Rosiglitazone Meeting: Advisory Committee Delivers Mixed Message to FDA

Shelley Wood

July 14, 2010

July 14, 2010 ( Updated July 15, 2010 ) (Gaithersburg, Maryland)It remains unclear whether the most closely watched FDA advisory panel in years will prove to be the death knell or a life raft for rosiglitazone (Avandia, GlaxoSmithKline): while a recommendation to pull the drug from the market received the most votes out of five options, the other options, all of which involved leaving rosiglitazone on the market, taken together received more votes.

The muddled answer leaves the FDA in a curious quandary. On the one side, a majority of votes recommended keeping the drug on the market, but most with beefed-up warnings or restrictions. Viewed from a different angle, a majority recommended either removal or severely restricting access to rosiglitazone.

Those complex recommendations were the best the panel could come up with after two days of exhaustive debate, polarized opinions, and a puzzling mix of data that seemed to confuse as much as it informed. A hint of how this vote might go came earlier in the day, when the majority of panel members voted "yes" to whether they believed the data were "sufficient" to raise safety concerns about ischemic cardiovascular events with rosiglitazone as compared with nonthiazolidinedione (non-TZD) drugs and, even more so, as compared with pioglitazone (Actos, Takeda Pharmaceuticals).

But by contrast, all but one panel member also said that there were insufficient data to convince them that mortality risks are higher with rosiglitazone compared with non-TZD drugs, and just seven viewed data as "sufficient" to raise mortality concerns vs pioglitazone.

Day-Two Drama

Ultimately, however, 12 out of 33 voting panelists from both the Endocrinologic and Metabolic Drugs Advisory Committee and the Drug Safety and Risk Management Advisory Committee agreed that the wide range of studies, many "fatally flawed," point to a safety signal too strong to allow rosiglitazone to stay on the market. The remaining 20 votes, saying the drug could stay on the market, were spread across three of the other four options.

No panel member voted to leave the drug on the market and remove the warning labels. A full 10 said the drug should stay on the market with additional warnings and severe restrictions on its use; seven suggested additional warnings should be sufficient. One member abstained--Dr Michael Proschan (National Institute of Allergy and Infectious Diseases, Bethesda, MD)--and three said no changes are needed to the labeling.

Of note, several members who voted for choice D (to leave the drug on the market, but with restricted access) characterized themselves as "wavering" or "oscillating," between full withdrawal and leaving rosiglitazone on the market with harsh restrictions--a position subsequently dubbed the "D-and-a-half-club."

Dr Ruth Day (Duke University, Durham, NC) explained why she didn't vote E, for the drug's withdrawal, by saying, "The main reason I caved and put D is because D-and-a-half was not available. . . . I would recommend that something very tough be instituted [for prescribing rosiglitazone]. If something like that is not put into place, then I would be pushed toward E."

Others, however, were strongly committed to withdrawal. Pharmacist Dr Mark Woods (Saint Luke Hospital, Kansas City, MO) explained: "The big reason I voted for E over D is that I'm very skeptical about [restricted prescribing] and black-box strategies changing prescribing habits. I'm just not sure that works when the rubber hits the road."

Decisions From the Heart (Doctors)

Cardiologists on the panel did not appear to vote as a block on the withdrawal question, despite predictions that they'd be less likely than endocrinologists to have concerns about a lack of good alternatives to rosiglitazone. Dr Curt Furberg (Wake Forest University Health Sciences, Winston-Salem, NC) and Dr Marvin Konstam (Tufts-New England Medical Center, Boston, MA) both voted to recommend the drug's withdrawal, while Dr John Teerlink (San Francisco VA Medical Center) voted for D. Dr Sanjay Kaul (Cedars Sinai, Los Angeles, CA), who chaired the 2010 AHA/ACC scientific advisory on TZDs, voted for C--to allow continued marketing, but with tougher warning labels. Finally, Dr Arthur Moss (University of Rochester, NY) and Dr David Capuzzi (Lankenau Institute for Medical Research, Philadelphia, PA) voted for B--to keep the drug on the market with no labeling changes.

Explaining his decision, which he said he expected to be "in the minority," Moss said he could have also voted C or D, but definitely not E. He chose B because "I wanted to send a message to the FDA, because I thought it would be appropriate that they [do] a proper trial, [which] should have been done going all the way back to 1993," when the drug developer first submitted its new drug application. "The trial still needs to be done."

Teerlink, likewise, said his D vote carried a message.

"I'd like to send a signal not only to these sponsors, but also to other future sponsors that the appropriate assessment of risk and safety is absolutely [essential]. And it's also a message to the FDA--we get in these binds because of surrogate end points and improperly powered initial studies."

Turning to TIDE

Those kinds of messages have implications for the topic tackled in the second major panel vote, where panel members were asked to weigh in on what should happen with the controversial Thiazolidinedione Intervention with Vitamin D Evaluation (TIDE) trial.

