Echocardiographic Examination of Women Previously Treated With Fenfluramine: Long-term Follow-up of a Randomized, Double-blind, Placebo-Controlled Trial

Background  Fenfluramine hydrochloride was withdrawn from the market in September 1997 after reports of heart valve abnormalities in patients who used it. The prevalence of echocardiographic abnormalities and the clinical cardiovascular status of patients who received fenfluramine monotherapy remains uncertain.

Methods  A long-term, follow-up evaluation was undertaken in subjects who were randomly assigned to receive either fenfluramine hydrochloride (60 mg daily) or placebo as part of a double-blind smoking cessation therapy study. Cardiovascular status was evaluated by echocardiography, medical history, and physical examination.

Results  From the group of 720 smokers who had originally participated in the smoking cessation therapy trial, 619 women were enrolled; data from 530 (276 in the fenfluramine group and 254 in the placebo group) were evaluable. No statistically significant differences were identified in the prevalence of aortic or mitral regurgitation by Food and Drug Administration criteria or by grade, aortic or mitral valve leaflet mobility restriction or thickening, elevated pulmonary artery systolic pressure, or abnormal left ventricular ejection fraction. No significant differences were demonstrated in cardiovascular status by physical examination, and no serious cardiac events were noted among fenfluramine-treated subjects.

Conclusion  There was no evidence of drug-related heart disease up to 4.9 years after anorexigen therapy in subjects who were randomly assigned to receive fenfluramine at the recommended dose for up to 3 months.

THE ANTIOBESITY agent fenfluramine hydrochloride (D,L-fenfluramine) was indicated as a short-term (a few weeks) adjunct as part of a regimen of weight reduction based on restriction of energy intake.¹ In preliminary smoking cessation studies, dexfenfluramine hydrochloride (D-fenfluramine) decreased food cravings, carbohydrate consumption, and withdrawal effects.^2-5 Therefore, a prospective, randomized, double-blind, placebo-controlled, single-center study (Seattle Smoking Cessation Therapy [SCT] Study) was initiated at Fred Hutchinson Cancer Research Center, Seattle, Wash, in 1993 to evaluate the role of fenfluramine in women attempting smoking cessation; the treatment phase was completed in 1994.

Fenfluramine and dexfenfluramine were voluntarily withdrawn from the market in September 1997 after observations of valvular abnormalities in patients taking anorexigens.⁶ These reports lacked pretreatment echocardiographic data and a control cohort, and predominantly involved patients who used phentermine hydrochloride in combination with fenfluramine. More recently, systematic and controlled clinical evaluations have been undertaken to determine the clinical significance of the initial findings. Several recent reports have indicated that short-term exposure to anorexigens (3 months or less) is not associated with an increased risk of cardiac valvular abnormalities,^7-10 with one recent report showing no increased prevalence of aortic or mitral regurgitation (AR and MR, respectively) as defined by the Food and Drug Administration (FDA)⁶ in patients who took the phentermine-fenfluramine combination for 6 months or less when compared with an untreated control group.¹¹ These and other echocardiographic studies have demonstrated an increased prevalence of aortic valve regurgitation that is generally mild¹² and not accompanied by serious cardiac events in patients who used anorexigens for longer than 3 months.¹⁰ The prevalence of clinically overt valvular disease is reported to be lower than in initial reports, even with more than 3 months of drug exposure ¹⁰ or 4 years of posttreatment follow-up.⁷

The present follow-up study is a single-center, retrospective cohort study designed to evaluate the prevalence of cardiovascular abnormalities in subjects who previously participated in the SCT study and had been randomized to treatment with either fenfluramine alone or placebo. The SCT study treatment phase was planned for a maximum of 3 months, and all treatments were completed by the end of 1994. Therefore, this evaluation permits as much as 4.9 years of posttreatment follow-up.

This study was conducted at Fred Hutchinson Cancer Research Center according to a prospectively designed clinical protocol and was in full compliance with all federal, state, and local regulations pertaining to human research and with the Good Clinical Practice Guidelines from the Fred Hutchinson Cancer Research Center. The study center had prior approval from the appropriate institutional review board.

