Anorectics on Trial: A Half Century of Federal Regulation of Prescription Appetite Suppressants

The First FDA-Approved Obesity Drugs

In November 1943, Abbott Laboratories of Abbott Park, Illinois, submitted a New Drug Application (NDA) for desoxyephedrine (Desoxyn) to the FDA's Drug Division. The company was seeking approval of their amphetamine for the treatment of narcolepsy, mild depression, postencephalitic Parkinson syndrome, chronic alcoholism, cerebral arteriosclerosis, and hay fever (3). The data submitted to support the drug's approval included review articles from academia, case reports from clinicians, and a 3-page testimonial from a patient with narcolepsy. Desoxyn was approved for all the proposed indications in December 1943 (4).

One year later, the director of the FDA's Drug Division authorized the approval of Hydrin (Endo Products, Garden City, New York), another desoxyephedrine compound. The indications for use of Hydrin were similar to those for Desoxyn, with 1 notable exception: Hydrin was approved as an adjunct in the treatment of obesity. No sooner, however, had the FDA approved Hydrin for obesity than those involved questioned the wisdom of their action. “The use of desoxyephedrine in [obesity] is wholly irrational and exposes the patient unnecessarily to a potent drug,” read a January 1946 letter from the new acting medical director of the FDA's Drug Division to the manufacturer of Hydrin (5).

What caused the FDA's abrupt turn of opinion regarding the use of Hydrin to treat obesity is unclear. The most likely explanation is that the FDA's acting medical director also served as a consultant to the American Medical Association's (AMA) Council on Pharmacy and Chemistry—a highly influential group whose opinions on the therapeutic value of drugs shaped clinical practice. In its 1946 edition of New and Nonofficial Remedies, the Council “went on record as disapproving general recognition of claims for the use of amphetamine in the treatment of obesity” (6).

The FDA clearly mandated that companies seeking to secure an obesity indication for a desoxyephedrine compound would have to submit evidence of the drug's safety when specifically used to treat obesity under the direction of a physician. Good fortune for the companies soon came in the form of an article titled “The Obese Patient,” which reported that 110 obese patients treated with 2 mg of desoxyephedrine 3 times daily lost up to 24.5 kg without apparent elevations in blood pressure or evidence of addiction (7). These data, along with the AMA Council's tepid endorsement of amphetamines for the management of obesity, led the FDA to approve Desoxyn and Hydrin “as adjuncts to the dietary management of obesity” in 1947 (8, 9).

In an attempt to develop drugs that would retain the anorectic effect of amphetamines without their stimulatory properties or the potential for addiction, industry chemists tinkered with the parent amphetamine molecule and synthesized 5 compounds known as the amphetamine congeners (Table). Applications for all of these drugs were submitted to the FDA soon after desoxyephedrine's approval, and all sought a single indication: the treatment of obesity. Reviewers from the FDA found no evidence that the amphetamine congeners were unsafe (particularly in comparison with the amphetamines), and by 1960 all 5 drugs were approved as adjuncts in the management of obesity.

The Kefauver-Harris Amendments and the Drug Efficacy Study

In 1962, Congress passed the Kefauver-Harris amendments to the Food, Drug, and Cosmetic Act (2). This legislation mandated, among other things, that new drug applications contain substantial evidence of a drug's efficacy from “adequate and well-controlled investigations.” While this law had immediate implications for new drugs, compounds approved between 1938 and 1962 were not covered by this legislation. The commissioner of the FDA therefore made the decision to retroactively apply the standard of “substantial evidence of effectiveness” to drugs approved before 1962. To assist in this ambitious endeavor, which became known as the Drug Efficacy Study, the FDA called on the National Research Council of the National Academy of Sciences (10). In 1966, 27 panels of academics began their reviews of the available data on the efficacy of nearly 3000 drug preparations. To account for the evolving definition of substantial evidence of efficacy and for the variation in the quantity and quality of the available data, the advisory panels categorized drugs as “effective,” “effective but” (drugs for which there was evidence of efficacy but more efficacious or safer drugs were available), “probably effective,” “possibly effective,” “ineffective,” or “ineffective as a fixed combination” (11).

