The best magazine
The recent sudden withdrawal of Vioxx (rofecoxib; Merck & Co., Whitehouse Station, NJ) from the market owing to safety concerns has created considerable public interest. The following brief article was written primarily for a concerned lay audience, but I believe that my professional colleagues may also enjoy considering some of the clinical issues that have been raised.
A large number of people have arthritis and other serious aches and pains, and so there is a strong interest in developing more effective ways of treating these conditions. The most widely used types of pain killers have been aspirin-like drugs called nonsteroidal anti-inflammatory agents, which are available under a variety of brand names at drug stores and supermarkets. These treatments, however, can have a serious drawback: in susceptible people they cause irritation and even more severe damage to the linings of the stomach and small bowel, leading to problems such as ulcers and bleeding, which can sometimes require major surgery and, on occasion, even prove fatal.
For this reason, 5 years ago, physicians and their patients were excited when a new class of drugs called selective cyclo-oxygenase-2 (COX-2) inhibitors came to the market. These newer drugs, usually referred to as the coxibs, effectively reduced pain and inflammation without affecting the cyclo-oxygenase-1 system, which is responsible for protecting the lining of the gastrointestinal system. In clinical trials, they were shown to have pain benefits similar to those of the older drugs, but with the advantage of being significantly less likely to cause ulcers or other serious gastrointestinal complications.
This would have been a great tale of medical progress had it not been for the chance finding that Vioxx (rofecoxib), one of the two coxibs initially released, was linked to a higher rate of myocardial infarctions when compared with one of the older, aspirin-like drugs called naproxen during a large study that had been designed primarily to compare the safety of these drugs in the gastrointestinal system.
This surprising discovery led to some obvious questions. Since a similar study involving Vioxx's competitor, Celebrex (celecoxib; Pfizer, New York, NY), did not show evidence for an increase in heart attacks, could the Vioxx discovery simply be a matter of chance? Also, even if it were true, how could it be explained? The manufacturer of Vioxx, Merck & Co., argued that Vioxx was not responsible for increasing the probability of heart attacks, but rather that the comparison drug naproxen, which somewhat akin to the well-known actions of aspirin, might have important protective effects against heart attacks and strokes.
At the time of that controversy, I recall that I served as an occasional advisor to Pharmacia, then the manufacturer of Celebrex, who saw a possible competitive advantage in the allegations regarding Vioxx's safety. At that time, I felt that Merck's argument might be plausible, and I recommended that Pharmacia be cautious in pursuing this marketing approach. But, in the end, despite some later evidence that naproxen may, in fact, be superior to other drugs of its type in preventing heart attacks, questions regarding the potential dangers with Vioxx, when used in relatively high doses, continued to be raised.
Observations of outcomes in large numbers of patients being treated by one of the large health maintenance organizations or who were being covered by one of the Medicaid plans also seemed to indicate increased heart attacks and strokes with Vioxx. Of course, such retrospective findings can be biased and misleading, as Merck, again not unreasonably pointed out. The United States Food and Drug Administration (FDA) also expressed concern, making it clear that they did not necessarily buy into Merck's defense of Vioxx and even pointing out hypothetical reasons for why an agent such as Vioxx could increase the likelihood of the blood clots that are associated with heart attacks. Indeed, in 2002, the official labeling for Vioxx was changed to include a warning about the dangers of increased cardiovascular events.
The final blow came very recently in a study designed to test whether Vioxx might have the ability to prevent a certain type of bowel polyp. What made this experience more definitive was that Vioxx was compared with a placebo, so that it was no longer possible to claim that differences in heart attack rates (as was the case in the earlier comparison with naproxen) could be explained by heart-protective properties of the other drug. Now Vioxx was found to double the incidence of myocardial infarctions. Even though the actual number of events was relatively small and there was no evidence for an increase in death rates, for the first time it became apparent that the potential dangers of this drug could no longer be refuted. Merck rapidly withdrew their product from the market.
This experience with Vioxx highlights a dilemma that not infrequently confronts clinicians. How can we know whether the benefits of a drug clearly outweigh its risks? The first issue is one of quantifying. In the case of Vioxx, we can probably make reasonable estimates of how many lives could be saved and how many major surgeries or hospitalizations avoided by the gastrointestinal benefits of this drug. At the same time, we might be able to make an estimate, from the risk point of view, of how many excess heart attacks or strokes might be inadvertently produced. Even if we could come up with reasonable calculations, however, how could we equate major gastrointestinal events against major cardiovascular events and decide, overall, whether this treatment is beneficial or harmful? Ironically, just a matter of days before the precipitous withdrawal of Vioxx, a commentary in Lancet, by Dr. Eric J. Topol, prompted by the publication of data on a new coxib, continued to question the wide use of this drug class despite uncertainty regarding its relative benefits and risks.
