by the Duke Research Institute investigators. I would like to see an independent look at the data available, and if the conclusion is that the original analysis shows no obvious harms of rosiglitazone vs. other medication used in the trial that needs to be publicized and given as much attention as all the previous attempts aimed to destroy the drug.

The impressive aspect to me is that the FDA has taken the opportunity to really look at the data very, very carefully. In my interpretation of what’s been going on so far, it’s more an issue of process as opposed to outcomes. What is the appropriate way to look at data? What are the appropriate ways of designing studies to answer the questions that we really need? And there has been a lot of controversy about the design of RECORD, but much of the discussion focused on the execution of RECORD. I think that actually becomes the more important aspect that's coming out of all of this. Can you actually have a regulatory agency, like the European Medicines Agency, design a trial to meet its needs and will that be acceptable to the scientists and to the FDA? That's a really critical issue. Having said that, what's very reassuring is that the FDA is going out of its way to really reinforce the integrity of the investigators and the analysis of the original RECORD study and to really validate the integrity of the re-evaluation as well. It's very true that there's nothing that's being changed. What it does is say: “This is the right process to evaluate the truthfulness of scientific studies.”

As for how the panel’s decision will affect the future of rosiglitazone, I have no idea how the committee is going to vote. I'm afraid that the horses have left the barn. I don't think there's going to be a substantial change in the business case for rosiglitazone. There’s been so much publicity and there’s such a negative impression that I don’t see a major future for rosiglitazone from a clinical standpoint, so I don't think that the vote is going to have a major change on how we end up treating people with diabetes. Again, I think the more important aspect of this particular meeting is that the agency is reinforcing the appropriate process by which data need to reviewed and how logical decisions as opposed to emotional decisions are made.