Decisions about drug safety and efficacy are far from easy. Tysabri, a multiple sclerosis (MS) drug that was voluntarily withdrawn from the market last year after the appearance of a previously unknown side effect, illustrates some of the conundrums that exist in drug treatment.In advance of the publication of three critical new studies on Tysabri in the current issue of the New England Journal of Medicine, a major news organization recently asked me, as a physician and former FDA official, whether I knew of examples of prescription drugs that have “efficacy but [also] serious safety issues.” That is the rule rather than the exception, I responded.Obvious examples include the antimetabolites used for traditional chemotherapy. Because these drugs are no more than poisons administered in a way intended to be more toxic to cancer cells than normal ones, it is not surprising that their side effects are often serious and even life-threatening. When I was a medical resident three decades ago, hospital gallows humor included referring to BCNU, an experimental cancer drug, as “Be seein' you.” (Approved in 1977, it is still widely used.)A more recent example is aldesleukin (also known as interleukin-2, or IL-2), a drug that has offered new hope to victims of kidney cancer and malignant melanoma. It is highly effective in a small proportion of patients but exhibits significant toxicity. The patient information booklet warns that those taking the drug “frequently experience serious, life-threatening, or fatal adverse events,” including dangerously low blood pressure and reduced organ perfusion, impaired function of infection-fighting white blood cells, disseminated infection, and autoimmune disease.Antibiotics are another class of wonder drugs that sometimes manifests significant toxicity. Chloramphenicol, a drug that is effective against a wide spectrum of bacterial infections, causes rare cases of fatal aplastic anemia, so it used only sparingly. The potent antibiotic gentamicin is often life-saving but can cause damage to the kidneys, nerves and ears. And significant numbers of patients are allergic to other important antibiotics, including the penicillins and cephalosporins. But let us return to Tysabri, only the sixth medication approved — and the first in several years — for the treatment of MS, a common and debilitating autoimmune disease that affects the central nervous system. Following impressive results of the drug's testing in clinical trials — the frequency of clinical relapses reduced by more than half — the FDA granted accelerated approval in 2004.However, by that time several thousand patients were being treated with Tysabri, three had contracted Progressive Multifocal Leukoencephalopathy (PML), a rare neurological disorder caused by a virus. (Because the drug suppresses certain components of the immune response regulators, clinicians and the product's developers had from the beginning been sensitive to the possibility of infections as a side effect.) Precipitously (some would say prematurely) the manufacturers of the drug voluntarily halted production and distribution and withdrew Tysabri from the market. MS patients and many neurologists were bitterly disappointed. The three clinical studies reported in the New England Journal of Medicine bolster our confidence about both the safety and efficacy of Tysabri. In a study of almost a thousand patients that compared Tysabri to placebo, the drug cut the rate of clinical relapses by 68 percent (from 0.75 to 0.24), reduced by 83 percent the number of new or expanding brain lesions found on MRI, and slowed the clinical progression of disease. (The other currently used drugs for MS lower the occurrence of acute relapses by roughly one third.) Similar results were obtained in a second trial which compared two-drug therapy with Tysabri plus interferon beta-1a to the interferon alone.Finally, a third study found no additional cases of PML in more than 3,000 patients (exposed to an average of 17.9 monthly doses) who had participated in clinical trials of Tysabri. The investigators concluded that the incidence of this serious side effect is approximately one in a thousand patients treated with the drug. However, it should be noted that all three of the original cases of PML occurred in patients treated with interferon beta or other immunosuppressive agents as well as Tysabri, so the risk might be significantly lower in patients treated with Tysabri alone. Many neurologists who are familiar with Tysabri in MS believe this will prove to be the case.The “safety” of a drug is a relative thing. Safety and efficacy, the two criteria required for marketing approval of a drug, are inextricably linked. The judgments of regulators (and practicing physicians) require a global and often difficult calculation of risk and benefits, including consideration of what alternative therapies are available. We are willing to tolerate greater uncertainty and more severe side effects for a potential cure for pancreatic cancer or AIDS, for example, than for a new drug that treats heartburn. When FDA grants marketing approval, the drug is deemed to be sufficiently safe and effective to be used for the conditions on the label. In light of previous studies and the new data published in the New England Journal of Medicine — to which the FDA should have had access months ago — the publicly available evidence certainly supports returning the drug to market, presumably with revised labeling, as quickly as possible. If that happens, physicians and patients will need to decide whether for a given individual the risk of the drug is acceptable in view of the possible benefits. Even in view of the just-published data, one eminent neurologist who has long and extensive experience with Tysabri remains cautious: “I myself would only use the drug in patients who are 'train wrecks,'” by which he meant patients who were severely symptomatic. He feels that it is only they who “can assume a 1/1000 risk of a fatal complication and for whom such a risk is reasonable. I know many in this category with MS, because of the nature of my practice. But I would not treat that hypothetical 20-year-[old] sophomore on campus with early MS.” I have far less clinical experience than he, but given that Tysabri reduces the number of new or expanding brain lesions as measured by MRI, and that it slows the progression of the disease, it may be that some of the hypothetical 20-year-old college sophomores with MS might opt to take the drug if it makes it possible for them to live more normal lives. When an FDA advisory committee convenes this week to make recommendations about Tysabri's fate, its members will likely confront a mishmash of testimony from patients, physicians and other experts, perhaps including even some mischief on the part of companies with competing products. As I learned during my own extensive service on advisory committees, there is astonishing variability in an individual's — even an expert individual's — interpretation of the data. The FDA has the final say. Although society has designated the FDA as a gatekeeper between drug companies and the marketplace, the agency must beware of excessive paternalism and strive to preserve the right of patients, in consultation with their physicians, to make informed individual decisions about the available therapeutic options. Inevitably, different people will have very different comfort levels about risks and benefits.If our society permits citizens to make risk-benefit decisions about whether or not to have elective surgical facelifts and stomach-stapling operations for weight control. Whether to buy the penis-expanders and weight-loss snake oil nostrums that are intensively advertised on TV, and whether to invest in certificates of deposit, equities or junk bonds. How can we prohibit patients from making life-and-death medical decisions about drugs that have been shown to be effective? The notion that FDA should “err on the side of safety” sounds like a tautology, but it is an affront to patients with incurable or poorly treatable diseases: for them, there is no safety in the status quo, and we only damage them further with paternalistic public policy that prevents individuals from exercising their own judgment. If FDA must err, it should be on the side of patients' freedom to choose.