FDA's Endocrinologic and Metabolic Drugs and Drug Safety and Risk Management Advisory Committees are calling for strengthened warnings for GlaxoSmithKline's type 2 diabetes drug Avandia due to cardiovascular safety concerns, but not its withdrawal from the market.

The joint committee convened July 30 to discuss the potential cardiovascular risk associated with Avandia (rosiglitazone), and voted 20-3 that use of the thiazolidinedione increases cardiac ischemic risk.

"The committee felt strongly enough that it was reflected in ... near unanimous concerns that Avandia increased cardiovascular risk, with some qualifications," acting committee chair Clifford Rosen, St. Joseph Hospital, said at a press briefing following the meeting.

The committee concluded there is an increased risk of myocardial infarctions based on a meta-analysis of rosiglitazone clinical data conducted by the sponsor, which was subsequently reanalyzed by FDA (¹ Also see "FDA Is Waiting On Internal Analysis Of Avandia Cardiovascular Safety Signal" - Pink Sheet, 28 May, 2007., p. 3).

Three meta-analyses, including one co-authored by Cleveland Clinic's Steven Nissen, found a 30 to 40 percent increased risk of myocardial ischemic events for rosiglitazone groups in Avandia clinical trials versus comparator groups. FDA's analysis found a statistically significant increase in the overall estimate of risk (odds ratio: 1.4), and that all rosiglitazone-containing regimens trended toward increased risk.

The committee observed that the safety signal was more pronounced for placebo-controlled trials than trials with active comparators.

"These trials address different questions. The placebo-controlled trial answers the question 'does rosiglitazone increase the risk of cardiovascular events?' And I think the answer is clear - there is an excess risk that is statistically significant. The active-control trial answers a different question, it asks 'how does rosiglitazone compare to others?' And in my view, I think it is totally inappropriate to pool those data," temporary non-voting member Curt Furberg, Wake Forest University, said.

"When active-control trials in the Avandia meta-analysis were separately looked at, they showed no significant risk," Office of Drug Evaluation II Director Robert Meyer noted.

Although the committee agreed that the results of the meta-analyses support the overall conclusion that Avandia increases the risk for cardiovascular events, which is of particular concern given that the majority of type 2 diabetics die from coronary heart disease, the panel expressed reservations about relying too heavily upon the meta-analyses because of concerns about the underlying clinical data.

In addition to the hypothesis-generating nature of the studies, the committee noted that many of the trials comprising the meta-analyses were short-term in duration, had few events and were not designed to evaluate cardiovascular events.

"We may be looking at an example where the quality of the analyses far exceeds the quality of the data," temporary member Nancy Geller, National Heart, Lung and Blood Institute said.

Other Data Do Not Clarify Cardiac Signal

Data from other large, randomized clinical trials and observational studies fail to definitively refute or confirm the safety signal seen in the meta-analyses.

Although Avandia frequently showed a trend toward increased myocardial infarction versus comparators in large, randomized clinical trials, rosiglitazone did not have a statistically distinct effect on cardiovascular outcomes relative to its comparators in the ADOPT, DREAM and interim RECORD results.

Furthermore, ongoing randomized clinical trials - including GSK's cardiovascular outcomes study RECORD and BARI 2D - are not expected to definitively establish Avandia's level of cardiovascular risk (² Also see "Avandia, Actos Add Black Box; RECORD Leaves Heart Attack Risk Unresolved" - Pink Sheet, 11 June, 2007., p. 3).

"I'm extremely disappointed, and I think this is really important that we're not going to get any really greater insights on this question from the randomized, placebo-control trials. I think that point has been fairly well established that we're not going to get the outcomes we expect, based on the data sets that have been coming from several of the other trials," Rosen said.

The committee was also presented data from three GSK observational studies, which FDA determined do not refute the signal seen in the meta-analysis.

