I suppose with the flurry of press on the ACCORD and ADVANCE trials presented at the ADA, the third similar study, the VADT, did not garner that much excitement in the media, with barely a mention. But the one headline from the good folks at MedPage today clearly caught my attention: VA Diabetes Trial Appears to Vindicate Rosiglitazone (Avandia) Safety
This 3rd trial was consistent with ACCORD and ADVANCE (which I blogged recently) in that it was unable to prove that aggressive diabetes management improved rate of heart attack and stroke. However, like the two prior studies, the expected event rate was lower than anticipated (likely due to better control of cholesterol and blood pressure) indicating in my opinion good news for the overall management of diabetes. However, unlike the Triple Crown, this 3rd leg did not disappoint with news about Avanida. However, before getting to this, let's review.
In May of 2007, Dr. Steve Nissen published a meta-analysis of diabetes studies containing Avandia. This now famous (or infamous) study started the entire Avandia safety controversy. Dr. Nissen analyzed 42 randomized trials lasting over 24 weeks involving Avandia, and found a 43% increase in risk of for myocardial infarction (for you stat folks, odds ration for MI was 1.43, 95% confidence interval [CI], 1.03 to 1.98; P=0.03).
1. Discrediting Nissen's study. In addition to concerns about how quickly the study was published in the NEJM and the political nature of the study's use as a lightning rod to shape public policy, scientific holes in the analysis were also raised. Diamond, et. al did a re-analysis of the same data published in the Annals of Internal Medicine in October 2007, and found no difference in heart attacks. As a side note, no scientist or statistician has ever been able to explain to me how if the rates of myocardial infarction were 0.58% in the Avandia group and 0.62% in the non-Avandia group, how it could be possible to have a 43% increase in myocardial infarctions with Avandia.
2. RECORD. The FDA had known about the concerns of myocardial infarctions, and was likely the primary source of Nissen's investigation. However, they were apparently holding off on any decisions, because the RECORD study, a large, randomize prospective trial to answer the very question of cardiac effects of Avandia was ongoing, and interim monitoring of the study seemed to indicate no cause for concern. However, when Nissen went public and started all the controversy, the RECORD investigators felt obligated to respond, and on July 5th, 2007, preliminary data from the RECORD study was published in hopes of answering any Avandia concerns. This study looked at over 4000 patients with uncontrolled diabetes not previously on Avandia. Half received Avandia and the other half did not, with the primary outcome of cardiac events. After 3.75 years there was no difference in heart attacks in the patients taking Avandia the on those not taking the medication. Not one, but three editorials accompanying the article suggested that these findings were not reassuring because the study was not statistically powered to show a difference.
3. The FDA investigates. In response to the controversy, and pressured by Congress, the FDA extensively reviewed Avandia safety data. In addition to a meta-analysis similar to Nissen's, the FDA looked at RECORD and two other large, randomized prospective trials; ADOPT (which showed Avandia to be superior to metformin and glyburide in keeping diabetes under control) and DREAM (which showed Avandia to prevent diabetes in patients who were at risk). These studies combined or individually, along with retrospective analysis from large data sources did not show any risk for Avandia related to heart attacks. In November 2007, the FDA added a boxed warning to Avandia which mentioned data discussed in Dr. Nissen's meta-analysis as well as data from ADOPT, DREAM and RECORD, stating, "In their entirety, the available data on the risk of myocardial ischemia are inconclusive."
4. ACCORD is stopped. In February of 2008, the intensive arm of the ACCORD study was stopped because of increased deaths. Since the results were just presented, we know that in the over 10,000 diabetics studied for 3.5 years, there was no difference in intense vs. usual care for prevention of heart attacks, and in fact there were more deaths in the intense group. According to the study, more than 90% of the patients in the intense group were taking a TZD (almost all Avandia) and more than half the patients in the standard care groups were taking a TZD. Three accompanying editorials to the recently published results imply that since more patients were on Avandia in the intense group, this might account for the increased risk of death in the intense group. However, we know this not to be the case because when the ACCORD was stopped, the NIH specifically looked at Avandia and found this not to be associated with increase risk. “Because of the recent concerns with rosiglitazone, our extensive analysis included a specific review to determine whether there was any link between this particular medication and the increased deaths. We found no link.”
5. VADT. June 8th, 2008, the results from the Veterans Affairs Diabetes Trial (VADT) were released. This study also looked of almost 2000 veterans for almost 8 years similarly found no improvement in heart attack and stroke in the intensively treated group as compared to the standard group. However, like the previous studies, the rate of cardiovascular events was substantially lower than expected, and about 80% of the patients in both groups were taking Avandia. When the investigators presented their results of their analysis at the ADA meeting, these results showed that, if anything, Avandia may have actually prevented heart attacks!
What does this all mean?
The gold standard for research is a large, prospective, randomized clinical trial (RCT). This is considered one of the best ways to answer a question about whether a drug causes a side effect or an improvement. A meta-analysis is done when large RCT's are not available. Meta-analysis uses statistical techniques to combine data from smaller trials in an attempt to statistically replicate a large, prospective RCT. Since Dr. Nissen's meta-analysis showing risk with Avandia that is fraught with controversy, we have had multiple large, prospective, randomized clinical trials that show the opposite result. Between RECORD, ADOPT, DREAM, ACCORD and VADT; there have been over 26,000 patients studied for over 3.5 years, of which more than 15,000 patients took Avandia and showed absolutely no difference in heart attacks or myocardial ischemia. I think it is clear based on the inherent limitations of Nissen's findings and the substantial evidence from the aforementioned studies that Avandia is not, and has never been associated with heart attacks or myocardial ischemia.