Another Leaked Rosiglitazone Manuscript? Controversy Spikes as JAMA, Archives Publish New Papers

Shelley Wood

June 28, 2010

June 28, 2010 (Chicago, Illinois) — The heat facing rosiglitazone (Avandia, GlaxoSmithKline) in the run-up to the drug's July FDA rereview is spiking to new heights on allegations from FDA scientist Dr David Graham that someone else at the agency leaked his rosiglitazone manuscript to a popular blog in the hopes that the Journal of the American Medical Association (JAMA) would opt not to publish it.

As previously reported by heartwire , the Pharmalot blog reported that it had received a copy of the manuscript and an email sent by Graham to FDA commissioner Dr Margaret Hamburg, the email stating concerns that his superiors at the FDA might stall or block the publication of observational data showing the risks of rosiglitazone to be higher than those of pioglitazone (Actos, Takeda Pharmaceuticals) in Medicare subjects. Pharmalot blogger and journalist Ed Silverman posted both the email and the draft manuscript online.

Now Graham's paper is, in fact, being published today as an early release in JAMA [1], and in an interview with heartwire , Graham explained that he has no idea who leaked his email and manuscript, but he believes it was a calculated attempt by someone at the FDA to prompt the journal to reject it, based on submission policies that restrict consideration of papers to those that have not been published previously in print or electronic format or shared with the media (outside of media activities associated with meeting presentations).

"In my view, and in the view of JAMA for sure, this was an attempt by people at the FDA to block the publication [of this data]," Graham said. "The idea was that JAMA would invoke the Ingelfinger rule, and the paper wouldn't get published, and it wouldn't have 'credibility.' "

When he found out about the leak, Graham says he called a JAMA deputy editor immediately and explained that he thought it was "an act of sabotage."

"Unfortunately, the FDA doesn't appear to be investigating who is responsible for the leak, and I'm disappointed about that," he said. "The whole point of the study was to address an important public-health issue, and this was an attempt by people who are charged with protecting the public health to protect an unsafe drug."

The JAMA editors, he continued, "were very reassuring and basically at the end of the day said that this is a poison pill and we aren't going to swallow it. We know it's an important study, and if this is the length the FDA is going to to stop it, it must even be more important than we recognized."

In Response

A JAMA spokesperson provided heartwire with a statement saying that the journal believed Graham's paper had been leaked to the media without his knowledge or consent. "In that case, the JAMA editors do not consider Dr Graham to have violated JAMA policy regarding prepublication release of information. More important, unlike the leaked paper, the study by Graham et al being published by JAMA has been peer reviewed, evaluated by content and biostatistical experts, and appropriately revised, just as is the custom for all published JAMA research papers."

An FDA spokesperson, meanwhile, denied that the agency was turning a blind eye. "We are investigating the leak and are very upset about it," FDA press officer Karen Riley told heartwire .

As to why anyone at the FDA's Center for Drug Evaluation and Research (CDER) would want to suppress or derail new data, Graham says he doesn't know, although the repercussions for an agency already under minute scrutiny by other branches of government no doubt play a role. "It's not only that [the FDA] approved it in the first place," says Graham. "They subsequently had the opportunity to take it off the market and didn't."

Graham's paper is one of two studies, plus an editorial, being released Monday morning by the JAMA/Archives publishing group, both showing that the cardiovascular risks of rosiglitazone exceed those of pioglitazone. The other paper is an updated meta-analysis by Dr Steven Nissen and Kathy Wolski [2], whose 2007 meta-analysis showing an increased risk of MI and a trend toward increased cardiovascular death with rosiglitazone was published in the Archives of Internal Medicine. Three years ago, the original Nissen-Wolski meta-analysis [3] precipitated the first rosiglitazone review by the FDA's joint Endocrinologic and Metabolic Drugs/Drug Safety and Risk Management advisory committees.

Medicare Data

Graham et al's paper is a nationwide retrospective analysis of 227 571 Medicare patients (mean age 74) treated with either pioglitazone or rosiglitazone over a three-year period, then followed for three years after initiation of treatment. The investigators, from the FDA and the Centers for Medicare & Medicaid Services (CMS), found that after adjustment for potential confounders, rosiglitazone was associated with a significantly increased risk of stroke, heart failure, death, and a composite of AMI, stroke, heart failure, or death, as compared with pioglitazone. As previously reported, only AMI was not significantly increased.

Adjusted Hazard Ratios and Numbers Needed to Harm, Rosiglitazone vs Pioglitazone

End point Adjusted HR 95% CI Number needed to harm
AMI 1.06 0.96–1.18 NS
Stroke 1.27 1.12–1.45 313
Heart failure (HF) 1.25 1.16–1.34 106
Death 1.14 1.05–1.24 222
AMI, stroke, HF, death 1.18 1.12–1.23 60

Discussing the results with heartwire , Graham explained that AMIs were captured only by data from patients who survived to reach the hospital; in this older cohort of patients, fatal out-of-hospital AMIs were not captured but likely drove the significantly increased risk of death and the combined end point.

The most important thing, he continued, was the number needed to harm. "If you treat 60 people with this drug for a year, one more will have a heart attack, death, stroke, or hospitalization event than they would if they had been treated with Actos," he said. "And that ends up being multiplied by hundreds of thousands of people taking the drug at any given time; you start talking about huge numbers."

