International Journal of Obesity (1999) 23, 926±928 ß 1999 Stockton Press All rights reserved 0307±0565/99 $15.00 http://www.stockton-press.co.uk/ijo Dose-effect of fen¯uramine use on the severity of valvular heart disease among fen-phen patients with valvulopathy

RLi1*, MK Serdula2, DF Williamson3, BA Bowman2, DJ Graham4 and L Green5

1Epidemic Intelligence Service, Epidemiology Program Of®ce and Division of Nutrition and Physical Activity; 2Division of Nutrition and Physical Activity; 3Division of Diabetes Translation, National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, USA; 4Epidemiology Branch, USA; and 5Reports Evaluation Branch, Division of Pharmacovigilance and Epidemiology, US Food and Drug Administration, USA

OBJECTIVE: To determine whether the severity of valvulopathy was associated with the dosage of fen¯uramine taken by fen¯uramine-phentermine users with valvulopathy. DESIGN: Out of 105 suspected valvulopathy case reports received by the US Food and Drug Administration (FDA) among fen¯uramine-phentermine users, 74 patients meeting FDA case de®nition for valvulopathy were included in this study. Patients with severe valvulopathy were classi®ed as those either undergoing valve replacement surgery or having severe aortic or mitral regurgitation; all other patients were considered to have less severe valvulopathy. RESULTS: The proportion with severe valvulopathy increased from 20 ± 66% with increasing fen¯uramine dosage from 40 mg=dto60 mg=d. Compared with patients taking<40 mg=d fen¯uramine, patients taking 60 mg=d had an adjusted odds ratio of 9.2 (95% con®dence interval ˆ 2.1 ± 40.8) for severe valvulopathy. CONCLUSION: Compared to patients with less severe valvulopathy, those with severe valvulopathy were substan- tially more likely to have taken 60 mg=d fen¯uramine.