Committee member Sean Hennessy, University of Pennsylvania, asked whether Office of Surveillance and Epidemiology Director for Science and Medicine David Graham thought there was sufficient information to make a decision on whether Avandia should continue to be marketed "or whether it would be more prudent to wait for some of these other data to come in and have the chance to analyze them in a more rigorous way."

Graham did conclude that Avandia should be withdrawn based on the current evidence, especially as Actos is available. He noted that there is no point in delaying a decision in order to wait for clinical data from ongoing trials that will not be powered or designed to answer questions about Avandia's cardiovascular safety profile. "At some point you're going to have to make a decision based on data that is not perfect," he said.

Comparative Data Do Not Support Withdrawal

Faced with a recommendation by the OSE staff to advise FDA to withdraw the drug, the committee focused on a subset of the meta-analysis data in its decision that the risk-benefit profile of Avandia supports its continued marketing.

"Patients with type 2 diabetes needing additional treatment to control their blood sugars don't have the choice between a drug and placebos. They and their doctors must choose between one drug and another," Meyer noted. The committee agreed that data from active-control trials was of greater clinical significance than that from placebo-controlled trials. Given that a significant safety signal did not emerge from the trials with active comparators, the committee voted 22-1 against withdrawal.

Temporary member David Schade, University of New Mexico, stressed the importance of having a TZD on the market. "If we remove rosiglitazone for what I consider a borderline data indication, then in one or two years we find out pioglitazone causes bladder cancer or something else, and that we then have to take it off with no choice, we're going to look back and say 'gee, why did we do this?' I am very concerned that to perform a draconian type of operation on this medication would probably not be advised in the long run," he said.

The landslide vote by the committee may help break an impasse within CDER, where the OND and OSE are engaged in a "fundamental disagreement" over whether Avandia should remain on the market.

Meyer characterized the committee discussion as providing "a fairly unanimous recommendation" during a press conference following the meeting. "It wasn't that there was a fracture along clinical lines. It was a fairly strong and consistent discussion. ... We heard fairly clear recommendations that some additional labeling was needed, and I often heard the word warnings. I did not uniformly hear the words 'boxed warning.'"

In an interview with "The Pink Sheet" Nissen suggests that labeling should include a "black box" warning about the overall cardiac ischemic risk.

In addition to a labeling change, FDA will also likely seek additional information. "We're going to need to ... decide what pieces are outstanding," Throckmorton said. "The committee made some suggestions about some additional work with regard to rosiglitazone. We need to talk in terms of exactly how much additional work will be needed before we can make that decision."

Rosen stressed that "what I think we really need to see is more data that gives us a clear cut answer - whether that's done by the sponsor or the NIH or by other organizations."

GSK's risk management proposal includes elements aimed at better characterizing Avandia's cardiovascular risk (see sidebar, p. 5).

Committee Identifies Spectrum Of Risk

Based on the available data, the committee identified several subpopulations that could be at an increased risk of a cardiac event.

"There are reasons not to use this drug in certain type 2 diabetic patients ... people who have congestive heart failure or are prone to it, those who have significant CV disease, particularly if they've been hospitalized previously for myocardial infarction. And I'd be concerned about ... the groups that are constantly using nitrates, where we found a significant interaction, and long-term users of insulin, who probably have more advanced disease," Rosen said.

"There is a fairly large spectrum of risk between someone who's pre-diabetic or very early diabetic and someone who's had the disease for 20 years," temporary member Peter Savage, National Heart, Lung and Blood Institute, said.

"It could be that the use of these drugs to either prevent the development of diabetes or slow the progression of diabetes and the loss of beta cell responsiveness could more than offset a very slight increase in absolute risk in CV disease at that end of the spectrum. At the other end of the spectrum where you're adding rosiglitazone to insulin therapy, the results could be completely different. It could be an unjustifiable increased risk."

- Brian Marson (b.marson@elsevier.com) and Carlene Olsen (c.olsen@elsevier.com)

Avandia Should Stay On Market With Warnings, Committee Says

More from Archive

More from Pink Sheet