Rosiglitazone Meta-Analysis, Revisited

The Nissen-Wolski analysis includes an additional 14 studies on top of the 42 included in the original meta-analysis, using only randomized trials of rosiglitazone that reported cardiovascular events. As with the earlier rosiglitazone meta-analysis, the updated paper found a significantly increased risk of MI but not CV mortality when all trials were included. When RECORD was excluded--because it was unblinded, open-label, and had implausibly low rates of MI, suggesting inadequate tracking of events, says Nissen--the hazard ratios for both of these end points climbed still higher.

"When you put all the pieces together, you see anything from a 28% to a 39% increase in the risk of AMI, and the number needed to harm is 37 or 52, depending on whether or not you include RECORD," Nissen told heartwire . "That's about as bad a number-needed-to-harm value as I've certainly ever seen, and I think it provides everything we know about the drug from randomized controlled trials."

MI and CVD Rates

End point Odds ratio 95% CI
MI 1.28 1.02–1.63
CV mortality 1.03 0.78–1.36
MI (RECORD excluded) 1.39 1.02–1.89
CV mortality (RECORD excluded) 1.46 0.92–2.33
MI, zero-event studies included 1.28 1.01–1.62
CV mortality, zero-event studies included 0.99 0.75–1.32

Addressing a criticism levied at their earlier meta-analysis by Drs Sanjay Kaul and George Diamond (Cedars-Sinai Medical Center, Los Angeles, CA) in a "rosiglitazone reanalysis" [4], Nissen and Wolski also reanalyzed their data using a different methodology, to take into account trials that had zero events. Here, too, says Nissen, results were similar.

Even in 2007, Nissen notes, the point estimates obtained by Kaul and Diamond in their reanalysis were similar to those of his and Wolski's paper, although Kaul and Diamond's confidence intervals were wider. "Frankly, I think their criticism [of the 2007 paper] may have hurt a lot of patients."

When the two studies are reviewed side by side, it's notable that MI rates were significantly higher in the Nissen-Wolski paper, but not in Graham's Medicare analysis; by contrast, heart-failure rates were higher in Graham et al's study, but not in the Nissen-Wolski paper.

Both Nissen and Graham attribute this disparity to the 20-year mean age difference in the two populations. Graham also added that he thinks the FDA has unduly downplayed the incidence of heart failure with rosiglitazone, moreover making the blanket statement that this is a class effect without actually testing whether heart-failure rates differ between rosiglitazone and pioglitazone.

Why Rosiglitazone?

Nissen also noted that the findings reflect the Ontario rosiglitazone study led by Dr David Juurlink (University of Toronto, ON), who wrote an editorial accompanying the Medicare data in JAMA [5]. Juurlink has previously told heartwire in no uncertain terms that he believes the ongoing Thiazolidinedione Intervention with Vitamin D Evaluation (TIDE) trial comparing the two thiazolidinediones should be stopped. Now, in his editorial, he sums up the available evidence by concluding that there is no direct evidence that rosiglitazone prevents vascular events in diabetics but there is mounting evidence that rosiglitazone is less safe than pioglitazone. While the American Heart Association and American College of Cardiology have called for more trials, the American Diabetes Association and European Association for the Study of Diabetes have advised against rosiglitazone use, he notes.

"The latter view incorporates a simple fact that has frequently gone overlooked: rosiglitazone confers no therapeutic advantage over pioglitazone. Whether rosiglitazone and pioglitazone really do have different cardiovascular safety profiles is an intriguing question, but one with a misplaced focus. Accumulating concerns about rosiglitazone make it difficult to advance a cogent argument regarding why, exactly, a patient might want to receive the drug or why a physician would choose to prescribe it when there is an available and quite possibly safer alternative."

Tension Builds

The fate of TIDE, rosiglitazone, and to some extent the FDA's reputation in the public eye now hinges on the upcoming two-day FDA meeting. People watching the developments closely have predicted that should the FDA accept the kind of data being put forward by Graham and Nissen, stop TIDE, and yank the drug, heads at the agency may roll.

Graham hopes for two things to come out of the hearing: "One, the TIDE trial needs to be stopped, and Avandia needs to come off the market," he said. Next, echoing a comment made many times by Nissen, Graham wants the agency to create an agency for postmarketing surveillance that is completely separate from the agency that approves drugs in the first place.

"And if the FDA isn't prepared to act, I'm hoping physicians and patients becoming aware of our study will decide that safe alternatives exist," he added.

GlaxoSmithKline, meanwhile, said last week that it is aware of the new data and provided heartwire with a statement echoing points it has made in the past--namely, that randomized trials are the "gold standard" and that it looks forward to the FDA hearing. It, too has updated a prior meta-analysis and says it "does not show an increase in myocardial ischemia [HR=1.1, 0.89-1.35; p=0.38].  There also have been several new observational studies published since 2007. These studies have tighter confidence intervals, with risk ratios that are very close to 1.00, indicating no difference in the risk of MI between rosiglitazone and pioglitazone." It will present that data at the meeting.

Nissen, who is expected to testify at the FDA meeting, seems to be looking forward to it and hinted that the JAMA/Archives publications are only the first of many "big surprises to come" in the weeks leading up to the meeting itself.

Graham, on the other hand, does not sound overly optimistic about the two-day hearing, which he described variously as "a media event, not a scientific event" and "kabuki theater, with the questions, presenters, and presentations carefully crafted."

He continued: "I've heard rumors that they might be inviting back to be on this panel the same people who voted on the drug the last time." In other words, the people who determined that while there were cardiovascular risks of rosiglitazone, the benefits outweighed these risks. "Talk about kangaroos back in the court," he quipped.

"Why would you vote that [rosiglitazone] increases risk and that it should stay on the market? If that doesn't disturb you, you're brain-dead."


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.