Female smokers enrolled in the SCT study between September 1, 1993, and April 30, 1994, and all study treatments were completed by August 31, 1994. Women aged 18 to 65 years, of any race or ethnic background, who smoked at least 10 cigarettes per day were randomly assigned to receive either fenfluramine hydrochloride (60 mg daily) or matched placebo for up to 3 months, provided they met the following criteria: their weight was between 85% and 150% of ideal based on the Metropolitan Life Insurance Tables; they were not pregnant or nursing at the time of study entry or anticipating pregnancy during the study treatment phase; they had no history of cardiovascular disease, cancer, or other significant disease; they had no previous or current long-term use of medication known to interact with fenfluramine; and they had no history of drug or alcohol abuse.

Eligibility for the current trial was limited to subjects who had participated in the SCT study, had no history of carcinoid tumor, and had documented informed consent for both studies. If, during the 5 years since the SCT study, a subject had received treatment with serotoninergic migraine headache medications or excluded anorexigens (dexfenfluramine, phentermine, phentermine and fenfluramine in combination, mazindol, phendimetrazine tartrate, sibutramine hydrochloride, herbal phen-fen, or liquid phen-fen), she was not eligible to participate.

Attempts were made to contact all SCT study subjects for possible enrollment in this follow-up study. Introductory letters and screening telephone calls were used to prequalify potential subjects, and clinic visits were scheduled for subjects who were tentatively eligible and willing to participate. At the clinic visit, each participant provided a detailed medical history and received a complete physical examination with eMPHasis on the cardiovascular system. The medical history collected in this study included dates of onset for medical symptoms and diagnoses (particularly related to the cardiovascular history and examination) and confirmed that obtained in the SCT. All SCT participants were asked to participate; data from those who did not meet eligibility criteria for the follow-up study were analyzed separately (ineligible population).

Two-dimensional, M-mode, color-flow, pulsed, and continuous wave Doppler echocardiograms were performed by trained sonographers who were blinded to all aspects of subject history. Echocardiograms were obtained according to a specified imaging protocol¹³ using standardized echocardiographic equipment (Sonos 2000 Imaging Systems; Hewlett-Packard, Andover, Mass) and recorded on videotape. The tapes were forwarded to an established central core laboratory (Boston Medical Center, Boston, Mass) staffed by board-certified cardiologists (R.D., G.J.B., and L.A.M.) specializing in echocardiography who were blinded to subject group and medical history.

Measures of valvular function, including (1) AR, MR, tricuspid regurgitation (TR), and pulmonic regurgitation and (2) aortic, mitral, and tricuspid valve leaflet thickness and mobility, were evaluated. Aortic regurgitation was visually graded on the basis of modified Perry et al criteria¹⁴ as follows: none, no regurgitant color flow; trace, regurgitant jet diameter in the parasternal (or apical) long-axis view 5% or less of the outflow tract diameter; mild, jet diameter greater than 5% and less than 25% of the outflow tract diameter; moderate, jet diameter 25% or more and less than 47% of the outflow tract diameter; moderately severe, jet diameter 47% or more and less than 65% of outflow tract diameter; and severe, jet diameter 65% or more of the outflow tract diameter, usually associated with diastolic flow reversal in the abdominal aorta. Both MR and TR were graded on the basis of modified Helmcke et al criteria¹⁵ as follows: none, no regurgitant color flow; physiologic, nonsustained jet within 1 cm behind the annular plane with a maximal jet 5% or less of the atrial area; mild, sustained color flow jet and maximal jet area greater than 5% and less than 20% of atrial area; moderate, maximal jet area 20% or more and 40% or less of atrial area; and severe, maximal jet area greater than 40% or jet reaches the back of the atrium. Eccentric jets (mitral and tricuspid valves) were upgraded 1 grade if jet impingement on a chamber wall precluded development of the full jet area. For all valves, regurgitation was graded not evaluable if assignment of a grade was considered impossible because of technical difficulty. The AR and MR were considered present in accordance with FDA criteria,^6,16 that is, subjects were AR positive when mild or greater AR was present and MR positive when moderate or greater MR was present.

