Avandia meeting concludes with 'mixed-bag'; restrictions suggested
The meetings July 14-15 with the FDA’s Encocrinologic and Metabolic Drugs Advisory Committee and the Drug Safety and Risk Management Advisory Committee are the second of their kind dealing with rosiglitazone after a meta-analysis featured in the June 2007 edition of the New England Journal of Medicine questioned the safety of the drug and whether it was a risk for cardiovascular (CV) events, including death, in patients who were administered it.
RECORD trial creates concern
To bolster its case, representatives of GSK presented results of the RECORD trial, which actively compared rosiglitazone to other drugs used to treat diabetes including metformin and sulfonylurea.
Results of the open-label trial found that patients administered rosiglitazone saw no increase in CV or all-cause mortality or stroke, reported Philip D. Home, DM, DPhil, a professor of diabetes medicine at the Newcastle University in the U.K., at the meeting. Home was part of the trial.
The “RECORD [trial] did not show an increased risk in major adverse cardiac events (MACE) or CV death or hospitalization versus standards of care,” said Murray Stewart, GSK's vice president of clinical development of biopharm research and development. "When used appropriately," said Stewart, rosiglitazone “has a positive effect … and should remain a treatment option for patients with type 2 diabetes.”
However, while Home concluded that the study did not increase the risk of recurrent events for those who were having MI and that rosiglitazone better managed glucose control for diabetics, he said that the drug does increase the risk of heart failure and distal fractures.
During Tuesday and Wednesday’s meetings, the RECORD trial results were a hotly contested item as FDA advisory members questioned whether or not its evidence was fruitful and unbiased.
Steve Nissen, MD, of the Cleveland Clinic, questioned the validity of the RECORD trial during his presentation and charged that GSK may have had access to treatment codes and knowledge of study data from the start. Additionally, Nissen charged that GSK removed and excluded silent MIs from the endpoint of the study, which could have changed the interpretation of the study and created bias. Nissen was not a member of the FDA advisers.
However, in a statement from GSK, Stewart called this a “misconception” and said that GSK never removed silent MIs from data as they were not an endpoint to begin with.
While Karen M. Mahoney, MD, of the FDA, said that the RECORD trial is the only new randomized controlled CV data since the 2007 committee meeting, “I think most scientists involved in the review of RECORD would agree that it cannot answer every single question one might have about the CV safety of rosiglitazone.”
Adding to the critiques of the RECORD trial, Thomas A. Marciniak, MD, medical team leader division of cardiovascular and renal products at the FDA, called RECORD “a very complex and confusing study,” and said that the trial cannot be depended upon for safety.
Marciniak reviewed 549 case report forms at 56 sites participating in the RECORD trial and found that one in four had a major problem in reporting (13 percent), and by a ratio of 4:1, these “problems” favored rosiglitazone, which included missing endpoints, missing information and adjudication disagreement.
Marciniak blamed the potential biases on the open label design of the RECORD trial and said, “If consulted in advance, I would have rejected this study design as completely inappropriate and biased.”
Ellis F. Unger, MD, deputy director of the Center for Drug Evaluation and Research (CDER) at the FDA, in response to Marciniak’s findings said, “We take these concerns very seriously. Whether or not particular adverse events were forwarded for adjudication, whether there was bias, uncertainty regarding dates, these are all important issues and the issue here is really truth.” However, Unger called Marciniak’s evidence regarding the adjudication bias “exploratory” at best.
According to GSK, 1,600 events during the RECORD trial were sent to adjudication and were peer reviewed.
In discussion, Nancy L. Geller, PhD, of the National Heart, Lung and Blood Institute, said that a total readjudication of all of the forms and events would be one of the only ways to “get anything out of RECORD.”
She offered that “evidence overall is missing,” specifically in regard to mortality. Because, most of the studies in the meta-analysis were short-term trials, there is no room to study death rates, she said. The majority of members agreed with Geller, judging that rates of mortality were unclear and data insufficient.
“We may not see an impact on mortality but we may see a potential impact on other things like MIs, which are detrimental,” said Abraham Thomas, MD, of the Henry Ford Hospital in Detroit, and a member of the FDA advisers.
However, many committee members agreed that mortality rates provide a clear-cut answer to many questions and deplete biases.