Keywords: dose ± effect; fen¯uramine-phentermine; severity of valvulopathy

Introduction provides information about the dose-effect. The pur- pose of this study was to determine whether the severity of valvular heart disease among fen-phen Fen¯uramine is a serotonin-based appetite-suppres- users was associated with the dosage of fen¯uramine. sant drug which is widely used in combination with phentermine, commonly called fen-phen. On 8 July 1997, the Mayo Clinic reported 24 cases of valvular heart disease in women treated with fen-phen,1 who Subjects and methods had no history of valvulopathy. Subsequently, the US Food and Drug Administration (FDA) issued a Public Because the striking feature of the case reports Health Advisory requesting that physicians report received by FDA was the consistent involvement of additional cases of valvulopathy through MED- left-sided cardiac valve lesions, FDA established its WATCH, a medical product reporting system estab- case de®nition as the existence of either mild or 2 lished by FDA. Based on results from the prevalence greater aortic regurgitation (AR), or moderate surveys, which consistently found a higher than or greater mitral regurgitation (MR), or both as expected prevalence of valvulopathy among persons determined by echocardiography. At the time when 3 exposed to fen¯uramine or dexfen¯uramine, FDA fen¯uramine and dexfen¯uramine were voluntarily announced the voluntary withdrawal of fen¯uramine withdrawn from the market, FDA had received 105 and dexfen¯uramine from the market on 15 Septem- fen-phen case reports. Among them, 88 (84%) met ber 1998. FDA de®nition of valvulopathy. After excluding 14 Although the most recent studies provide further patients without complete information on drug expo- support to the association of appetite-suppressant sure and covariates, the ®nal study sample consisted 4±6 medications with valvulopathy, none of them of 74 fen-phen patients with valvulopathy. Patients with severe valvulopathy were classi®ed as those either undergoing valve replacement surgery or *Correspondence: Ruowei Li, Division of Nutrition and Physical Activity (Mailstop K26), Centers for Disease Control and having severe AR or severe MR; all other patients Prevention, 4770 Buford Hwy, N.E., Atlanta, were considered to have less-severe valvulopathy. GA 30341-3717, USA. w2 test was used to examine the difference in the E-mail: [email protected] Received 23 November 1998; revised 5 February 1999; accepted proportion of severe valvulopathy. Logistic regression 31 March 1999 analysis was used to determine whether the severity of Severity of valvulopathy and fen¯oramine use R Li et al 927 valvulopathy was related to the drug use, after adjust- Table 2 Odds ratios (OR) for severe valvulopathy ing for age, body weight at report, and the duration of Crude OR Adjusted OR a drug use. No interaction terms between duration and (95% CI) (95% CI) drug use were included in the model, because of their Age (10 y) 1.2 (0.7 ± 1.9) 1.3 (0.7 ± 2.4) statistical insigni®cance (P > 0.01). All statistical ana- Weight at Report (10 kg) 1.2 (1.0 ± 1.4) 1.1 (0.8 ± 1.4) lyses were done using SAS version 6.12.7 Duration (1 month) 1.0 (1.0 ± 1.1) 1.0 (1.0 ± 1.1) Fen¯uramine (mg=d) <40 1.0 (ref) 1.0 (ref) 40 1.4 (0.2 ± 8.4) 1.5 (0.2 ± 10.2) Results 60 9.0 (2.2 ± 36.4) 9.2 (2.1 ± 40.8) 95% CI ˆ 95% con®dence interval. aAdjusted for all other variables. Thirty-®ve out of 74 fen-phen patients (47%) were classi®ed as having severe valvulopathy, whereas 39 (53%) were classi®ed as having less severe valvulo- Discussion pathy. Among patients with severe valvulopathy, 17 underwent value replacement surgery, ®ve had severe AR, and 13 had severe MR. In all patients, the median This is the ®rst study to suggest that the severity of dosage for fen¯uramine and phentermine was cardiac valvulopathy was associated with fen¯ura- 60 mg=d (range ˆ 10 ± 220 mg=d) and 30 mg=d mine dosage, but not with phentermine dosage. (range ˆ 15 ± 60 mg=d) respectively, whereas the Among patients with valvulopathy, those with median duration of fen-phen use was 11 months severe valvulopathy were substantially more likely (range ˆ 1 ± 39 months). Table 1 shows that patients to have taken a higher dose of fen¯uramine with severe valvulopathy tended to use higher fen- (60 mg=d) than those with less severe valvulopathy. ¯uramine doses than those with less-severe valvulo- The mechanism of the drug-associated valvulo- pathy (82% of severe vs 38% of less-severe patients pathy is unclear. It has been speculated that serotonin took 60 mg=d, w2 ˆ 15.08, P ˆ 0.0001). The propor- might be the link, because the injured valves look tion with severe valvulopathy increased from 20% identical to those observed in carcinoid-induced valv- (6=30) to 66% (29=44), as fen¯uramine dosage ular disease.1 An additional hypothesis is that phen- increased from 40 mg=d to 60 mg=d. Compared termine may impair the clearance of serotonin by the with patients taking <40 mg=d fen¯uramine, patients lung, which would allow abnormally high concentra- taking 60 mg=d had an adjusted or of 9.2 (95% tions of serotonin to reach the left side of the heart, con®dence interval ˆ 2.1 ± 40.8) for severe valvulopa- when it is taken together with fen¯uramine.9,10 thy (Table 2). The Hosmer and Lemeshow goodness- Because all the subjects in this study were treated of-®t tests indicated that the logistic model ®ts the with both fen¯uramine and phentermine, we cannot data well with goodness-of-®t statistic ˆ 1.3 reject the possibility that the dose-effect of fen¯ur- (P ˆ 0.9961).8 In contrast, both phentermine dosage amine on the severity of valvulopathy may have a and duration of drug use did not appear to be related phentermine-related threshold. to the severity of valvulopathy. Because data on initial body weight were not provided, weight at the time of report was used as a proxy. Patients with severe valvulopathy had a sig- Table 1 Description of cases of severe and less-severe ni®cantly higher weight at the time of report than valvulopathy those with less-severe valvulopathy, which may Severe Less-severe re¯ect differences in initial weight. Nevertheless, the valvulopathy vavulopathy effect of fen¯uramine dosage on severity of valvulo- (n ˆ 35) (n ˆ 39) pathy cannot be ascribed to differential body weight, Female 35.