Valve leaflets were considered abnormally thickened if, on visual inspection, mitral leaflets were greater than 4 mm during diastole, or if echodensities were detected involving the tip, body, and/or base of aortic leaflets. Valve leaflet mobility was considered abnormal if morphological changes resulted in moderate or greater mobility impairment according to the following criteria for the aortic valve: normal, full excursion of leaflets to the margins of the aortic annulus; mildly impaired, minimal restriction with maximal cusp separation greater than 1.5 cm; moderately impaired, restriction with cusp separation of 1.0 to 1.5 cm; and severely impaired, restriction of all 3 leaflets with cusp separation less than 1.0 cm. Mitral valve mobility was graded by means of the following criteria: normal, full excursion; mildly impaired, slightly decreased mobility that does not impair excursion to the posterior wall by more than 50%; moderately impaired, impaired motion by more than 50% with some mobility of leaflet body maintained; and severely impaired, complete immobility. Pulmonary artery systolic pressure (PAP) was calculated by means of the modified Bernoulli equation: PAP = 4v² + RAP, where v was the peak systolic velocity of TR jet recorded by continuous wave Doppler and RAP (right atrial pressure) was assumed to be 10 mm Hg.¹⁷ Left ventricular and left atrial dimensions were measured according to American Society of Echocardiography convention.¹⁸ Left ventricular ejection fraction (LVEF) was visually estimated by integrating information from all views and was considered normal if greater than 0.50.

The number of echocardiographic readings performed was equally distributed among 3 cardiologists (R.D., G.J.B., and L.A.M.) at the core echocardiography laboratory. A second, independent, blinded reading of an echocardiogram was performed when AR or MR (FDA criteria), leaflet thickening, or restricted leaflet mobility was detected on the initial reading. Disagreements between readers were resolved by consensus readings.

Study end points included echocardiographic variables, cardiovascular signs on physical examination, and serious cardiovascular events. Primary echocardiographic end points included AR and MR by FDA criteria (see the "Subjects and Methods" section). Secondary echocardiographic end points included cardiac valve regurgitation by grade, and aortic and mitral valve leaflet mobility and thickness. A sample size of 250 subjects per group was anticipated, to provide 90% power to detect a difference between the 2 study groups of at least 7% (α = .05) with the use of a 2-tailed test of proportions, assuming a control group prevalence of 2%. Means and SDs were calculated for numerically continuous variables, and percentages were calculated for categorical and ordinal data. Demographic variables for the fenfluramine and placebo groups were compared with a 1-way analysis of variance for numerically continuous variables and the Fisher exact or χ² tests for categorical variables.

Prevalence of AR and MR (FDA criteria) were compared in the fenfluramine and placebo groups by means of the Fisher exact test. Regurgitation grade data were compared with the Kruskal-Wallis test. Valve leaflet mobility and thickening data were also compared with the Fisher exact test, and calculated data on pulmonary artery systolic pressure were compared with analysis of variance. Unadjusted relative risks with 95% confidence intervals were calculated for the prevalence of FDA-defined AR and MR, and abnormal thickening and mobility of aortic and mitral leaflets for the fenfluramine group vs the control group.

A series of multivariate logistic regression analyses were performed to identify predictors of valvular regurgitation by means of FDA criteria for AR and MR. Independent variables tested in the models were selected because of study hypotheses, clinical relevance with respect to cardiac valvular regurgitation, or bivariate statistical association with regurgitation. Variables tested were treatment group, compliance, age, body mass index (calculated as weight in kilograms divided by the square of height in meters), hypertension, other medical and cardiac history, previous echocardiogram, menopausal status, and previous medication use. Two-way interaction terms were considered among variables. The analyses were conducted first with both treatment groups and then for the fenfluramine-treated patients only. A .05 level of significance was used to determine both entry and removal of a term in the stepwise selection process. The Hosmer-Lemeshow goodness-of-fit test was used to assess the models.

The Cohen κ statistic and percentage exact agreement were calculated as a measure of interreader variability with the use of second readings performed to verify an abnormality and a 5% randomly selected sample of tapes.

Comparisons between treatment groups were considered statistically significant for P values less than .05, and all statistical tests were 2-tailed. All statistical analyses were performed with SAS statistical software, version 6.09 (SAS Institute Inc, Cary, NC).

