Appetite suppressants and valvular heart disease

Three articles from the New England Journal of Medicine (September 10, 1998).



The serotoninergic appetite suppressant drug fenfluramine, its d-isomer dexfenfluramine (Redux) and the sympathomimetic drug phentermine have been widely used for the treatment of obesity. In particular, the combination of fenfluramine and phentermine (fen-phen) began to be widely used for longterm treatment after the publication of several articles in 1992. Dexfenfluramine, alone or in combination with phentermine, became very popular in the U.S. immediately after its introduction in April, 1996. Until July, 1997, the greatest concern with these agents was their potential for causing rare cases of pulmonary hypertension.

In July, 1997, the FDA issued a health advisory concerning the potential of fen-phen for causing valvular heart disease (see article summarized here). A case series of 24 patients was described in the NEJM (summarized here) in August, 1997 and the three drugs were subsequently withdrawn from the market. Since the morphology of some of the valvular lesions that were seen at surgery resembled valve damage seen with the carcinoid syndrome, which is postulated to be due to high serotonin levels, it has been speculated that the appetite suppressants may cause heart valve damage through a serotonin-related mechanism. The risk for valvular lesions associated with these drugs has been estimated to be as high as 20 to 30 percent, but many feel that this greatly overestimates the risk, particularly for short-term users.

The three important studies summarized here were undertaken to shed light on the actual risk of valvular heart disease associated with exposure to fenfluramine, dexfenfluramine and phentermine.

The prevalence of cardiac valvular insufficiency assessed by transthoracic echocardiography in obese patients treated with appetite-suppressant drugs

Authors Khan M, Herzog C, St Peter J, Hartley G, Madlon-Kay R, Dick C, Asinger R, Vessey J. 
Source New England Journal of Medicine. 339:713-8. September 10, 1998. 
Institutions Multi-institutional in Minneapolis, Minnessota, USA. 
Support National Institutes of Health; Centers for Disease Control and Prevention.



This study was an echocardiographic study of patients who had previously taken appetite suppressants and matched controls. 

Patients had all previously participated in one of three studies of appetite suppressant medications. These studies utilized fen-phen (fenfluramine 60-120 mg daily plus phentermine 30 mg daily), or dexfenfluramine (30 mg daily) or dexfenfluramine (30 mg) with optional phentermine (30 mg). Patients were at least 30% above ideal body weight and were without overt cardiovascular disease. 

Controls were recruited by advertisement and were matched for age, sex, height and body-mass-index. 

Echocardiograms were interpreted by at least two observers. Valvular regurgitation was graded -- aortic regurgitation was judged significant if it was at least mild, mitral regurgitation if it was at least moderate. 



Of the 295 patients enrolled in the three original studies, 257 agreed to participate in this study. 6 were excluded because of protocol violations and another 18 could not be matched to controls, leaving 233 matched pairs of patients and controls.

Patients (and controls) were predominantly women (87%), with a mean age of 45, mean weight of 247 lbs and mean BMI of 40.5.

Significant valvular regurgitation (as defined above) was present in 1.3% of controls and in 22.7% of patients. Among patients, regurgitation was present in 12.8% of the 39 patients who had been treated with dexfenfluramine alone for a mean of 4.9 months, in 22.6% of the 31 who had been treated with dexfenfluramine-phentermine for 9 months and in 25.2% of the 163 who had been treated with fenfluramine-phentermine for a mean of 26.5 months.


A population-based study of appetite-suppressant drugs and the risk of cardiac-valve regurgitation

Authors Jick H, Vasilakis C, Weinrauch L, Meier C, Jick S, Derby L. 
Source New England Journal of Medicine. 339:719-24. September 10, 1998. 
Institutions Boston University Medical Center; Harvard Medical School. 
Support Food and Drug Administration; indirect support by multiple pharmaceutical companies.



This was a case-control study, utilizing a computerized system from the UK, the General Practice Research Database, which contains clinical information on over 4 million patients in private practices.

Subjects were identified from the database who were under 70 years of age, without overt cardiovascular disease and who had been given at least one prescription for an appetite suppressant (fenfluramine, dexfenfluramine, phentermine) after January 1, 1988.  Controls were selected from the same database and matched for sex, age, weight and medical practice.

The incidence of new valvular heart disease, occurring after the date of the first appetite suppressant prescription (or the same date for matched controls), was compared between patients and controls.  Valvular heart disease was identified by diagnostic codes for valve disease entered in the database and by reviewing any available echocardiograms.



9765 patients were identified who had been prescribed appetite suppressants. Of these, 1001 had taken more than one agent but were assigned to one of the three drugs for the purpose of analysis (based on quantity prescribed or, if the quantities were equal, which one had been taken last).  In this fashion, 6532 were classified as dexfenfluramine users, 2371 as fenfluramine users and 862 as phentermine users. A total of 9281 matched control subjects were identified from the database.

Among the 19,046 patients and controls, 22 patients with apparently incident cardiac valve abnormality were found. After record review, 11 of these were excluded because they had evidence of prior valvulopathy, leaving 11 with newly incident valve abnormalities.

These 11 had been referred for cardiac evaluation because of symptoms (8) or a murmur (3). The valvulopathy was confirmed by echocardiography in 8 out of the 11 patients. Two of the eleven had only mitral regurgitation, the remainder had only aortic regurgitation (6) or combined aortic and mitral regurgitation (3).

