In some ways, it is enigmatic that the association of valvular heart disease with fenfluramine did not occur or was unrecognized for more than two decades. In retrospect, similar problems were clearly identified with the serotonin antagonist methysergide, which was introduced as breakthrough preventive therapy for migraine in the early 1960s. In that era, it became widely recognized that inflammatory fibroplastic lesions in the heart and lungs occurred as a consequence of prolonged treatment with methysergide [1-3]. Among these complications was cardiac valvular regurgitation, originally reported by Graham in 1967 [1]. At this time, M-mode echocardiography was in its infancy, and use of two-dimensional and color Doppler echocardiography would not become widespread until more than a decade later. Of interest, because methysergide has recognized problems with fibrosis, the drug's current prescribing information states that "there must be a drug-free interval of 3-4 weeks after every 6-month course of treatment" [4]. In the past 10 years, similar reports of aortic and mitral regurgitation emerged after years of ergotamine use [5]. Because the chemical structures of serotonin, methysergide, and ergotamine are related and because the valve lesions in these three conditions are identical, a common mechanism for ergot alkaloid-associated heart disease and carcinoid valve disease was suggested.

Another striking feature of this diet-drug problem was the high prevalence of valvulopathy in patients who had received fenfluramine or dexfenfluramine; this prevalence was based on data from five separate echocardiographic surveys reported to the FDA in September 1997 [6]. This followed the original report from the Mayo Clinic published 1 month earlier [7] citing a relation between fenfluramine-phentermine and valvular disease in 24 patients. In the Mayo Clinic report, cardiac valves excised at surgery to correct severe aortic and mitral regurgitation in 3 of 5 patients showed fibroplastic encapsulation indistinguishable from the lesions previously reported with carcinoid heart disease as well as with methysergide and ergot alkaloids. The remaining two patients had mitral valve repairs with lesions on gross inspection that showed "glistening white, thickened and tethered" leaflets. The five FDA surveys found an almost 33% prevalence of valvular lesions in 284 patients, predominantly women younger than 50 years of age. Because more than 14 000 000 prescriptions for fenfluramine or dexfenfluramine had been written after 1995, this finding had astounding public health and economic implications.

Late in the summer of 1998, three more reports addressing the diet-drug heart problem emerged. Jick and coworkers [8] conducted a population-based, longitudinal, retrospective analysis of practices in the United Kingdom and found 11 new cases of clinically recognized valvular regurgitation among 9765 patients who received monotherapy (6 of 2371 patients taking fenfluramine, 5 of 6532 taking dexfenfluramine, 0 of 862 taking phentermine, and 0 of 9281 matched controls). The 5-year cumulative incidence of valvular regurgitation was 7.1 per 10 000 patients taking fenfluramine or dexfenfluramine for less than 4 months and 35 per 10 000 patients in those taking these medications for 4 or more months.

Another study evaluated 1072 patients by using echocardiography an average of 34 days after termination of a 2- to 3-month obesity trial. Patients had been randomly assigned to two different dexfenfluramine preparations or placebo. Using the echocardiographic criteria set forth in the previous FDA surveys, the authors found that medication was no more likely than placebo to produce valve damage [9]. These results contrast with the findings of investigators from the University of Minnesota [10], who performed a retrospective echocardiographic evaluation of 233 patients previously enrolled in obesity trials. Thirty-nine patients were given dexfenfluramine, 31 were given dexfenfluramine-phentermine, and 163 were given fenfluramine-phentermine for a mean duration of 4.9, 9.0, and 26.5 months, respectively. Matched controls who had never received diet drugs were similarly evaluated. The prevalence of valvular regurgitation was 22.7% among patients and 1.3% among controls. The odds ratios for valvular abnormalities were 12.7, 24.5, and 26.3 with use of dexfenfluramine, dexfenfluramine-phentermine, and fenfluramine-phentermine, respectively.

Given this background, the article by Wee and colleagues [11] in this issue presents valuable and sobering findings. These investigators performed echocardiography in 46 patients (74% women; mean age, 53 years) who had undergone echocardiography before the institution of therapy with fenfluramine or dexfenfluramine. The patients had been using fenfluramine or dexfenfluramine for a mean duration of 163 days. Of note, 8 of the 46 patients (17.4%) had baseline echocardiographic abnormalities (that is, before use of diet drugs) that met the criteria for valvulopathy. Only 2 of the 46 patients (4.3%) met the predefined criteria for new or worsening valvular disease; 1 had been receiving fenfluramine-phentermine for 307 days and the other had been receiving this drug combination for 250 days.

Can these widely discrepant findings, ranging from valvular involvement of less than 1% to more than 30%, be reconciled? They can, at least in part. First, the FDA surveys and the University of Minnesota study report point prevalences and inherently overestimate the association of drug with valvulopathy because a certain percentage of patients will have preexisting valve lesions, particularly regurgitation (which the Doppler echocardiographic technique is especially sensitive in detecting). Moreover, case–controls are no substitute for objective evidence of the status of cardiac valves before drug exposure, a fact that is nicely documented in the report by Wee and colleagues. Second, the duration of drug exposure varied widely in these reports. Previous experience with methysergide and results of the studies cited previously indicate that valvulopathy is related to treatment duration. Finally, the method of detection plays a crucial role: Echocardiography is far more sensitive than clinical examination in detecting valvular regurgitation. The issue may also be confounded by lesion regression after discontinuation of therapy with these diet drugs, which occurred in some patients who received methysergide and has been reported in one case of valvulopathy associated with fenfluramine-phentermine [11].

To return to the original puzzle, why was diet-drug-associated valvulopathy not recognized earlier? Changes in medical practice seem to have played a role as longer-term and widespread use of these diet drugs evolved in the 1990s [13]. Concurrently, physical examinations have become more superficial, and cardiac murmurs can be more difficult to perceive in obese persons. The cause of echocardiographically detected valvular lesions is more likely to be questioned now that rheumatic heart disease is less prevalent. Combination therapy may play an important role. Although fibrotic valve lesions have not been recognized or reported with phentermine as monotherapy, concurrent use of the drug with fenfluramine or dexfenfluramine may involve a heretofore unrecognized interaction that facilitates a fibrotic response affecting heart valves. Alternatively, phentermine may be an innocent by-stander that has come under scrutiny because it was so frequently used in conjunction with fenfluramine or dexfenfluramine. Further evaluation of large patient populations treated with these drugs singly and in combination for similar durations might shed further light on this subject.

In conclusion, diet-drug-associated cardiac valvular incompetence is a real problem that, to date, seems to have a distinctly lower occurrence than originally suggested in the FDA surveys. The interplay of case selection, sensitivity of Doppler echocardiography in detecting valvular regurgitation, and duration (and, possibly, combinations) of therapy have probably played important roles in the original estimates of the magnitude of the problem. Fortunately, the problem has been recognized and steps to address it have been taken [14]. Most important, the recall of fenfluramine and dexfenfluramine will help prevent serious valvular heart disease from becoming a more widespread problem.