FDA Advisory Panels Acknowledge Signal of Risk With Rosiglitazone, but Stop Short of Recommending Its Withdrawal

July 31, 2007 (Gaithersburg, MD) - Rosiglitazone (Avandia, GlaxoSmithKline [GSK]) is associated with a clear signal of cardiac ischemic risk in type 2 diabetics, the available data suggest, but this signal is not enough to justify yanking the drug from the market. This was the near-unanimous conclusion of the FDA's joint Endocrinologic and Metabolic Drugs/Drug Safety and Risk Management advisory committees at Monday's hearing.
A long day of confusing, often conflicting data was capped by the seemingly inconsistent conclusions of the committee members who agreed 20:3 with the statement that available studies supported a signal of harm, but voted 22:1 to keep rosiglitazone on the market. Dr Thomas Pickering (Columbia University, New York, NY), who was one of the three members who did not agree that there was a clear increased risk of ischemic events, pointed to the apparent contradiction: "I'm puzzled as to how people can vote yes for both questions," he mused.
No choice but to have choices
But the vast majority of voting members on the panel seemed to agree with the sentiment raised repeatedly by presenters, panelists, and open public-hearing speakers, that it was important for physicians to have rosiglitazone in their arsenal of treatments for type 2 diabetes. Indeed, some of the day's discussion revolved around an emerging hypothesis that the other thiazolidinedione (TZD) on the market, pioglitazone, might not carry the same safety concerns as rosiglitazone. Those data, however, comes predominantly from an as-yet unpublished analysis conducted by pioglitazone manufacturer Takeda, and has not yet been reviewed by the FDA, nor was it provided in full to panel members.
A review of that data, according to FDA statistician Dr Joy Mele, was on track to be completed in time for the panel's review of the cardiovascular ischemic/thrombotic risks of TZDs, which had originally been scheduled for later in the year. The hearing, however, was bumped up after the publication of a controversial meta-analysis in the New England Journal of Medicine (NEJM) [1]--pointing to a significant 43% increase in myocardial infarction with rosiglitazone. This paper was followed swiftly by a congressional hearing scrutinizing the FDA's role in evaluating rosiglitazone's safety, as reported by heartwire. During today's session, panel members heard from the sponsor and the FDA, both of whom had conducted their own meta-analyses of the randomized controlled trial data and turned up findings that were surprisingly consistent with the NEJM analysis--a 40% increased risk of serious ischemic events by the FDA's reckoning and a 31% increase in myocardial ischemic events in GSK's meta-analysis.
All of the yes-votes reiterated the same concerns: that the evidence linking rosiglitazone with increased risk of cardiovascular death or MI was weak or inconsistent, particularly in trials that had active control arms rather than placebo comparators; that ischemic risk appeared higher in older patients, patients with heart failure, patients with preexisting coronary disease, and patients taking insulin--and that this should be reflected in the labeling. In fact, several panelists pointed out that current labeling lists rosiglitazone as being indicated for diabetics taking insulin; they felt this should be removed and a black box should be added warning against its use in this group. Others emphasized that the inconclusiveness of the existing studies and the fact that trials are still ongoing should also be mentioned in the packaging.
But time and again, the experts on the panel bemoaned the fact that, not for the first time, the FDA had not had the foresight to mandate appropriate trials, leaving the committees to try to draw conclusions from meta-analyses and observational studies. Even the ongoing RECORD, ACCORD, and BARI-2D trials, by the FDA's own review, are underpowered or not designed to answer key questions about whether ischemic events will be higher than with other diabetes drugs, and if they are, which patient subsets will be affected.
The mere fact that most of the studies included in the meta-analyses were only six months in duration underscores the paucity of solid information, Dr Nancy Geller (NHLBI, Bethesda, MD) pointed out. "The quality of the meta-analyses far exceeds the quality of the data that went into the meta-analyses," she quipped.
A lone no-vote
The sole no-vote on the key question of whether rosiglitazone should remain on the market came from Dr Arthur Levin (Center for Medical Consumers, New York, NY).
"It seems to me that given the evidence of a strong safety signal, given the fact that around this table and at the FDA there are doubts about the ability of ongoing clinical trials to definitively answer the question about the CV safety of the drug, and given the enormity of the potential public health risk of allowing this drug to continue to be marketed and used by millions of people for the rest of their lives, I logically can't find any way to justify leaving this drug on the market."
Levin's opinions have the support of at least two FDA insiders, Dr Gerald Dal Pan and Dr David Graham, both in the FDA's Office of Surveillance and Epidemiology, Center for Drug Evaluation and Research. Graham presented a risk/benefit assessment of rosiglitazone, pointing out that although his views were his own, his findings had been reviewed and were supported by others in his department, such that he was not just speaking as "David Graham the FDA Whistle-Blower."
Graham showed projections to back up his claim that ongoing studies "will not change our state of knowledge. The poor design and low power of RECORD, and the low power of BARI 2D mean a very high risk of a false-negative conclusion exists. Waiting for these studies will insure an additional 1600 to 2500 adverse cardiac events per month," he predicted. "Waiting for these studies does not make sense."
As such, he concluded, "rosiglitazone should be withdrawn from the market."
That recommendation fell largely on deaf ears today, but Graham's concerns about the quality of the existing rosiglitazone data, and the flawed studies in progress, struck a chord with panel members, who called for stricter standards for pre-approval and post-marketing studies.
"I would have to say, the FDA has to take some responsibility for the dilemma in which we find ourselves, for approving less than optimally designed trials in the past," Dr Arthur Moss (University of Rochester, NY) observed. "I do think there is a problem that needs to be rectified in the future."
Vindication for Nissen
Commenting on the day's deliberations to heartwire, Dr Steven Nissen (Cleveland Clinic, OH), one of the authors of the NEJM analysis that first sparked the current controversy, seemed satisfied that the FDA and sponsors' meta-analyses had confirmed his own findings.
"The committee was convinced enough by this that they voted overwhelmingly that there is an increased risk and I think that was the right thing to do. Clearly they also overwhelmingly voted that there should be a strong warning placed on the drug, which will have the impact of identifying for physicians and patients the potential of this drug to cause ischemic heart disease," he said.
Nissen predicted that the panel's conclusions could have "a major effect on utilization of the drug." But, sounding somewhat deflated after a long day spent on the sidelines of the discussion as a non-voting consultant--answering the occasional question from the committee members--Nissen also anticipates "continued discussion and debate."
"One concern I have is that black-box warnings do not always result in huge changes in prescribing practices, so time will tell," he told heartwire. "History has shown there is not always major respect for those warnings . . . and we'll have to see what happens with this drug."
Nissen SE and Wolski K. Effect of rosiglitazone on the risk of myocardial infarction and death from cardiovascular causes. N Engl J Med 2007; 356:2457-2471.