  • 5 of the cases were from the 6532 dexfenfluramine users. 2 of these were from 5086 who had used the drug for 1-3 months; 3 from 1446 who had used it for more than 4 months.
  • 6 were from the 2371 fenfluramine users. 2 of these were from 1831 who had used the drug for 1-3 months; 4 from 540 who had used it for more than 4 months.
  • none were from either the 862 phentermine users or the 9281 controls.
The cumulative incidence of valvular disease was 7.1 per 10,000 subjects among those who had taken dexfenfluramine or fenfluramine for one to three months and 35 per 10,000 subjects among those who had taken dexfenfluramine or fenfluramine for four or more months.

An assessment of heart-valve abnormalities in obese patients taking dexfenfluramine, sustained-release dexfenfluramine, or placebo

Authors Weissman N, Tighe J, Gottdeiner J, Gwynne J, for the Sustained-Release Dexfenfluramine Study Group. 
Source New England Journal of Medicine. 339:725-32. September 10, 1998. 
Institutions Georgetown University Medical Center; Wyeth-Ayerst Research.
Support Wyeth-Ayerst Laboratories.



This study was initially designed as a randomized, double-blind trial comparing a sustained-release formulation of dexfenfluramine (30 mg daily), conventional dexfenfluramine (15 mg twice daily) and placebo. The drugs were to be administered for 16 weeks, but the trial was discontinued early when dexfenfluramine was withdrawn from the market. The protocol was then modified to include ehocardiographic evaluation of all subjects.

Echocardiograms were performed and evaluated for valvular morphology and regurgitation. When these were judged to be abnormal, they were re-evaluated by a second echocardiographer.


1212 patients were randomized, and 1072 underwent echocardiography, a median of 33-34 days after drug discontinuation. Patients were predominantly white women, with an average age of 45 and an average body mass index of 38. The average duration of treatment was 71-72 days.

Valve regurgitation

Using the FDA criteria for significant valvular regurgitation (at least mild aortic regurgitation and at least moderate mitral regurgitation), and combining the two treatment groups (sustained release and regular dexfenfluramine), the prevalence of significant aortic regurgitation was 5.4% in dexfenfluramine-treated patients vs. 3.6% in those who had received placebo (p=0.27, NS). Similarly, the prevalence of significant mitral regurgitation was 1.8% vs. 1.2% (p=0.60) and of significant mitral or aortic regurgitation was 6.9% vs. 4.5% (p=0.16).

When all degrees of valvular regurgitation were considered, including trace aortic regurgitation and  physiologic or mild mitral regurgitation, the differences did reach statistical significance. Aortic regurgitation was present in 17% of the treatment group vs. 11.8% of the placebo group (p=0.03). Mitral regurgitation of any severity was present in 61.4% of the treatment group vs. 54.4% of the placebo group (p=0.01).  Tricuspid regurgitation was not correlated with dexfenfluramine use.

Valve morphology

There was no significant difference between the groups in aortic or mitral valve thickening, or in aortic leaflet mobility. The posterior mitral leaflet demonstrated less mobility in the combined dexfenfluramine groups than in the placebo group.



Although all three of the studies found some increase in valvular regurgitation among patients who had been treated with appetite suppressants, the degree of risk varied greatly. This variation is attributable to differences in the design of the studies and in the length of exposure to the drugs.

The greatest incidence of valvular regurgitation attributable to appetite suppressant use (about 20%) was found in the first study by Khan et al.. This study examined patients who had had a relatively long exposure to the drugs (about 20 months on average). Patients all had echocardiograms, and mild aortic regurgitation was considered significant. Thus, the high incidence of valvular abnormalities caused by the drugs is in part attributable to the long duration of exposure and to the fact that some lesions which may not be clinically significant were included.

In the study by Jick et al., based on a review of a clinical database from the U.K., the incidence of attributable valvulopathy was only 0.35% among those who had used the drugs for over 4 months. However, these patients came to be diagnosed because of symptoms or the appearance of a new murmur, and thus represent clinically more significant disease. Furthermore, the average duration of drug use in this study was substantially less than 20 months.

In the study by Weissman et al., all patients underwent echocardiography. The increase in detectable valvular regurgitation did not reach statistical significance unless even the mildest degrees of leakage were included. However, these patients only took the drugs for less than 3 months, on average. 

In an accompanying editorial, Dr. Richard Devereux points out that these three studies are important because they confirm the association between the drugs and valvular heart disease and exclude obesity as a confounding explanation. He also notes that the studies are reassuring for those patients who have taken these drugs for only a few months, since the incidence of valvulopathy was quite low in this group.

Unfortunately, the three studies do not help tease out the contribution of phentermine to the development of valvular lesions. Although phentermine does not act by a serotoninergic mechanism and was not associated with any valvulopathies in the second study (Jick et al), the number of phentermine users in this study was too small to draw any clear conclusions. Since much of the published data on valvular damage is from fen-phen users, phentermine is likely to remain "guilty by association", at least until data to the contrary emerges.

November 15, 1998 


References related to this article from the NLM's PubMed database. 

Reader Comments

February 22, 1999

Letters to the editor about these articles, from the NEJM.

Submit a comment

Journal Club homepage

Site Meter