Feature: Why drugs get pulled off the market
by Michelle Meadows, for FDA Consumer Magazine
"But aren't drugs supposed to be safe?"
According to Janet Woodcock, M.D., director of the Food and Drug Administration's Center for Drug Evaluation and Research (CDER), people tend to ask that question a lot when a drug is taken off the market. The FDA's mission is making sure that drugs are "safe and effective." So what does "safe" really mean?
When it comes to any drug, "safe" means that the benefits of the drug outweigh the risks for the population the drug is intended to treat and for its intended use. "Safe does not mean harmless," Woodcock says. "Every drug comes with risks, and our tolerance for risk is higher for drugs that treat serious and life-threatening illnesses. There is no question that cancer drugs can be highly toxic. But they also save lives."
If the FDA decides that a drug's benefits outweigh its risks, the agency approves the drug for marketing. Approved drugs continue to be evaluated through postmarketing surveillance -- a system that monitors a drug's safety on an ongoing basis. Postmarketing surveillance seeks to identify problems that weren't observed or recognized before approval and any problems that might arise because a product isn't being used as anticipated.
The goal is to catch any bad news right away so that the FDA and drug companies can act quickly and communicate new risk information to consumers and doctors. CDER evaluates required reports from drug companies, which must promptly pass on any report they receive of a serious adverse reaction that isn't already described in the drug's labeling. CDER also relies on MedWatch, the system through which consumers and health professionals voluntarily report adverse events associated with all products the FDA regulates.
When the FDA receives reports of significant new adverse events, the agency evaluates them for their seriousness and the likelihood that they were caused by the drug. To the extent possible, the agency also considers how the toxicity compares with other treatments for the same disease. Ultimately, of course, the critical question is: Do the benefits of this drug still outweigh its risks for the population described in the labeling? In many cases, that question cannot be answered immediately, and more reports must be considered. Sometimes, the impact of labeling revisions needs to be assessed.
Usually, when important new risks are uncovered, the risks are added to the drug's labeling and doctors are informed of the new information through letters and other education. It's only rarely that the approval decision on a drug needs to be reassessed and changed. A conclusion that a drug should no longer be marketed is based on the nature and frequency of the adverse effect and how the drug compares with treatment alternatives.
When the FDA believes it is clear that a drug no longer has a place in treatment, it will ask the manufacturer to withdraw the drug voluntarily. Companies have agreed to withdraw the drug in all cases except one -- the case of an antidiabetic drug called phenformin, which was taken off the market in 1976 as an imminent hazard, despite the company's objections. If a company does not agree, the FDA can bring formal procedures to require withdrawal.
At first glance, one might assume that every time a drug comes off the market, it means that somewhere along the way somebody made a horrible mistake -- that the drug never should have been on the market in the first place. But FDA experts say that would not be correct. Most often, the withdrawal occurs because of adverse effects that were not seen prior to marketing. Sometimes, there was no clue at all. In other cases, one can see hints of the problem in retrospect, but not the serious events that eventually led to the withdrawal.
Many complex factors go into making judgments about benefits and risks, and into ultimately deciding whether a drug should be taken off the market. Here are some major issues, often overlapping, that weigh into the decision-making process...
Rare, unpredictable problems
Most drugs on the market are well-tolerated and their adverse effects are known. Known side effects cause more injuries and deaths than unrecognized side effects. But some problems happen so infrequently that they can't be seen or predicted before a drug gets on the market. Serious drug-induced liver disease, for example, is the leading single reason drugs have been pulled from the market. But it is rare, occurring at a rate of 1 in 5,000 to 1 in 10,000 exposures or less. This will not show up in clinical trials, which will pick up relatively common problems.
"If we want reasonably rapid access to needed drugs, it's not practical to require that they be tested in 15,000 to 30,000 people, which is what you'd need to be reasonably sure you saw even one case that occurs at a rate of 1 in 5,000 to 10,000," according to Robert Temple, M.D., director of the FDA's office of medical policy. "And the case would need to be recognized as drug induced," he says.
So drugs are typically tested in several thousand subjects, allowing detection of relatively common serious adverse events, such as those affecting 1 in 1,000 people. This practical size of clinical trials means we can't know everything about a drug when it gets on the market. Rare events will only surface when the drug is used in larger numbers of people. Temple says, "Sometimes less severe events that are seen in trials can be used to predict the occurrence of rare, more serious events, but that is not always the case, and such predictions have considerable uncertainty."
The number of subjects in clinical trials is increasing in some areas of drug development, says Peter Honig, M.D., director of the FDA's office of drug safety. "But the numbers will never be large enough to eliminate the need for postmarketing surveillance." The FDA is working on ways to better predict rare events, especially those related to the liver and heart.
But some uncertainties will always be there, including the possibility of rare characteristics that make some people particularly susceptible to an adverse reaction.
More toxic than expected
There are also times when a drug's toxicity is known, but the drug turns out to be more toxic than the clinical trials suggested, which again may only be seen when the drug is used in larger numbers or in different ways.
Initially approved in 1997, Baycol (cerivastatin) was a member of a class of cholesterol lowering drugs known as "statins." Baycol and the other five drugs in its class -- Lipitor (atorvastatin), Lescol (fluvastatin), Mevacor (lovastatin), Pravachol (pravastatin), and Zocor (simvastatin) -- have all been associated with rare reports of rhabdomyolysis, a condition that causes marked breakdown of muscle cells and can sometimes lead to fatal kidney failure and other problems.