The task of assessing the weight-loss efficacy of the amphetamines and the amphetamine congeners fell to the Psychiatric Drug Panel. After 3 years of review, the Panel concluded that as treatments for obesity, the amphetamines were “possibly effective” and the amphetamine congeners were “effective but” (12, 13). Reasons given for considering these drugs less than effective included the short duration of the studies and the lack of evidence showing that the drugs altered the natural history of obesity.

The FDA considered the Psychiatric Drug Panel's findings and ultimately agreed that the available data did not support an “effective” classification for the amphetamines or the amphetamine congeners. Thus, in 1970, all manufacturers of the anorectics were given 6 months (later extended to 12 months) to obtain and submit substantial evidence of their drug's effectiveness from adequate and well-controlled clinical studies. Absent definitive efficacy data, the FDA threatened to revoke the obesity indications or to remove the drugs from the market (14).

As the companies began studies to demonstrate their drugs' efficacy, the FDA began its search for criteria to define this yet-to-be-demonstrated efficacy.

The FDA's Struggle To Define the Efficacy of Weight-Loss Drugs

For guidance on how to define the efficacy of the anorectics, the FDA turned first to Thaddeus E. Prout, an endocrinologist and associate professor of medicine at Johns Hopkins University (15). Regulators met with Prout, 8 other academics, and the medical director from Abbott Laboratories to discuss the development of the anorectics in general and the definition of their efficacy in particular (16). Prout's working group reached many conclusions, the most influential of which was a recommendation that the efficacy of the anorectics be defined as statistical superiority of drug versus placebo. In other words, as long as the average weight lost by patients taking the drug was greater than the average amount lost by those taking placebo and the difference was statistically significant (that is, P < 0.05), the drug should be considered effective. Prout's group declined (or was unable) to define clinically significant weight loss.

Still seeking to determine how much weight must be lost to reap clinical benefit, the FDA next turned to one of its advisory committees for help. Instead of offering an answer to this question, however, the committee dodged the issue by referring to Prout's recommendation that efficacy be defined as statistical superiority of drug to placebo (17).

Why would no one define clinically significant weight loss? Perhaps the prevailing mindset and available evidence didn't lend themselves to the task. By the early 1970s, much data existed to link obesity with excess mortality (18), type 2 diabetes (19), elevated serum cholesterol levels (20), and hypertension (21). Many years would still need to pass, however, before large end point trials (such as the Lipid Research Clinics) would provide the medical community with evidence that drugs, through their effects on biomarkers such as serum cholesterol or blood pressure, could substantially alter the clinical course of a chronic disease (22). Without the availability of such data, how would one even begin to define clinically significant weight loss? Nonetheless, concluding that an obesity drug was effective if it caused statistically significantly more weight loss than placebo would have been attractive to drug regulators for 2 reasons: It was objective, and it left no room for argument.

The Amphetamine Anorectic Drug Project and the Balance of Benefits versus Risks

In June of 1972, the FDA publicly discussed the results of their Amphetamine Anorectic Drug Project—a crude meta-analysis of weight-loss data from more than 10 000 patients who had participated in 200 weight-loss studies involving all of the major amphetamine and amphetamine congeners. Among these drugs was fenfluramine, which had been under regulatory review since 1967 (23). The studies, which had been conducted in response to the FDA's 1970 request for “substantial evidence” of the anorectics' efficacy, ranged in duration from 3 weeks to 6 months, although few patients were exposed to a drug for more than 12 weeks.

The meta-analysis indicated that obese patients treated with active drug lost “a fraction of a pound more a week” than those treated with placebo, a “trivial” yet statistically significant difference (24). The results from the Amphetamine Anorectic Drug Project led the FDA to officially declare that the amphetamines and the amphetamine congeners were effective for the treatment of obesity (25, 26).

