After Avandia: Does the FDA Have a Drug Problem?

A TIME investigation of Avandia's journey from lab to pharmacy reveals serious flaws in the way the Food and Drug Administration protects consumers from harmful medicines

  • Illustration by Sean McCabe for TIME

    (2 of 3)

    In public, SB executives defended the safety of their drug. At an April 1999 FDA public hearing featuring outside experts charged with recommending whether to approve Avandia, SB's head of research and development, Dr. Tadataka Yamada, maintained that Avandia had a "risk-neutral lipid profile" and "cardiovascular safety ... comparable to placebo and active comparators." FDA scientists disagreed. Concerned about the boost in LDL, FDA pharmacologists recommended against approving the drug. Dr. Robert Misbin, the FDA's medical officer, said he would support approval only if the company committed to a thorough safety trial that would include monitoring for cardiovascular risks.

    The hearing committee, three of whose eight voting members had declared financial conflicts of interest in the case, debated the heart issue and eventually recommended that the FDA approve the drug. (TIME requested the forms that waived the conflicts of interest; an FDA official declined to release them and said none of the conflicts involved a relationship with SB.) Then came the horse trading. After outside experts weigh in on a new drug, but before it receives final approval from the agency, the FDA and the drugmaker negotiate which tests the company will perform once large numbers of people are taking it on the open market.

    On May 5, 1999, SB sent its proposal for testing Avandia to the FDA. The company didn't want to do a long-term safety test at all. Less than a week later, in a letter to his superior, Misbin threatened to withdraw his approval recommendation, saying the risk of heart disease may be increased by treatment with Avandia and accusing SB of attempting to divert attention from dangers that Avandia might pose to patients, according to parts of the letter read to TIME.

    Then, right at the May 25 deadline for FDA approval, SB made an offer to focus its testing on the drug's ability, as compared with competitor drugs, to lower blood sugar. It was a side step from the question the agency wanted to answer about the drug's safety. Instead of focusing on finding out if Avandia posed a heart risk, SmithKline Beecham was going to run a trial its sales representatives could use to promote the drug. "It was really a marketing study," says Misbin now. But later that day, Dr. John Jenkins, the FDA's director for new drugs, accepted SB's proposal for testing the drug on the market and approved Avandia for sale. By agreeing to the company's version of the postmarket trial, scientists say, the FDA abdicated its responsibility to collect reliable data on Avandia's safety.

    Even with the FDA's help, the company had its hands full. In 1999, Dr. John Buse of the University of North Carolina at Chapel Hill, a diabetes expert, using slides that SB officials had presented at their approval hearing, did his own calculations based on the data. In speeches, he highlighted the fact that Avandia users experienced a more than fourfold rise in cholesterol compared with those taking a placebo. Because elevated cholesterol levels are a risk factor for heart disease, Buse wrote to the FDA commissioner, warning that Avandia could cause "adverse cardiac outcomes." In March 2000, officials with the newly merged GlaxoSmithKline got a copy of the letter and, Buse tells TIME, contacted his boss, accusing Buse of being a liar and being for sale, and saying he needed to be muzzled. The company's stock had dropped, and "they threatened to sue me for something like $4 billion, which was the loss of the company's valuation," he says.

    In the meantime, the company took measures to promote Avandia. In 2001, GSK worked on an article, later published in the American Heart Association's journal Circulation by Dr. Steven Haffner of the University of Texas Health Science Center at San Antonio, arguing that the class of drugs that includes Avandia could significantly reduce cardiovascular risk factors in animals. At meetings with doctors in 2001, GSK sales representatives denied Avandia had cardiac side effects, prompting the FDA to issue a public letter of warning against the company.

    Keeping the Public in the Dark
    By 2004, Avandia sales were earning GSK more than $1.5 billion a year in the U.S. alone. But as more people went on the drug, the picture on cardiovascular risk began to get clearer. GSK began a review of the drug's heart risks, and in 2005 and 2006 the company produced internal analyses showing 29% and 31% jumps in negative heart events. On May 9, 2006, the company provided these results to the FDA. The agency didn't immediately release those studies to the public, because its officials "didn't necessarily agree with some of the methodology used," says Dr. Janet Woodcock, head of the FDA Center for Drug Evaluation and Research. Instead, the FDA put its own statistician on the job. Just before Christmas that year, Misbin looked at the statistician's spreadsheet and found that "in virtually every trial, there were more cardiac events with Avandia than with the comparator," Misbin says. He was convinced enough to call his uncle, who was on Avandia, and advise him to ask his doctor to switch him to another drug.

    It was seven years after the drug was approved, and the dangers of Avandia had still not been made sufficiently clear to the public. The FDA was sitting on the new analyses, and GSK, the FDA discovered during an investigation by its inspections unit in the fall of 2007, had failed to report clinical data and other material from 15 tests of Avandia by the end of 2006, according to a March 25, 2008, warning letter to the company. With the company and the FDA maintaining tight control over the full database of information on Avandia's effectiveness and safety, there was little independent scientists could do to assuage their growing concerns about the drug.

    1. 1
    2. 2
    3. 3