Nissen said, “We cannot, and we should not let data of this quality influence a decision that affects the lives of so many hundreds of thousands of people. This trial was not properly handled and cannot be used for regulatory purposes.”
Is rosiglitazone the black sheep?
During the meetings, the FDA had 10 additional trials to pull from for its meta-analysis, but still, data were marginal at best, the majority concluded.
“Due to the way that medicine is practiced today, patients rarely stay on TZDs for an extended period of time,” said David J. Graham, MD, of the office of surveillance and epidemiology at the FDA. He said that diabetes patients on average are on the class of drugs for four months before they move onto other therapies.
"That’s important to keep in mind when you are looking at these risks and thinking of what the long-term benefits might be of a drug that you stay on for only a short period of time,” he said. “If you don’t stay on the drug long enough to get a benefit, then all you are buying is the risk.”
The FDA advisors also voted on whether or not there are sufficient data to be able to raise concerns regarding ischemic CV events in type 2 diabetes patients administered rosiglitazone relative to those on pioglitazone. During the vote, 21 voted those data exist and that should warrant concern, three voted there were not enough data and nine voted they could not make a conclusive decision.
During Nissen’s presentation, he compared rosiglitazone to pioglitazone (Actos, Takeda). He found that forest plots showed rosiglitazone had significantly higher hazard rates for MI: 1.80 versus 0.91, respectively.
“There are no known benefits for rosglitazone that are not produced by the companion drug, pioglitazone,” said Nissen. “We have 13 class drugs that reduce blood sugar; we have a safe alternative in the same class.”
Lamont G. Weide MD, PhD, from the University of Missouri, Kansas City, and an FDA committee member, said, “Perhaps too much weight is placed on [the] RECORD [trial]. It seems to me that it ends up being that rosiglitazone is about as good as the non-thiazolidinediones.”
However, other members saw concerns with comparing rosiglitazone to other drugs in its class as it is not necessarily proven whether or not all TZDs have the same risk-benefit ratio. “There are so many classes,” said Thomas. “Some of these [drugs] are being tested, some of these are under testing and some of these may never be tested. We don’t really know what the risk profiles are. Lumping TZDs together is probably not a safe assumption."
“Pioglitazone is a perfectly acceptable alternative,” said Morris Schambelan, MD, at the San Francisco General Hospital and University of California San Francisco, who voted that the drug be pulled from the market.
“I think the quality of pharmaceutical care that is delivered to patients is not as it could be and I think we have a great opportunity here to protect patients from additional harm by not withdrawing any treatment from them because there are alternatives available,” said Almut G. Winterstein, PhD, an associate professor at the College of Pharmacy Epidemiology & Biostatistics College of Public Health and director of the FDA/CDER graduate training program in POP Research in Gainesville, Fla., who also voted Avandia be pulled from the shelves.
Will the TIDE study be in or out?
The advisory committee in a 19-11 vote also agreed that the TIDE trial should continue despite the skepticism the committee found with rosiglitazone.
The much debated TIDE (Thiazolidinedione Intervention With Vitamin D Evaluation) trial will look to answer outlying questions regarding rosiglitazone and its potential to cause CV events compared to pioglitizone and placebo. The trial so far has enrolled 1,000 patients, a low number from 16,000 trial enrollees expected.
Hertzel Gerstein, MD, of McMaster University in Hamilton, Ontario, who presented the randomized, double-blind TIDE trial, estimated that the enrollment number would be 50 percent higher if the politics surrounding rosiglitazone were different.
In criticism, Nissen calculated that the TIDE trial would take at least another eight years to complete. “Given the drug’s [rosiglitazone] current usage, are we willing to wait that long with a drug with this hazard to wait for the completion of this trial?” he said.
Graham called the TIDE trial “misleading” and said that it is a “CV endpoint trial with no reference to CV disease” in its name. Additionally, he said the trial shows “no mention of Avandia as the sole source of CV concern.”
However, committee members say the trial is necessary if the FDA decides to keep rosiglitizone on the market, but question how the trial will play out if further restrictions and warning labels are placed on the drug.
“Everyone already agrees that it’s [rosiglitazone] bad for congestive heart failure so the question is how many more things does it [rosiglitazone] have to be bad for before you say enough is enough?” concluded Michael A. Proschan, PhD, of the National Institute of Allergy and Infectious Diseases, and an FDA adviser.