Dangerous practices: critics see flaws in drug-safety monitoring.

It was the best of statins, it was the worst of statins. The anticholesterol drug cerivastatin, sold under the brand name Baycol, was the most potent medicine in its class in the late 1990s. In 2001, however, it was ripped from the marketplace after being linked to an unusually high incidence of a rare, but sometimes deadly, side effect. In certain respects, the compound's meteoric career portended the recent deluge of bad news about one drug after another.

Baycol, made by Bayer of Leverkusen, Germany; appeared safe enough in the manufacturer-funded studies that the Food and Drug Administration considered before approving it for prescription use in the United States. In clinical trials, researchers had specifically looked for certain side effects that were already, albeit rarely, associated with other statin drugs. In reports in scientific journals, they concluded that Baycol was no riskier than the other statins. What's more, it was effective at much lower doses than the other drugs.

Between Baycol's nationwide launch in 1998 and its sudden retirement in 2001, U.S. doctors wrote nearly 10 million prescriptions for the pill. It was during that flurry of use that the drug's dark secret gradually emerged. Baycol caused a potentially deadly form of muscle degeneration called rhabdomyolysis more often than the other statins did.

All pharmaceuticals carry some potential for causing side effects. In vetting therapeutics, FDA hews to what it has called a pragmatic standard: Do the demonstrated benefits outweigh the known, risks? However, risks sometimes don't become evident until a drug has been used by millions of people. Another problem is that tight regulation can, while reducing the risks for many people, deny others an irreplaceable benefit. Recent studies of drugs that have been removed from the market are providing information on the shortcomings of regulations and on the risk-benefit dilemma.

SECOND LOOK A reexamination of Baycol's history illustrates the problems. Using records from II U.S. health care plans that paid for statin prescriptions for more than 250,000 people between 1998 and 2001, David J. Graham of FDA's Office of Drug Safety in Rockville, Md., and his colleagues determined the risk of rhabdomyolysis for users of Baycol and the other statin drugs available during that period.

For every 10,000 person-years of therapy; Baycol caused more than 5 cases of rhabdomyolysis. Other statins carried just one-twelfth as much risk. When Baycol was used with a triglyceride-lowering drug--a fibrate named gemfibrizol--the rhabdomyolysis rate skyrocketed to more than one case per year for every 10 people treated.

Graham's team reports those findings in the Dec. 1, 2004 Journal of the American Medical Association (JAMA). The study is consistent with two reports, both published the year after Bayer pulled Baycol from the market, that suggested a higher incidence of rhabdomyolysis with Baycol than with other statins.

Critics of Bayer claim that the company had become aware of the hazards even as it continued to sell Baycol and that it imperiled its customers by choosing not to disclose those risks fully. Several thousand lawsuits have been brought against the company.

Four researchers who were paid by prosecutors to testify as expert witnesses studied the medical literature and internal company documents that were made public during the legal proceedings. Reporting in the same issue of JAMA in which Graham's report appears, they conclude that the data Bayer released while marketing Baycol didn't tell the whole truth about the drug.

The company selectively disseminated data on the drug's effectiveness and closely guarded information about its hazards, say Bruce M. Psaty of the University' of Washing*on in Seattle and his colleagues.

For example, they say that by December 1999, the company's analyses of physicians' reports suggested that, in the absence of fibrate use, the rhabdomyolysis risk associated with Baycol was 10 times that of other statins. With fibrate use, Baycol users were also "at a remarkable disadvantage," a company document stated in March 2000.

Postmarketing regulations require that companies collect, review; and forward to FDA all physicians' reports of possible drug side effects. The data, which are sometimes reanalyzed by FDA, are available to the public upon formal request.

Bayer communicated all of the required information to FDA, says Joseph D. Piorkowski Jr., a Washington, D.C.-based lawyer for Bayer.

In approving a new drug, FDA may demand that a company conduct additional safety trials, but the agency can't enforce these postapproval studies, and more than half of those agreed to by manufacturers never occur, according to a Department of Health and Human Services report in the March 15, 2004 Federal Register.

The story of Baycol is just one example of a pharmaceutical company keeping quiet on ominous findings, Psaty and his colleagues say. Although it jeopardizes public health, this kind of data filtering is unlikely to end unless government regulators get more control over the monitoring of drug safety, the critics argue.