(100%) 38.(98%) because weight was not signi®cantly related to sever- Age (y) 44.3 (s.d. ˆ 9.6) 43.1 (s.d. ˆ 8.5) Weight at report (kg) 95.0 (s.d. ˆ 28.3) 87.1 (s.d. ˆ 18.6)* ity of valvulopathy in the logistic models. Fen¯uramine (mg=d) Although a previous study showed an association of <40 3.(9%) 14.(36%) valvulopathy with the duration of drug use,5 duration 40 3.(9%) 10.(26%) 60 24.(68%) 13.(33%) of drug use did not appear to be related to the severity > 60 5.(14%) 2.(5%) of valvulopathy in this MEDWATCH data. There are Phentermine (mg=d) several explanations. First, the duration of exposure <30 11.(32%) 13.(33%) 30 18.(51%) 21.(54%) depended upon the reporting physician's recall and > 30 6.(17%) 5.(13%) may have been underestimated for those patients who Duration of drug use (months) had previously used appetite-suppressants under the 6 9.(26%) 12.(31%) > 6 ± 12 15.(43%) 14.(36%) care of other physicians. Second, cardiac disorders > 12 ± 24 6.(17%) 10.(26%) may occur as early as a few weeks after drug expo- > 24 ± 5.(14%) 3.(8%) sure,6 whereas 93% of all patients in this study had Percentage may not add up to 100 because of rounding-off. taken the drug for at least 3 months, with a median *P<0.05. duration of 11 months. Therefore, this data set lacks Severity of valvulopathy and fen¯oramine use R Li et al 928 suf®cient data on short duration use to determine the Acknowledgements independent effect of duration. We thank the following persons for their support and Because not all fen-phen users with valvulopathy collaboration in this study: Murray Lumpkin and see physicians, and not all cases of valvulopathy seen Robert O'Neill, Food and Drug Administration; by physicians were reported to MEDWATCH, selec- Susan Yanovski, National Institutes of Health; tion bias could have occurred in this study. However, Wayne Giles, William Kohn, William H Dietz, Cen- because the most frequently used dosage for fen¯ur- ters for Disease Control and Prevention. amine was 60 mg=d, which was the standard dosage, selection of severe valvulopathy cases for reporting References was unlikely to be related to the dosage. In this case, 1 Connolly HM, Crary JL, McGoon MD, Hensrud DD, Edwards 11 our results would be unbiased. Because echocardio- BS, Edwards WD, Schaff HV. Valvular heart disease asso- graphy was not standardized across cardiologists, ciated with Fen¯uramine-phentermine. N Engl J Med 1997; diagnostic misclassi®cation may also have occurred. 337: 581 ± 588. If this misclassi®cation was unrelated to the dosage 2 Lumpkin MM. FDA health advisory. FDA Bull 1997; 27: 2. 3 Bowen R, Glicklich A, Khan M, Rasmussen S, Wadden T, (non-differential misclassi®cation), then our results Bilstad J, Graham D, Green L, Lumpkin M, O'Neill R, Sobel 11 would be biased towards the null. S, Hubbard VS, Yanovski S, Sopko G, Centers for Disease This study also suggests common off-label drug use Control and Prevention. Cardiac Valvulopathy Associated in the USA. For fen¯uramine, the recommended with Exposure to Fen¯uramine or Dexfen¯uramine: U.S. dosage was 60 mg=d for a few weeks with a total Department of Health and Human Services Interim Public dosage not to exceed 120 mg=d.12 However, 9% of all Health Recommendations, November 1997. Morbidity and Mortality Weekly Report 1997; 46(45): 1061 ± 1066. patients in this study took more than 60 mg=d fen- 4 Khan MA, Herzog CA, Peter JV, Hartley GG, Madlon-Kay R, ¯uramine (maximum dosage ˆ 220 mg=d) and 72% of Dick CD, Asinger QW, Vessey JT. The prevalence of cardiac all patients took fen-phen for more than 6 months valvular insuf®ciency assessed by transthoracic echocardio- (maximum duration ˆ 39 months). Although the com- graphy in obese patients treated with appetite-suppressant bination use of fen-phen was not approved by FDA, it drugs. N Engl J Med 1998; 339: 713 ± 718. 5 Jick H, Vasilakis C, Weinrauch LA, Meier CR, Jick SS, Derby was also widely used and the majority of case reports LE. A population-based study of appetite-suppressant drugs received by FDA through MEDWATCH system were and the risk of cardiac-valve regurgitation. N Engl J Med fen-phen users. 1998; 339: 719 ± 724. 6 Weissman NJ, Tighe JF, Gottdiener JS, Gwynne JT. An assessment of heart-valve abnormalities in obese patients taking dexfen¯uramine, sustained-release dexfen¯uramine, or placebo. N Engl J Med 1998; 339: 725 ± 732. 7 SAS Institute Inc. Modeling Techniques for Categorical Response Data. SAS: Cary, NC, 1995. Conclusion 8 Hosmer DW, Lemeshow S. Applied Logistic Regression. John Wiley & Sons: New York, 1989. 9 Maher TJ, Ulus IH, Wurtman RJ. Phentermine and other Based on our results we conclude that, among patients monoamine-oxidase inhibitors may increase plasma serotonin with drug-associated valvulopathy, the risk of clini- when given with fen¯uramine. Lancet 1999; 353: 38. cally severe valvular heart disease increased markedly 10 Fishman AP. Aminorex to Fen=Phen. Circulation 1999; 99: at a fen¯uramine dose 60 mg=d. This ®nding is 156 ± 161. 11 Rothman KJ, Greenland S. Modern Epidemiology, 2nd ed. important in understanding the nature of the drug- Lippincott-Raven: Boston, 1998, Chapter 8. associated valvulopathy and providing appropriate 12 Anonymous. Physicians' Desk Reference. Medical Economics care to patients exposed to fen¯uramine. Data Production Company: NJ, 1997.