Another good example of this problem, again pertaining to a product withdrawn from the market by its manufacturer, was with the diabetes drug troglitazone. This drug was effective in treating type 2 diabetes, a very common condition in the United States that increases risks, among other things, of heart disease, strokes, and kidney failure. Troglitazone, however, despite its clear value in treating diabetes, at times had toxic effects on the liver which, in some cases, were fatal. Clinical experts, however, seeing the value of troglitazone in treating diabetes, argued that this drug (provided, of course, that physicians carefully monitored liver tests in their patients) should continue to be available. Again, there was the difficult issue of comparing one set of estimated benefits with a totally different set of estimated adverse outcomes.
Trying to put numbers on benefits and risks is an imprecise and difficult art that depends partly on known facts but also, to a large extent, on assumptions and speculations. Even so, it is an exercise worth undertaking. Certainly, data on a drug's benefit:risk ratio would help clinicians make decisions, on a patient-by-patient basis, about whether the supposed benefits of a drug outweigh its dangers. It might be very valuable for the FDA, when evaluating new products, to require that experts make their best efforts to provide quantitative benefit and risk estimates, however approximate they might be, that could be useful in describing drugs and guiding their usage.
The experiences with Vioxx and troglitazone help underscore another issue, in particular the advantages of having more than one drug in a class. In fact, in withdrawing Vioxx, Merck made the point that since other drugs of its type were now on the market, they could remove their product without depriving patients of an acceptable alternative. Likewise, it became feasible to withdraw troglitazone from the market once newer competitors, which were less likely to cause liver problems, became available. There are other strong clinical and scientific reasons for encouraging the availability of choices within drug classes, but certainly those people who argue against so-called "me too drugs" because of what they regard as the high costs of developing new products should be aware of just how important it can be to have apparently similar but, at the same time, clearly different agents available.
The safety of drugs is a legitimate concern, not just for physicians and patients, but also for the pharmaceutical industry and obviously the FDA. One of the reasons commonly given for the relatively slow process of drug approval is that the FDA is so meticulous in studying safety issues and ensuring that clinical trial experience provides reasonable confidence that a newly available drug can be considered acceptably safe. Obviously, the Vioxx story reminds us that the process of safety evaluation does not end with a drug's approval for marketing, but might need to be continued for as long as the drug is used.
Even so, what is troublesome about the Vioxx story is that >20 million patients were exposed to this drug, generating literally millions of years of patient exposure. Despite this extraordinarily large experience, in the end it was the unexpected findings of a relatively small trial (not even based on the types of patients who typically take Vioxx) that finally drew attention to the specific safety issue. In a commentary in the New England Journal of Medicine, Dr. A. Garret FitzGerald argued that even small-scale studies of underlying mechanisms of drug action could have anticipated the clinical problems and brought the issue of Vioxx's safety to a head far sooner. As an expert in the field, Dr. FitzGerald makes a fair point, but drugs can cause problems in such unexpected areas that it is often difficult to know where to start looking.
Solutions to this difficult problem of drug safety have included the creation of registries where all, or at least a substantial number, of patients exposed to a product are followed systematically by clinicians for a prolonged period (after the drug has been marketed) to ensure that any untoward effects, even those that might not have been anticipated based on previous knowledge of the drug, can potentially be identified.
Another useful approach might be practice-based studies in which large numbers of prescribing physicians enroll patients into relatively simple clinical trial protocols. This type of approach not only helps to familiarize relevant members of the physician community with a drug that they are likely to be using but also provides information on what actually might be happening in the practice setting rather than in the somewhat atypical academic environment in which most clinical trials traditionally have been performed. Convincing a large number of interested doctors to systematically build a large base of experience creates a practical and powerful way of expanding our knowledge of both the benefits and risks of treatments.
This whole issue surrounding Vioxx, however, will help solidify the FDA's agenda for the foreseeable future. High on the list of ideas to be considered are methods for better measuring the benefits and risks of therapies, albeit an imprecise science, so that physicians and patients can make more informed therapeutic choices. Second, the FDA must maintain its willingness to actively consider newer or later generation drugs in the same classes as drugs already approved. Vioxx may be an important example of how competitive products can be essential in providing alternative treatment choices. Finally, more thought must be given to postmarketing surveillance of drug safety. It now seems appropriate that some form of active monitoring, rather than the passive spontaneous reporting on which we have traditionally depended, should be required throughout the life of new drugs.
Source: ...