Yet efficacy was only half of the story. After passage of the Kefauver-Harris amendments in 1962, drug regulation was governed by evaluations of benefit versus risk. For more than a decade, the major perceived risk for the anorectics, as emphasized in a series of high-profile congressional hearings, was addiction (27). Although the amphetamines clearly posed a risk for addiction, the addictive potential of the amphetamine congeners was not as well studied and remained open to debate. Nevertheless, on the basis of structural similarities and some anecdotal evidence, many believed that the amphetamine congeners also posed a risk for abuse and addiction.

The FDA discussed many options to deal with its concerns regarding the balance of benefits and risks for the anorectics, including removing the obesity indication, removing the drugs from the market, requiring additional studies of efficacy and safety, or imposing greater restrictions on production and distribution. In the end, a compromise was reached. All of the amphetamine and amphetamine congeners would remain on the market and keep their obesity indication, but all would be restricted to short-term use (a few weeks), and all would be prominently labeled to warn against the risk for addiction (26). Although use of the anorectics for only a few weeks theoretically reduced the risk for addiction, this restriction also eliminated the potential for clinical benefit vis-à-vis sustained weight loss with extended use of the drugs. Regardless of whether the new labeling restrictions were right or wrong, they marginalized the anorectics and contributed to the eventual decline in their use.

A Transition to Long-Term Treatment of Obesity

The decrease in the use of anorectics during the 1970s and 1980s came to an abrupt end when prescription rates for phentermine and fenfluramine skyrocketed in the mid-1990s (28). This revival was stimulated by the juxtaposition of a dramatic increase in the prevalence of obesity with publication of a single study in which 121 obese individuals received treatment with placebo or phentermine plus fenfluramine for up to 4 years (29, 30). Although less than one third of the patients completed this study (and most regained weight during its latter stages), the findings, published in 1992, were cast in a very favorable light by the lay press, fueling the phen-fen craze (31).

In addition to popularizing off-label use of 2 aging anorectics, the phen-fen studies presaged a transition from short-term to long-term drug treatment of obesity. The first drug to garner FDA approval for the long-term treatment of obesity was dexfenfluramine, an isomer of fenfluramine.

When an FDA advisory committee met in September 1995 to discuss the dexfenfluramine application, the agency finally had working guidelines for the development of obesity drugs. Recommendations stipulated that at least 1500 obese patients be studied for 1 year under placebo-controlled conditions and that 200 to 500 of these patients continue drug treatment for a second year in an open-label manner. The 2 criteria used to define an obesity drug as effective were that a mean difference in weight loss of at least 5% between the drug- and placebo-treated patients after 1 year was noted or a greater proportion of patients lost at least 5% of their weight after 1 year of treatment with the drug than with the placebo.

These efficacy criteria were based on 2 tiers of evidence linked by speculation. First, data indicated that as little as a 5% reduction in weight improved blood pressure, serum cholesterol levels, and blood glucose control (32). Second, evidence from clinical trials of some medications demonstrated that modest drug-induced decreases in biomarkers, such as blood pressure and cholesterol, substantially reduced cardiovascular events and, in some cases, death (33-36). Thus, one only needed to take a small leap of faith, argued some researchers, to expect that modest drug-associated reductions in weight would reduce the risk for irreversible morbidity and mortality.

At face value, the clinical evidence from the dexfenfluramine application supported the drug's efficacy. In a 1-year trial involving 822 obese patients, 64% of patients treated with dexfenfluramine lost at least 5% of their baseline weight, compared with 43% of placebo-treated patients (37). The most common treatment-emergent adverse events in this trial were drowsiness, dry mouth, and diarrhea, problems certainly not worthy of serious concern.

