(3 of 3)
Then came a bit of legal serendipity. As part of a settlement with the state of New York over GSK's nondisclosure of possible heightened suicide risk among teenagers taking its antidepressant Paxil, the company agreed to put all its recent clinical studies on a website. Aware of the growing concerns among clinicians about the risks posed by Avandia, in April 2007, Cleveland Clinic cardiologist Dr. Steven Nissen Googled the site and downloaded all of the available Avandia trials. After analyzing the 42 trials, he wrote up his findings and in May submitted them to the New England Journal of Medicine. He had found what GSK and the FDA already knew: a 43% higher rate of cardiac events among Avandia patients compared with those taking other drugs or placebos.
By chance, the New England Journal of Medicine chose as a prepublication reviewer of the Nissen article Haffner, the University of Texas doctor who was the lead author on the 2001 paper that had suggested that Avandia's class of drug could decrease cardiovascular risk. He faxed a copy of Nissen's article directly to GSK. Now GSK faced the threat of broad public awareness of the hazards of its drug. So with the clock ticking until Nissen's article was to be published on May 21, 2007, GSK harvested data on cardiac events from the recently completed efficacy trial Jenkins had signed off on back in 1999. Because the trial had been designed to show efficacy, not safety, its cardiac data were inconclusive.
As it prepared for the pivotal May 16, 2007, meeting in White Oak with FDA regulators, GSK came up with an additional counterattack to Nissen's study. Unlike the FDA, European regulators had insisted on a long-term cardiac-safety study, called RECORD, when they approved the drug. So GSK argued that the only prudent approach would be to let the RECORD trial run to completion in 2009 to reach a definitive answer on cardiovascular risk. To top agency officials, it seemed like a reasonable solution at the time. But three years and hundreds of millions of dollars in Avandia sales later, it turns out the RECORD trial may not be as reliable a study of cardiac risk as agency officials had hoped.
Regulator, Regulate Thyself
By 2008, after the painkiller Vioxx and the cholesterol-lowering medication Baycol were pulled from the market because of side effects and complications, House and Senate overseers both began investigations of the drug-approval process and the relationship between the drug industry and its federal minders. The Senate Finance Committee concluded in January 2010, after a two-year review, that GSK failed to promptly alert the FDA about Avandia's drug risks. In response, FDA commissioner Dr. Margaret Hamburg initiated another review of whether to keep Avandia on the market. As part of that review, FDA investigator Dr. Thomas Marciniak presented a devastating report on RECORD's shortcomings, detailing how the RECORD study minimized Avandia's heart risks: one death among the drug takers, for example, was missing from the final tally, and discrepancies in some cardiovascular data favored Avandia by a ratio of 4 to 1. The congressional investigation also uncovered e-mails indicating that GSK executives had managed to persuade the trial's supposedly independent steering committee to publish interim results that demonstrated how inconclusive the heart risk was. The trial's design, Marciniak found, was "completely inappropriate and biased."
For its part, GSK insists the drug is indeed safe and says it has played fair with the data. It lists multiple studies that are inconclusive or show no increase in heart risk for Avandia. It says it has disciplined the sales representatives who triggered the FDA admonition; it updated its report on Avandia tests after the FDA's 2008 warning letter. "GSK continues to stand behind Avandia," says spokesman Kevin Colgan. "The facts will support our position." Using the most powerful argument of all, GSK says diabetics desperately want and need drugs to lower their blood-sugar levels. All of this persuaded the FDA advisory group this July to narrowly vote not to pull Avandia off the market, citing a lack of strong evidence that it should be withdrawn. Eleven years after the drug was approved, neither GSK nor the FDA could yet prove Avandia was safe.
GSK certainly had reason to dispel the uncertainty if it could. In 2001, it ran a calculation of what the "net sales downside" would be if the cardiovascular "safety issue intensifies" and found that for 2002‑04, potential lost revenue amounted to $600 million, according to the civil-court proceedings obtained by TIME. GSK reported a 10% drop in profits for the fourth quarter of 2007, partly as a consequence of a drop in Avandia sales following the publication of the Nissen article. Now GSK has other financial concerns. In March the company put aside $3.5 billion for "legal and other disputes." In May it paid $60 million to settle 700 Avandia civil cases; in July it reportedly offered to pay $460 million to settle civil cases claiming the drug caused heart attacks. GSK declines to discuss the costs of or anything else about the investigation the FDA is undertaking against it.
That investigation may indicate that change is coming to the FDA. Government and independent watchdogs say the agency has made some progress. In 2009, government auditors found that the FDA had begun to bolster the role of drug-safety monitors. The Health and Human Services inspector general said last March that the FDA has boosted its prescription database of postmarket reports.
But none of that addresses the issue at the heart of the Avandia case. Science is often inconclusive, and the FDA rightly argues that surveys like Nissen's are sometimes wrong. Where the FDA fell down on Avandia was in allowing GSK to perpetuate the uncertainty about safety rather than clarify it. In 2007, the FDA gained new powers to require postmarket safety trials, but FDA leaders admit they're still learning how to use them. Grassley wants to give FDA safety monitors even more power, and former FDA chief Mark McClellan says the agency should use newly computerized medical records to track safety data in near real time.
But with the FDA ever more dependent on industry user fees, and with new drug-safety concerns emerging year after year, it will take more than faster data retrieval to restore the reputation of an agency that was once synonymous with trust in the public mind. The FDA will have to start forcing companies to be transparent and call them out on it when they're not. Says the medical reviewer Misbin: "Companies are always going to present their best face. It's our job to say no." In that sense, the FDA just needs to perform the task it was charged with more than 100 years ago: protecting the public interest by keeping industry honest.
With reporting by Susan Weill / New York