Of the 720 women who participated in the SCT trial, 619 were enrolled in the present study. Of these 619, 89 were found to be ineligible, most because they had taken serotoninergic migraine medications, other prescription anorexigens, or herbal phen-fen that were prohibited by the study eligibility criteria (including control subjects who had exposure to fenfluramine). Of the 101 women who were not enrolled, 65 were unavailable for follow-up and no further information was obtainable for them, 31 declined participation, and 5 had died. As a result, 530 women are included in the evaluable population of study subjects. Study participant flow is described in Figure 1. Data presented herein describe the evaluable population unless otherwise specified.

Study participants were generally white (92%), in their fifth decade (48.8 ± 8.2 years of age), and overweight (mean body mass index, 26.5). In both treatment groups, slightly more than one third of subjects were overweight, defined as body mass index greater than 25 but less than 30, and an additional one fifth were obese, defined as BMI of 30 or more.¹⁹ Most participants (about 75%) were current smokers; this was expected, because they had all previously participated in the SCT study, which required that they smoke an average of more than 1 pack per day at the time of enrollment (D.J.B., unpublished data, 1994). Demographic data are presented in Table 1.

More than 4 years (range, 4.1-4.9 years) elapsed from the start of the original smoking cessation trial to echocardiographic assessment in the present study (4.4 years, mean and median in both treatment groups). Because specific fenfluramine dose and duration data were not available from the SCT, compliance was estimated from returned pill counts at the 3-month visit or subject-reported usage data if pill counts were unavailable. Compliance estimates were available for 93% of study subjects with evaluable data; mean compliance was comparable among the study groups (53.4% ± 36.6% in fenfluramine-treated subjects and 56.6% ± 36.5% in placebo-treated subjects).

Overall, no statistically significant difference was demonstrated in comparisons between fenfluramine-treated subjects and placebo-treated control subjects (evaluable and ineligible populations) for any echocardiographic variable measured. Echocardiogram quality was high; of the 617 echocardiograms performed on enrolled subjects, 97% were considered by the readers to be of fair quality or better and more than 97% of the 530 echocardiograms performed on subjects in the evaluable population were of sufficient quality to assess all echocardiographic end points.

At median posttreatment follow-up of 4.4 years, no statistically significant difference was demonstrated between study groups in the prevalence of left-sided cardiac valvular regurgitation assessed by FDA criteria (Table 2). Aortic regurgitation was present (ie, mild or greater severity) in 6.2% of the treated group and 4.3% of the control group (relative risk vs control, 1.42; 95% confidence interval, 0.68-2.98; P = .44), and mitral valve regurgitation was present (ie, moderate or greater severity) in 5.1% of treated subjects and in 4.7% of control subjects (relative risk vs control, 1.07; 95% confidence interval, 0.51-2.28; P>.99).

Similarly, there were no significant differences between study groups when cardiac valvular regurgitation was compared by grade (Table 2). There was a trend toward increased prevalence of AR in the fenfluramine group, but the difference vs the control group was not significant (P = .26). The MR and TR prevalence rates were very similar in the anorexigen-treated and placebo groups. One fenfluramine-treated subject had severe MR associated with mitral annular calcification. Tricuspid regurgitation of moderate or greater grade was present in 7.2% of fenfluramine-treated subjects and 5.1% of control subjects (P = .37). In addition, there was no statistically significant difference in the prevalence of pulmonic regurgitation in the fenfluramine vs the placebo group (P = .51).

Comparisons of AR and MR prevalence among participants in the fenfluramine group vs placebo by compliance category demonstrated no statistically significant differences (for AR, P = .44; for MR, P>.99; Table 3). The AR prevalence appeared to increase with increasing compliance (P = .04); however, there was a significant age-compliance interaction (P = .03), where compliance increased with increasing age. In a multivariate logistic regression analysis with compliance, age, and the interaction of age and compliance, there was no statistically significant association with age (P = .46) or compliance (P = .80; adjusted odds ratio, 1.25). No other factors entered the model.