What did concern some regulators and members of the advisory committee were animal data linking dexfenfluramine to neurotoxicity and epidemiologic data linking dexfenfluramine (particularly when used for more than 3 months) to an increased risk for primary pulmonary hypertension, an invariably fatal disease (38). Proponents of dexfenfluramine argued that the finding of neurotoxicity in preclinical models was not clinically relevant because the animals received extremely high doses of the drug. In support of this position, dexfenfluramine had been widely used in Europe for years, and no evidence of serious neurologic damage had come to light.

To put the primary pulmonary hypertension risk into perspective, an academic consultant remarked that the risk for fatal anaphylaxis from penicillin was much higher. Furthermore, echoing the rationale supporting the FDA's efficacy criteria for obesity drugs, dexfenfluramine had been shown to induce a 5% weight reduction and subsequent improvement in cardiovascular risk factors in a significant proportion of patients treated for up to 1 year. By extrapolating from available evidence, clinicians could have expected these changes to reduce the risk for serious morbidity and even death. As a professor of pulmonary medicine later argued, “the risk of developing PPH [primary pulmonary hypertension] [from dexfenfluramine] is about 1000-fold less than the risk of dying from the complications of obesity” (39).

Similar arguments convinced 6 advisory committee members to vote in favor of approving dexfenfluramine. Five members, however, did not believe that the available data supported a favorable balance of benefit to risk and voted against approval. As the advisory committee's split vote made clear, the availability of long-term data on body weight did little to illuminate whether the benefits of dexfenfluramine outweighed its risk—no more than when regulators, on the basis of short-term data, had to address this question for the amphetamines and amphetamine congeners in the early 1970s. In both cases, the FDA ultimately concluded that the drugs' benefit-risk profiles were favorable when, and only when, the drugs were used in accordance with the approved labeling.

When dexfenfluramine was approved in 1996 for the long-term treatment of obesity, it was labeled only for patients who were at substantially increased risk for illness because of their weight. This risk was defined as either a body mass index greater than 30 kg/m² or a body mass index of at least 27 kg/m² in the presence of comorbid conditions, such as hypertension, diabetes, and hypercholesterolemia (40). To reduce needless long-term exposure (and therefore the risk for primary pulmonary hypertension), the labeling recommended that patients who did not lose at least 4 pounds during the first month of treatment should stop taking the drug because they were unlikely to achieve a 5% reduction in weight with continued treatment. Furthermore, the increased risk for primary pulmonary hypertension, particularly when the drug was taken for more than 3 months, was highlighted in the labeling in a large, boldface font—1 step removed from the most restrictive labeling, a black box warning.

Within a year of its approval, dexfenfluramine was being dispensed at a rate of 85 000 prescriptions per week (41). Within a year and a half of its approval, the drug was off the market, as was fenfluramine. Reports implicated the 2 drugs in a wave of unusual cases of left-sided valvular degeneration—a risk that no one saw coming, and to this day, one that eludes a biomechanistic explanation (42).

The void created by the withdrawal of dexfenfluramine in September 1997 was quickly filled with sibutramine. Like dexfenfluramine, sibutramine's regulatory path to approval involved intense debates over the balance of its benefits and risks. One of these debates played out in a 1996 advisory committee meeting in which regulators, their academic advisors, and sibutramine's manufacturer and its consultants discussed the drug's approvability (43).

Few disputed that sibutramine was an effective drug, at least as defined by the FDA's efficacy criteria. Following a year of treatment, approximately 60% of 320 obese patients treated with the drug lost at least 5% of their baseline weight; in comparison, about 30% of 160 placebo-treated patients achieved that goal. The point of contention, as regulators repeatedly emphasized, was what to make of the drug's tendency to increase blood pressure and pulse rate. In the preapproval trials, treatment with sibutramine was associated with mean increases in systolic and diastolic blood pressure of approximately 1 mm Hg to 3 mm Hg, respectively, and an average increase in heart rate of about 5 beats/min (44).

The company openly conceded that sibutramine, as a sympathomimetic, did have pressor effects. However, they claimed that the small average increase in blood pressure would be offset by the favorable changes in lipid levels that accompanied sibutramine-induced weight loss.

