Avandia in the Hot Seat as Key Vote Nears

The diabetes drug Avandia is about to get its reckoning, as advisers to the U.S. Food and Drug Administration (FDA) prepare for a 2-day meeting starting tomorrow that will culminate in a vote about whether Avandia (whose generic name is rosiglitazone) should be removed from the market. Things aren't looking good for the drug, especially after the FDA released more than 700 pages of documents on Friday in which the agency's own officials expressed grave concerns about the cardiac safety of Avandia and the conduct of the company that makes it, GlaxoSmithKline (GSK).

Especially damning was FDA's assessment of a clinical trial called RECORD, which Glaxo was asked to run by European drug regulators—a study designed to give an extended view of Avandia's safety once the drug was on the market. In 2007, Glaxo released an interim analysis saying Avandia didn't cause heart attacks or other cardiac problems; the company has frequently cited RECORD to defend the drug's safety.

In its Friday documents, FDA took a very different position. The RECORD study had "many design limitations," wrote Thomas Marciniak, medical team leader of FDA's Division of Cardiovascular and Renal Products, in an exhaustive review of the trial. Among other things, patients and doctors knew whether they were receiving Avandia or a different drug, which can lead to bias, and the FDA team was also concerned that the definitions of different "endpoints"—whether patients developed heart problems—were ambiguous.

In addition, Marciniak wrote, there were "study conduct problems." In one case, according to documents from GSK that appear in FDA's documents, the company requested that a patient who died of what appeared to be a pulmonary embolism also be listed as having had a heart attack; the individual was not receiving Avandia, so more heart attacks in this cohort would help the drug's safety profile. (An autopsy had revealed a heart attack, but the study protocol didn't allow for inclusion of heart attacks that turned up only on autopsies but weren't detected by physicians.)

Although FDA would have counted this as a heart attack without GSK's intervention, that's not the point, Marciniak wrote: "We judge that it was highly inappropriate for GSK to have attempted to influence the work of the CEC Coordinating Center [which independently assesses side effects in the trial] in this way. We have no mechanism for verifying whether such influences were common or rare."

Marciniak also found, in reviewing the RECORD data, "four cases whose handling we consider to be completely unacceptable. All four of these cases are rosiglitazone cases; the mishandlings of all four favor rosiglitazone." They include a heart attack that was "deleted" from the record 16 months after it happened.

FDA also found 66 other "case errors" in how study data were reported, including questionable diagnoses.

In 166 pages of documents from Glaxo released by FDA, the company defended its drug and the RECORD trial. "The results from RECORD remain robust and reliable," it wrote, though it did not address Marciniak's specific allegations about the company's conduct.

Avandia has been in the hot seat since 2007, when an independent meta-analysis of Avandia trials reported that the drug increases the risk of heart attacks. FDA advisers voted overwhelmingly that the drug carries this risk but also decided that the risk wasn't enough to pull it from the market. Instead, FDA added a warning label to the drug. A couple weeks ago, new published studies asserted again that Avandia was risky.

In light of the latest furor, Glaxo announced Friday, European regulators are also reviewing Avandia's safety profile. FDA advisers will vote Wednesday afternoon whether to recommend keeping the drug on the U.S. market or removing it.