At the day's dimly remembered outset, Dr Herzel Gerstein (McMaster University, Hamilton, ON) had delivered an animated and passionate rationale for why the TIDE trial should continue. Tuesday, several speakers, most notably FDA reviewer Dr David Graham, had argued with equal zeal that the trial was unethical and must be stopped.

Gerstein, for his part, argued that randomized controlled trial data were essential to resolve the safety issue once and for all: "We can be seriously misled by the types of data that were presented and discussed all day yesterday," he said.

Later, when asked to weigh in on TIDE's continuation, panel members voted 19 to 11 that the trial should continue, with two abstentions, and one "no" vote (absent). (An earlier version of this story reported the vote as 20 to 10, which was how the votes first came in. One panel member later said that he had hit the wrong button.)

The TIDE question was particularly tricky for the 12 panel members who thought the drug should disappear altogether. Konstam summed up the view of several on the panel saying, "The dilemma is, if the drug is still on the market then the trial has to be done, because there's this adverse safety signal and we can't just leave it hanging there." The problem is that the point of doing such a trial would be to prove that the safety concern is real. "I'm not sure you can really [get] adequate consent for that kind of proposition," Konstam said.

Several panel members suggested that ideally TIDE could have its design overhauled, with Moss taking it one step further and suggesting that the trial be scrapped altogether.

"I would look upon the first 1000 patients [enrolled in TIDE] as kind of a preliminary pilot study," Moss said. "A definitive study needs to be done, and this is the time that the FDA can demand such."

One of the 11 members who voted no, Dr Howard Mann (University of Utah, Salt Lake City) said he did so for multiple reasons, but chief among them was the inevitable lack of equipoise among practicing clinicians, regardless of what the FDA decides.

What Next?

The FDA does not need to follow the advice of its advisory committees, although it usually does. In this case, which has received intense public scrutiny, it's difficult to predict how the FDA will move forward. Media reports propagating like mushrooms as soon as the withdrawal vote was announced largely interpreted the split decision as a win for rosiglitazone and GlaxoSmithKline, whose stock price bumped higher when the vote first passed.

Should the FDA follow the bulk of today's votes, most of which recommended no withdrawal, but tougher labeling and restrictions, it remains to be seen just how tough those restrictions can get. And as several panel members noted, the tougher these restrictions, the more TIDE will struggle to enroll and the more patients will be leery of taking the drug.

The More Things Change . . . 

Rosiglitazone was first approved in 1991 and swiftly rose to blockbuster status, taken by millions of patients with type 2 diabetes and rushing up revenues for the manufacturer. A controversial meta-analysis published by Dr Steven Nissen and Kathy Wolski (Cleveland Clinic, OH) precipitated an earlier FDA advisory committee meeting in 2007, where panel members ultimately acknowledged seeing a signal of risk with rosiglitazone but advised in a vote of 22 to 1 to keep the drug on the market.

Three years later, with more studies and issues to consider, a committee including many of the same members as the one in 2007 has not managed to decisively clarify that earlier advice.

Speaking with heartwire after the meeting wrapped up Wednesday, Nissen expressed the hope that the FDA had built in enough "wiggle room" to be able to withdraw the drug, despite the mixed advisory-panel recommendations. After all, he pointed out, a clear majority of panel members voted either for withdrawal or "such severe restrictions that it would be virtually like withdrawal." Option D, he pointed out, which received 10 votes during the withdrawal question, entailed having physicians require specialized training to prescribe the drug and obtaining specialized informed consent from patients. "It's sort of a compassionate-use provision. I think that 99% of the people on the drug now wouldn't be eligible to get it under this restricted-use provision."

It's an "age-old" problem, he said, sounding dispirited about the day's events. "No FDA advisory panel in history, to my knowledge, has ever voted to withdraw a drug. This is about as close as it ever gets."

A Broken RECORD?

Explaining their votes on both the withdrawal and TIDE questions, many panel members returned to an issue that had consumed them Tuesday--namely, the trustworthiness of the RECORD data and whether the company was unduly involved in the trial's conduct.

Of note, Wednesday morning, GlaxoSmithKline representatives, as well as Dr Philip Home (Newcastle University, UK), RECORD steering committee chair, sought to reassure panel members, specifically rejecting statements from Tuesday that suggested the company had tweaked end points after peeking at the data. That notion, Home explained, came from a typo in the steering-committee minutes and was categorically "not true."

GlaxoSmithKline issued a statement following the meeting Wednesday saying it will work closely with the FDA going forward and advised patients to take any questions to their physicians [1]. Pending any formal decisions, Avandia remains "available to physicians and appropriate patients," the statement reads.

For the full minute-by-minute record from the day's meeting, read's account on twitter.

Editor's note: An earlier version of this story erroneously identified Teerlink as the abstaining vote for the withdrawal question.