The system needs reform, says .lAMA Editor-in-Chief Catherine D. DeAngelis. Today, postapproval monitoring is "almost strictly in the hands of the pharmaceutical companies" she says, which gives companies too much opportunity to put a spin on both the promising and the foreboding data that they might have on a drug.

Furthermore, Psaty and others say, trends in the pharmaceutical business have accelerated the approval of new drugs in the United States, thereby putting them in widespread use before long-term safety data have accumulated.

The pharmaceutical industry and FDA disagree with the charges that eagerness to sell drugs is blinding them to safety concerns and say that they're addressing problems with drug safety.

In a direct response in JAMA to the account by Psaty's team, Piorkowski points out that in late 1999, Bayer changed package labels to warn against using Baycol with a fibrate. According to the best analysis available when Baycol was on the market--an unpublished company study--the danger appeared to be limited to people using both drugs at once, Piorkowski maintains.

Determining that Baycol's risks exceeded those of other statins was a scientific riddle that took time to solve, says Brian L. Strom of the University of Pennsylvania School of Medicine in Philadelphia. Side effects data that doctors report to drug companies are too piecemeal for reliable comparisons of different drugs' risks, he says.

Once rigorous studies revealed the unacceptable risk, Baycol got buried promptly, Strom says in a JAMA article that Bayer submitted to the journal. "Although hindsight always raises questions about whether a problem could have been detected earlier, the events surrounding cerivastatin serve as a clear example of how the system should work," he says.

Nevertheless, Strom adds, "there is indeed a conflict of interest in asking industry to monitor its own drugs."

PAINFUL OUTCOMES One particular shift in the relationship between the pharmaceutical industry and FDA has exacerbated that conflict, says Sidney M. Wolfe of Washington, D.C.-based Public Citizen. Since 1992, federal law has required pharmaceutical companies to pay fees to FDA to partially defray the expense of reviewing new-drug applications. After these "user fees" were instituted, the average time that a new drug takes to obtain FDA approval fell from 27 months to 15 months, enabling doctors to prescribe new medications a year earlier than they otherwise would have.

But user fees generate a conflict of interest at the agency because it is now partly funded by the companies it regulates, says Wolfe. The proportion of drugs that later prove too dangerous to remain on the market climbed from 2.0 percent for approvals granted between 1989 and 1992 to 3.5 percent for those made between 1993 and 2000, according to a 2002 report of the Government Accountability Office.

"To me, that's a red flag," says DeAngelis.

Perhaps the most widely felt drug withdrawal has been that of the arthritis pill rofecoxib, or Vioxx. Manufacturer Merck of Whitehouse Station, N.J., initiated that recall last September when it announced that an ongoing, company-sponsored trial showed increases in the risks of heart attack and stroke associated with the drug (SN: 10/30/04, p. 286).

What's more, later in 2004, FDA issued a warning about heart risks linked to a related drug, valdecoxib (Bextra), and manufacturer Pfizer stopped consumer advertising of a third such drug, celecoxib (Celebrex), because of worrisome new cardiovascular findrags. An earlier study, completed in 2000, had suggested a link between Celebrex and cardiovascular problems, but Pfizer did not make the relevant data publicly available until this week.

The first clinical study to hint at a heart attack risk associated with Vioxx was funded by Merck and published in 2000, the year after the drug's approval. However, the study concluded that Vioxx was not at fault. Several smaller, earlier studies had not shown a significant sign of problems.

Combining data from that trial and from the earlier studies, researchers at the University of Berne in Switzerland recently conducted their own analysis. They concluded that the cumulative data provided significant evidence of Vioxx's hazards by late 2000. Around that time, advocacy groups such as Public Citizen and some eminent U.S. cardiologists began speaking out about the drug's risks.

Yet neither Merck nor FDA took Vioxx off pharmacy shelves then. The Swiss team, which reports its analysis in the Dec. 4, 2004 Lancet, suggests that FDA's relative inaction during that interval "has raised major concerns about the undue control of industry over postmarketing safety data."

In response to public furor following the Vioxx recall, FDA commissioned an external audit of its practices and announced steps to improve drug-safety monitoring and to communicate uncertainties to drug makers, academic researchers, and the public. "[T]hese initial steps," DeAngelis and two colleagues state in a Dec. 1, 2004 JAMA editorial, "are insufficient to dispel the perception that the agency appears to be unduly influenced by industry."