Aortic and mitral valve leaflets were evaluated for thickening and mobility restriction in a high proportion (97% or greater) of subjects with evaluable data in both treatment groups; no significant differences between treatment groups were identified (Table 4). Aortic valve leaflets were thickened in approximately 8% of subjects in each treatment group, and leaflet mobility was normal in all aortic valves evaluated. Thickened mitral valve leaflets were observed in 5.8% of fenfluramine-treated subjects and 2.8% of those who received placebo (P = .09). Mitral leaflet mobility restriction (moderate or greater) was not statistically significantly different when the groups were compared (0.7% for fenfluramine, 0% for placebo; P = .50). When all grades of mobility impairment were considered, the fenfluramine and placebo groups were also not statistically different (P>.99).

Pulmonary artery systolic pressure could be estimated in 358 subjects (68%) with TR sufficient for measurement of peak jet velocity (see the "Subjects and Methods" section; Table 4). Mean estimated pulmonary artery systolic pressure was 30 mm Hg in both treatment groups and was 40 mm Hg or higher in 7 placebo-treated subjects and 5 fenfluramine-treated subjects.

No subject in either treatment group had an abnormal LVEF, ie, an LVEF less than 0.50, and comparisons of active treatment with placebo identified no significant differences in left ventricular or left atrial dimensions (Table 4).

Interreader agreement was evaluated by comparing valve regurgitation and morphology ratings made on 182 echocardiograms with second readings performed for abnormality verification and quality control purposes. The κ coefficients (and percentage exact agreement) for interreader concordance (using FDA criteria for AR and MR, and moderate or greater grade to define TR and pulmonic regurgitation) were as follows: AR, 0.78 (93%); MR, 0.79 (95%); TR, 0.48 (93%); and pulmonic regurgitation, 0.66 (99%). Interreader consistency was addressed by a consensus adjudication process.

No significant difference was demonstrated between treated and control groups for any cardiovascular physical finding, including third and fourth heart sounds, murmur, peripheral edema, jugular venous distention, rales, rhonchi, and wheezing (Table 5). Vital signs (systolic and diastolic blood pressures, pulse, and respiration rate) were similar in the 2 treatment groups, with the exception of systolic blood pressure, which was higher in the fenfluramine-treated subjects than in the placebo group (P = .03).

Serious cardiovascular events occurring after study treatment were rare and not statistically different between the fenfluramine and placebo groups. There were no reports of cardiac valve replacements among the study participants. A single case of congestive heart failure was reported in a fenfluramine-treated subject and a single case of endocarditis occurred in a subject who received placebo. One fenfluramine-treated subject experienced cardiac arrest related to pneumonia and adult respiratory distress syndrome. There were 5 known deaths among the smokers who originally participated in the SCT trial, all of which occurred from 1 to 3 years after their SCT study treatment was completed. Four deaths occurred in women in the fenfluramine group who were aged 49 to 57 years at death; 3 of these were related to cancer, with 1 case each of pancreatic carcinoma, acute myelogenous leukemia, and lung cancer. One non–cancer-related death occurred in a 56-year-old woman with end-stage chronic obstructive pulmonary disease. In addition, a 42-year-old placebo-treated subject died of disseminated non-Hodgkin lyMPHoma.

As a group (n = 89), the ineligible subjects (fenfluramine group, n = 35; placebo group, n = 54; Figure 1) were slightly younger (fenfluramine group, 45.3 ± 7.7 years; placebo group, 44.2 ± 6.9 years) and heavier (fenfluramine group, 74.9 ± 10.8 kg; placebo group, 76.6 ± 13.0 kg) than the women in the evaluable population (Table 1), but were otherwise similar. (Data for weight and age are given as mean ± SD.) In the ineligible population, the prevalence of AR (FDA criteria) was 2.9% in the fenfluramine group and 5.8% in the placebo group, and MR prevalence (FDA criteria) was 2.9% in the fenfluramine group and 1.9% in the placebo group. No statistically significant differences were noted between ineligible study groups for any echocardiographic variable.