This concept of negating risk factors found a quantitative voice in a professor of epidemiology who spoke on behalf of the company (45). Using mathematical models of Framingham data, the speaker showed the advisory committee calculations of risk for coronary heart disease for various hypothetical clinical scenarios of sibutramine use. In a population of 40-year-old nondiabetic, nonsmoking women, for example, the increase in coronary heart disease risk associated with a 2-mm Hg increase in blood pressure caused by sibutramine would be offset by the reduction in risk associated with the reduction of 0.26 mmol/L (10 mg/dL) in total serum cholesterol level. An increase in serum high-density lipoprotein cholesterol level of 0.05 mmol/L (2 mg/dL) would also accompany a 5-kg sibutramine-induced reduction in body weight. Therefore, the overall 8-year risk for coronary heart disease in this population of patients would actually decrease by 11.0%.

Response to this line of reasoning was generally favorable, although several people thought the use of 0.26 mmol/L (10 mg/dL) as the standard decrease in total cholesterol level was generous given the inconsistent lipid changes observed in the sibutramine preapproval trials. Nevertheless, when the members of the advisory committee were asked whether they believed the benefits of sibutramine outweighed its risks, 4 voted yes and 5 voted no.

Lacking a clear mandate, regulators left the advisory committee meeting once again faced with the difficult task of making a regulatory decision based on a rough estimation of the long-term risk-benefit profile of an obesity drug. As long as sibutramine met accepted standards of efficacy and safety and the labeling accurately described the drug's potential benefits and risks (in particular, the need to monitor blood pressure and pulse), some regulators held that physicians, as learned intermediaries, were the appropriate final arbiters of whether the balance of benefits and risks for sibutramine was favorable for a given patient. Sibutramine was approved for the long-term treatment of obesity in November 1997, just weeks shy of the 50th anniversary of desoxyephedrine's approval for the treatment of obesity in 1947.

Emblematic of the polarization over anorectics, a consumer advocacy group derided the news of sibutramine's approval as a prelude to “another diet drug disaster,” whereas a seasoned academic hailed FDA's decision as “great news for dieters” (45, 46).

Conclusion

To be sure, polarization remains the legacy of the first half-century of the FDA's regulation of anorectics. Yet, 8 years since the approval of sibutramine, use of the drug remains steady at about 50 000 prescriptions a month, suggesting that the drug has found favor with some dieters; meanwhile, no evidence has surfaced to suggest that sibutramine has become “another diet drug disaster” (47).

This is not to say that some did not try to make that case. Following the deaths of 2 young women taking sibutramine in Italy in 2002, that country temporarily suspended the drug's marketing license (48). This news then triggered some of the drug's opponents to question whether sibutramine should remain on the U.S. market (49).

European drug regulators were quick to conclude, on the basis of an assessment of the 2 deaths in Italy as well as other safety data, that sibutramine's risk-benefit profile was still favorable and that the drug should remain on the European market (50). This exoneration, however, was accompanied by a proviso with worldwide regulatory ramifications: Abbott Laboratories, the manufacturer of sibutramine, would have to conduct a large trial to definitively examine the drug's risk-benefit profile in obese patients at risk for cardiovascular disease (51). The Sibutramine in Cardiovascular Outcomes (SCOUT) trial is underway and aims to study 9000 patients for up to 5 years. This will be the first trial to verify or refute the long-held assumption that drug-induced weight loss—in this case, with sibutramine—reduces the risk for fatal and nonfatal cardiovascular disease.

Not only is SCOUT a landmark study, it reminds us that there is no substitute for data from large, long-term controlled trials for making the most accurate assessment of a drug's risks and benefits. This fact will weigh heavily on the minds of FDA regulators as they, amid calls to reduce the size and scope of obesity drug registration trials, begin the process of updating the agency's Guidance for the Clinical Evaluation of Weight-Control Drugs(52).