Knowing this about the statin drugs, the FDA made sure to look for the problem when deciding to approve Baycol. But the agency didn't find unusual risk associated with the drug at that time. "In its first few years, Baycol had a small market share," says Sandra Kweder, M.D., acting director of the FDA's office of review management. "But when FDA approved a higher dose of the drug after initial marketing, use of the drug grew and we could see clearly that Baycol caused the problem more frequently than the other drugs in its class."
Problems with Baycol were reported most frequently when it was used at higher doses, when used in elderly patients, and particularly, when used with another lipid-lowering drug called Lopid (gemfibrozil). Baycol was voluntarily withdrawn in the summer of 2001.
When safer options are available
When the FDA approved Seldane (terfenadine) in 1985, the drug became the first prescription antihistamine to relieve allergies without causing drowsiness -- a side effect that can cause accidents and injuries. A few years after approval, it was discovered that Seldane could cause fatal heart rhythm irregularities when it was used together with drugs that followed its elimination from the body, or in patients with liver disease.
Major efforts to warn against use in such patients were partly successful, but fatal rhythm abnormalities continued to be reported. According to Temple, removal of the drug was considered, but that would have left only one non-sedating antihistamine, so Seldane remained available.
"But the equation shifted when Allegra came on the market in 1997," Temple says. "Allegra provided exactly the same benefits of terfenadine but without the risk of the potentially fatal heart condition." So the new availability of Allegra (fexofenadine) weighed heavily in the decision to withdraw Seldane.
Like Seldane, a heart drug called Posicor (mibefradil) posed problems mainly when used with other drugs. Although Posicor itself did not have unusual toxicity, it was taken off the market in 1998 because of its interactions with at least 25 drugs. It markedly increased the blood levels of those drugs, leading to potentially fatal side effects of the other medications.
When Posicor was first marketed in 1997, its labeling warned of possible interactions with three drugs. Two more drugs were later added, but reports of interactions and resulting adverse reactions with even more drugs kept coming. There was concern over whether it was realistic to expect physicians to be able to use Posicor safely, given the many drugs it interacted with. In the absence of any special benefit of Posicor compared to other members of its class, such as effectiveness in people who don't respond to other treatments, the FDA concluded that the drug should be removed from the market.
When taken at a higher than recommended dose and when taken with other drugs, Hismanal (astemizole), another non-sedating antihistamine approved in 1988, posed risks similar to Seldane. The drug was withdrawn in 1999, as safer alternatives became available.
Beginning about 1990, many potentially harmful interactions between drugs and even between drugs and foods (such as grapefruit juice) were noted with Seldane and other drugs. The discovery led to greater attention by the FDA and drug manufacturers to such interactions before drugs are marketed, Temple says. This represents a significant enhancement of safety assessment.
The term "safe" also depends on whether a drug is used according to the labeling. This is why the FDA makes sure labeling and advertising for prescription drugs are accurate and balanced -- presenting both the benefits and the risks.
The major problem with Duract (bromfenac), a nonsteroidal anti-inflammatory drug, was that the directions were not followed. The pain drug was withdrawn in 1998 after liver failure occurred in patients who took the short-term treatment for pain for more than the 10 days recommended in the labeling. Clinical trials indicated that a higher incidence of elevated liver enzymes was associated with longer use. Duract's manufacturer, Wyeth-Ayerst Laboratories, Philadephia, added a new warning to the labeling and sent letters out to doctors, but reports of long-term use of the drug continued.
When other risk management options fail
The day you hear news about a drug coming off the market, it may appear to be a sudden, drastic step. But several other options to manage risks usually have been attempted before that point. The main risk management tools employed by the FDA are education through letters to health-care professionals (known within the FDA as "Dear Doctor" letters), and labeling changes, such as new warnings, sometimes boxed in black for emphasis. Also used are required Medication Guides, labeling specifically for patients that emphasizes significant risks and advises patients how to detect or avoid them.
In some cases, a drug is labeled as "second line," meaning it is to be used only in patients for whom other treatments fail. In other cases, a drug that is known to be dangerous is still made available under certain circumstances through what's known as restricted distribution.
Sometimes these risk management techniques are effective, and other times they aren't. "We have our anecdotes, but there is little systematic study on the effect of drug labeling changes on physician behavior," says Temple.
Labeling changes were a partial success with the allergy drug Seldane. Studies showed use of Seldane with inappropriate drugs declined almost 90 percent, but that left considerable exposure to the dangerous combinations, some of which could be lethal.
The label of the heartburn treatment Propulsid (cisapride) was changed several times in 1998. The FDA cosponsored a study to evaluate the effect of various regulatory actions, and found that the percentage of patients inappropriately exposed to the drug was unchanged.
"We know that the farther out we are from the initial approval, the less likely we are to change behavior," Woodcock says. "Once a prescribing pattern has been established, it's hard to change it."
Clearly, the more special care that is required, the more physicians must remember, and the more we need other safeguards like spotting dangerous combination uses at the pharmacy level, the more of a challenge risk management becomes. "We do consider whether we are being unreasonable in our expectations, but sometimes that can't be known beforehand," Temple says.
Currently, the FDA is involved with several drug safety initiatives, including revamping the drug labeling for physicians to create a highlights section, a relatively short section that will describe the most critical information. Better education is a high priority. "We're looking into better ways to educate the public and doctors about changes in risk information, and to get information out faster," says Honig.
But FDA experts say the agency can't do it alone. The FDA judges drug risks for a population, doctors judge risks for individual patients, and patients judge the risks they'll take based on personal values. Ultimately, drug safety requires involvement of all parts of the health system.
U.S. Food and Drug Administration
FDA Consumer magazine
Posted: April 2002