We examined subjects previously given fenfluramine or matched placebo for up to 3 months to evaluate the prevalence of cardiac valvular abnormalities at a median of 4.4 years after treatment and found no statistically significant differences between the fenfluramine and placebo groups in any cardiac valvular measure evaluated. Study participants were predominantly white and in their 40s, and more than half were overweight or obese. In this investigation, aortic and mitral regurgitation according to FDA criteria were present among placebo-treated subjects in 4.3% and 4.7%, respectively, and among fenfluramine-treated subjects in 6.2% and 5.1%, respectively. No serious cardiac events, valve replacements, or endocarditis were reported in fenfluramine-treated subjects followed up for up to 4.9 years after drug exposure.

In this study we report a lower prevalence of valvular regurgitation among those in the fenfluramine group than did initial reports.^6,20 Although we observed numeric increases in the prevalence of AR and MR among the fenfluramine-treated subjects, in neither case did these differences approach statistical significance. To put our study in perspective, the prevalence of AR demonstrated in anorexigen-treated patients in published cohort studies is shown in Table 6. By study design, all of our subjects were treated for no more than 3 months, and this is in contrast to the longer duration in most of the other studies. The age ranges are similar. The prevalences of mild (4%), moderate (1.4%), and severe (0.7%) AR are not different from those in our control group and are contrasted in Table 6 with results of the other reported studies.

Recent reports of clinical investigations have concluded that treatment with fenfluramine-based anorexigens for 3 months or less is not associated with an increase in the prevalence of echocardiographic cardiac valvular abnormalities.^8,10 Physical findings of cardiovascular disease are rare among patients who used these drugs⁷ and not different from findings in untreated obese patients.¹⁰ However, unlike this study, the time elapsed between drug discontinuation and follow-up echocardiogram has been relatively short in the other investigations that reported echocardiographic data (median, 1-6.6 months).

The present clinical trial differs from previous investigations in several meaningful ways, including (1) the substantial length of follow-up after anorexigen discontinuation, (2) recruitment from a cohort of subjects who had previously been prospectively randomized to treatment or matched placebo, and (3) the examination of a population who took fenfluramine alone (rather than in combination therapy with phentermine). While on the market, most fenfluramine was used in combination with phentermine (Wyeth-Ayerst Research Division, Philadelphia, Pa, unpublished data, May 1997) after reports by Weintraub and colleagues^22,23 of comparable efficacy with an improved side effect profile when the agents were combined.

The present study enrolled subjects who had been treated in the context of a previous randomized placebo-controlled trial; therefore, duration of treatment was limited to 3 months or less and baseline echocardiographic data were not available for the subjects. Also, possibly because of the nature of the original trial (smoking cessation), study drug compliance was somewhat lower than average. However, there was no statistically significant relationship between drug compliance and the echocardiographic abnormalities observed.

It is now generally agreed that treatment with fenfluramine-based anorexigens for 3 months or less does not increase the prevalence of valvular abnormalities. However, concerns regarding long-term follow-up have not previously been addressed. This study provides additional information regarding cardiovascular status almost 5 years after drug discontinuation.

Accepted for publication November 8, 2000.

This study was supported by a grant from the Wyeth-Ayerst Research Division of Wyeth Laboratories, Philadelphia, Pa.

Presented at the Second Annual and Plenary Meeting of the Working Group on Echocardiography of the European Society of Cardiology, Trieste, Italy, December 10, 1998.

We acknowledge the contributions of the following people: Fred Hutchinson Cancer Research Center: Margaret M. Hanrahan, MN (clinical services and coordination), and Diane Powers, MA (clinical study management). Boston Medical Center (echocardiography core laboratory): Bruce Talbot Joziatis, RDCS, Quirino Orlandi, MD, and Patricia Ray, RDCS (sonography). Wyeth-Ayerst Research: M. Linda Clarke, RN, Anita B. Hudak, Denise Mestichelli, and Karen P. Taylor (trial management); Harvey Kushner, PhD (statistical support); Stacy L. Hankin and Madelyn R. McGee (administrative assistance); and Laura L. Snyder (writing assistance).

Corresponding author and reprints: Ravin Davidoff, MB, BCh, Section of Cardiology C8, Boston University Medical Center, 88 E Newton St, Boston, MA 02118-2393 (e-mail: ravin.davidoff@bmc.org).