However, even among those who agree with that assessment, little consensus exists on what should be done.

"We certainly need an independent office of drug safety" says Psaty. The current system discourages the government from recalling a dangerous drug, he says, because in issuing a recall, FDA must essentially admit its initial authorization was mistaken. He and his colleagues also advocate government reassessment of all drugs 5 years after initial approval, a practice already in place in Europe.

Strom, on the other hand, argues that carefully selected academic centers should receive more government funds to formulate and publicize drug-prescribing guidelines.

Academic institutions, medical associations, and government agencies are all mired in industry ties, says Jerome P. Kassirer, former editor-in-chief of the New England Journal of Medicine and author of On the Take: How Medicine's" Complicity with Big Business Can Endanger Your Health (Oxford University Press, 2004).

Kassirer says that a cautionary example comes from the National Institutes of Health-affiliated panel that last July promoted broader use of statin therapy for people with high cholesterol (SN: 7/24/04, p. 62). Nine of the panel's 10 members acknowledge financial relationships with at least one statin manufacturer. Such conflicts of interest may make many guidelines presented by expert panels biased toward encouraging enthusiastic prescribing of drugs, he says.

To make financial conflicts transparent, all physicians' ties to drug companies, including gifts received, should be listed in a public database, Kassirer says. The Center for Science in the Public Interest, a Washington, D.C.-based nonprofit group, has taken a step in this direction (www.cspinet.org/integrity).

LOSING PATIENTS Inappropriate use of a drug can upset the scales of safety by increasing the population's risk of side effects without producing additional benefits. The consequences can adversely affect the users who most need the medication.

Recalling a drug that has substantial side effects spares some people from injury, but it may deprive others of a powerful tool that would probably have done them more good than harm.

For example, when Merck recalled Vioxx, DeAngelis estimates, "roughly 20 million people were taking it, and it might have been the best drug for, say, 1 million of them." The smaller group includes people for whom other pain relievers are either ineffective or cause side effects that outweigh Vioxx's risks.

Strom agrees that Vioxx was "vastly overused." But he adds, "For the very small minority of people who truly needed it, [Vioxx] would still be useful [and] should still be on the market today."

Inappropriate prescribing also forced Baycol into retirement, Strom says. "If Bayer had been able to eliminate use of Baycol with fibrates, Baycol would probably still be on the market," he says.

Regulatory action may yet kill other effective medications that, used more discerningly, could genuinely benefit an important minority, of patients. One drug that might face that fate is the statin rosuvastatin, which manufacturer AstraZeneea of London launched in September 2003 under the brand name Crestor.

Although his supervisors at FDA disagreed, Graham warned the Senate last November that he considers rosuvastatin unsafe because evidence ties it to kidney failure and a higher incidence of rhabdomyolysis than with the other statins on the market.

Of all statins, Crestor was the one most comprehensively studied before it was marketed, counters AstraZeneca's Gunnar O. Olsson of Molndal, Sweden. A high dose that caused side effects in early trials was never sold, he notes. Postmarketing statistics compiled by FDA show that through August 2004 at least 65 Americans had developed rhabdomyolysis while taking Crestor, Wolfe reported in the Oct. 30, 2004 Lancet. Public Citizen petitioned FDA to ban the drug last March and lists it on www.worstpills.org.

Strom says that, if prescribed appropriately, Crestor may produce benefits that outweigh its risks but the drug should be used only in cases where older statins have failed to lower cholesterol sufficiently.

Not everyone is adhering to that principle. Researchers at the Utrecht Institute for Pharmaceutical Sciences in the Netherlands reported in the Oct. 30, 2004 Lancet, that 42 percent of people using Crestor in a region of that country were receiving a statin for the first time. Furthermore, 23 percent of Crestor users there are older than 70 or simultaneously use a fibrate--factors that should rule out therapy with Crestor because they could raise the risk of rhabdomyolysis, according to Eibert R. Heerdink and his Utrecht colleagues.

If inappropriate use causes preventable side effects, as Wolfe and Graham suggest, then Crestor could get the hook just as Baycol and Vioxx did. For people who truly need Crestor, says Strom, its premature exit would be a step back in medical progress.