Fluoxetine Treatment of Anorexia Nervosa: Title and subTitle BreakImportant but Disappointing Results

Anorexia nervosa is a severe, treatment-resistant illness primarily affecting women. It has one of the highest all-cause mortalities, and the highest suicide rate of any psychiatric illness.^{1 - 2} Restriction of food intake, refusal to maintain adequate body weight, and disturbed thinking about food, weight, and body image are hallmarks of this illness. A prominent associated feature is profound ambivalence about therapeutic efforts (whether nutritional, psychological, or pharmacological) aimed at weight restoration. In addition to the classic pattern of extreme food restriction, a subset of patients also report binge eating and purging behaviors (the “binge-purge subtype”). Co-occurring psychopathology is common, especially depression, obsessive-compulsive disorder, and other anxiety disorders.³

In severely malnourished patients with anorexia nervosa, nutritional rehabilitation with weight restoration is the cornerstone of treatment, and this intervention often requires hospitalization. Psychotherapy is also a regular part of treatment. Preliminary evidence supports use of a specific family-based psychotherapy in adolescents with anorexia nervosa who still reside at home⁴ ; effective alternative psychotherapies for adolescents have not yet been determined. For adults with anorexia nervosa there is some evidence supporting the use of cognitive behavioral therapy,⁵ and other potential psychotherapies for adults are in development.

Psychopharmacology is also a common treatment strategy, although the evidence base for psychopharmacologic approaches for anorexia nervosa is quite limited.⁶ The existing literature on pharmacotherapy for anorexia nervosa is modest in size and striking for the diversity of medications that have been studied. Agents as diverse as lithium,⁷ various antipsychotics,⁸ cyproheptadine,⁹ tricyclic antidepressants,⁹ and tetrahydrocannabinol¹⁰ have been examined with generally negative results in small trials. This wide diversity of agents is clearly an indication of the remarkable challenges inherent in treating individuals who have anorexia nervosa, and also reflects the generally negative results to date in pharmacotherapy trials for this illness.

Recently, there has been much interest in the use of serotonin reuptake inhibitors for acute treatment and relapse prevention in anorexia nervosa. Trials of acute treatment (ie, while patients are still at low weight) with fluoxetine have failed to demonstrate clinical benefit.^{11 - 12} Modest-sized studies of fluoxetine for anorexia nervosa relapse prevention (after an initial weight gain) have shown both positive¹³ (n = 35) and negative¹⁴ (n = 33) results.

In this issue of JAMA, Walsh and colleagues¹⁵ report the results of a placebo-controlled trial of fluoxetine in the relapse prevention treatment of anorexia nervosa. Participants were 93 women aged 16 to 45 years who were completing multimodal inpatient or day hospital treatment and had weight restoration to a body mass index (calculated as weight in kilograms divided by the square of height in meters) of at least 19.0. The patients were randomly assigned to either placebo or fluoxetine, with a dosage goal of 60 mg/d (with decreases permitted for adverse effects and an increase to 80 mg/d permitted for clinical deterioration). A total of 12 months of treatment was provided, during which time all patients also received cognitive behavioral therapy. Relapse was defined as weight loss to a body mass index less than 16.5 or development of medical complications or imminent suicide risk or the onset of another severe psychiatric disorder.

The results of the trial by Walsh et al¹⁵ did not support efficacy for fluoxetine in relapse prevention for anorexia nervosa. Response to fluoxetine was not different from placebo, whether measured in time-to-relapse, percentage of patients maintaining a body mass index of 18.5 or higher, or the percentage of patients meeting modified Morgan-Russell criteria for at least fair outcome (Morgan-Russell criteria provide a slightly broader assessment of anorexia nervosa treatment outcome, extending beyond simple weight restoration).¹⁶ Response by subtype of anorexia nervosa (restricting vs binge-purge) was also examined. The authors note that since a previous fluoxetine trial with positive findings enrolled only participants with the restricting subtype of anorexia nervosa,¹³ perhaps this subtype would respond to fluoxetine. In addition, given the efficacy of fluoxetine for purging in bulimia nervosa, patients with the purging subtype of anorexia nervosa might have been expected to have fared better with fluoxetine. However, analyses by subtype did not affect the results.

