Appetite Suppressants

APPETITE SUPPRESSANTS

In contrast to the experience with dofetilide, in which an adverse drug effect was identified during product development, the consequences of appetite suppressants were detected serendipitously. Phentermine was approved by the FDA for appetite suppression in 1959, fenfluramine in 1973, and dexfenfluramine in 1996. The former two were approved for short-term use, and all three drugs were approved for use as single agents. In the 1990s, prescription of fenfluramine in combi-nation with phentermine and for periods longer than a few weeks was widespread. From 1995 to 1997, 14 million prescriptions were written for fenfluramine or dexfenfluramine, exposing an estimated 1.2–4.7 million persons to these agents (MMWR, 1997).

In July 1997, physicians in Minnesota reported 24 women with valvular heart disease who had taken fenfluramine–phentermine for 2–28 months (Connolly et al., 1997). The women were identified during evaluation of conditions such as congestive heart failure, heart murmurs, or arrhythmias. Echocar-diographic features of the dysfunctional valves resem-bled those seen in carcinoid heart disease; in five, valvular incompetence was severe enough to require cardiac surgery. Because of the morphologic simi-larity to carcinoid valvular disease, which has been attributed to high circulating levels of serotonin, the authors hypothesized that the valvular damage seen with fenfluramine might be due to its promotion of serotonin release and inhibition of serotonin reuptake. Phentermine is a nonadrenergic agent which impedes pulmonary clearance of serotonin, and which might potentiate the effect or concentration of circulating serotonin.

The Minnesota report, in conjunction with an FDA public health advisory, rapidly spawned additional case reports (Cannistra, Davis and Bauman, 1997; Graham and Green et al., 1997; Kurz and Van Ermen, 1997). A trio of larger clinical studies, each of differ-ent design, was published in September 1998 in the New England Journal of Medicine.

In the first study, echocardiograms were performed on 257 of 295 participants in prior appetite suppres-sant studies at Hennepin County Medical Center and in gender, age and body mass index-matched controls (Khan et al., 1998). Study participants had taken fenfluramine 60–120 mg + phentermine 30 mg daily, dexfenfluramine 30 mg daily alone, or in combi-nation with phentermine 30 mg daily. Mean dura-tion of treatment was 20 5 ± 12 months. Echocardio-graphic aortic and mitral insufficiency was scored (none, trace, mild, moderate or severe) by two blinded readers; if they disagreed, the study was reviewed by a third reader. FDA criteria for valvular abnormal-ity were applied, that is, aortic valvular disease of mild or greater severity and mitral valvular disease of moderate or greater severity ( = 0 79 for corre-lation between readers). Valvular insufficiency was identified in 3/233 control subjects (1.3%) and 53/233 appetite suppressant consumers (22.7%). In multivari-ate analysis which included age, gender, body mass index, blood pressure, and diabetes as covariates, and control subjects as the reference group, the odds ratios (95% confidence intervals) for valvular abnor-mality were 12.7 (2.9–56.4), 24.5 (5.9–102.2), and 26.3 (7.9–87.1) for dexfenfluramine, dexfenfluramine + phentermine, and fenfluramine + phentermine use, respectively.

The  second  study  compared  patients  who  had been prescribed dexfenfluramine= 6532 , fenfluramine (n = 2371) , or phentermine (n = 862) with age, gender, and weight-matched controls in the UK General Practice Research Database (Jick et al., 1998). During follow-up of about 4 years, the database identified 22 new diagnoses of valvular abnormality. Eleven patients were excluded when other causes of valvular disease, such as rheumatic heart disease or mitral valve prolapse, were identified by medical record review. The remaining 11 subjects had been referred to cardiologists for recent symptom onset or new heart murmur. In eight patients, valvular insuf-ficiency was confirmed by echocardiography and in three by clinical examination. All 11 patients had been prescribed dexfenfluramine or fenfluramine, a cumu-lative incidence of 14.2 per 10 000 (95% confidence interval 7.8–26.2). No valvular abnormalities were identified in untreated subjects or those prescribed phentermine.

The third study performed echocardiograms on participants in a randomized, double-blind trial comparing dexfenfluramine (15 mg bd), sustained-release dexfenfluramine (30 mg daily) and placebo, which was ongoing at the time dexfenfluramine was withdrawn from the US market (Weissman et al., 1998). Echocardiograms were performed on 1072 of 1212 randomized participants and interpreted by blinded readers; mean exposure was 72 days. Using FDA criteria for valvular abnormality (aortic insuffi-ciency of mild or greater severity and mitral insuf-ficiency of moderate or greater severity), valvular disease was not significantly more prevalent in the combined dexfenfluramine groups compared with placebo. When any degree of valvular insufficiency was compared between the treatment groups, aortic = 0 03 and mitral insufficiency = 0 01 were more frequent in the combined dexfenfluramine groups compared with placebo.

In November 1998, the American College of Cardi-ology and American Heart Association (ACC/AHA) recommended evaluation of appetite suppressant users, including history and physical examination, with echocardiography in those with signs or symp-toms of valvular disease (Bonow et al., 1998). Subse-quent meta-analyses have tempered initial estimates of the frequency of valvular insufficiency associ-ated with appetite-suppressant use, supporting the ACC/AHA statement that routine echocardiography was not recommended for all of the millions of indi-viduals exposed.

Pooled data from six controlled cohort studies yielded a relative risk ratio of 2.32 for aortic insuf-ficiency (95% confidence interval 1.79–3.01) and 1.55 for mitral insufficiency (95% confidence interval 1.06–2.25) (Loke, Derry and Pritchart-Copley, 2002). A second analysis which included ten studies found a prevalence odds ratio of 2.2 (95% confidence interval 1.7–2.7) for aortic insufficiency meeting FDA crite-ria among individuals treated for at least 90 days with fenfluramine derivatives (Sachdev et al., 2002). The odds ratio for mitral insufficiency of moderate or greater severity was 1.6 (95% confidence interval 1.05–2.3).

The experience with appetite suppressants illustrates the role of fortuitous observation in identifying adverse drug effects and the potential for inaccuracy of early risk estimates due to methodologic weak-nesses.