Premature study termination, mainly due to treatment failures and patient-initiated withdrawals, was common, but the rate of premature termination did not differ between fluoxetine- and placebo-treated patients. A variety of secondary outcomes were examined, including measures of eating disorder psychopathology, depression, anxiety, self-esteem, and quality of life. Greater reductions in Beck Anxiety Inventory¹⁷ scores were seen in fluoxetine-treated patients, and this difference was statistically significant. There were no differences between fluoxetine- and placebo-treated patients for any of the other secondary outcome measures.

What, if any, current role is there for medication treatment in anorexia nervosa? The results of this study suggest there is little role for fluoxetine. While the results of previous relapse prevention trials have been mixed,^{13 - 14} the report by Walsh and colleagues¹⁵ has many strengths and appears convincingly negative. Even though Beck Anxiety Inventory scores decreased, suggesting a possible role for fluoxetine in diminishing anxiety in such patients, this was an isolated finding among a number of secondary measures, the remainder of which were negative. Intense development efforts are under way aimed at identifying new pharmacologic agents that affect weight and appetite as well as anxiety, mood, and other types of psychopathology in eating disorders.¹⁸ However, much of this effort is focused on strategies to diminish, not increase, food intake to treat obesity. For anorexia nervosa, successful pharmacologic approaches might ultimately be those that target pathological thoughts and attitudes about eating, weight, and body shape, rather than attempting to directly change eating behavior.

Dropout rates in the study were substantial, but generally consistent with previous trials in anorexia nervosa. Treatment adherence is a major challenge in treating this illness; the symptoms of anorexia nervosa interfere with its treatment in a way not typically encountered with other illnesses. Thus, dropout rates are a particularly important aspect of outcome data in anorexia nervosa treatment. In this study, the design may have helped to limit dropouts, as there is evidence that providing medication plus cognitive behavioral therapy is associated with greater persistence in treatment than providing medication alone.¹⁹

Although the results of the current trial are negative, the choice of drug studied was obvious. Fluoxetine does have well-substantiated efficacy in bulimia nervosa^{20 - 21} and has been studied widely enough in obsessive-compulsive disorder to receive a US Food and Drug Administration indication for that use. Given the seemingly close relationship of anorexia nervosa to these illnesses, it was reasonable to have hoped that fluoxetine would provide some benefit in anorexia nervosa. The authors speculate that providing the drug treatment later in the course of recovery (that is, a longer period after hospitalization and weight restoration) might have yielded different results. Perhaps this is true, but clinically the transition from higher- to lower-intensity treatment appears to be a time of high risk for relapse.

Initiating medication at the end of weight restoration is logical, and this choice of study design reflected a clear clinical need that remains unmet. Perhaps the selection of a younger patient sample might have yielded different results, but the age range studied here is fairly typical for this illness. Furthermore, given growing evidence for the effectiveness of family-based therapy in younger individuals with anorexia nervosa, the need for treatment development is particularly great in the postadolescent age range.

The study by Walsh et al¹⁵ is an important contribution that addresses a major gap in research on anorexia nervosa and provides vital information about a fairly common treatment practice for this illness. Unfortunately, it appears that fluoxetine provides no benefit in the relapse prevention treatment of anorexia nervosa. Despite a prevalence rate similar to many other psychiatric illnesses, and particularly in light of the high mortality associated with anorexia nervosa, there is a serious underrepresentation of anorexia nervosa in biomedical research. Much more information is needed on the treatment of individuals with anorexia nervosa while they are at low weight. In addition, strategies must be developed to help individuals who recover to stay in recovery.