A placebo-controlled pilot study of the novel opioid receptor antagonist ALKS-33 in binge eating disorder




To assess preliminarily the effectiveness of a novel opioid antagonist, ALKS-33, in binge eating disorder (BED).


In this randomized, placebo-controlled, flexible dose, proof-of-concept trial, 62 outpatients with BED and obesity received ALKS-33 (N = 26) or placebo (N = 36) for 6 weeks. Outcome measures of binge eating, body weight, and eating pathology were assessed.


A large decrease in binge eating episode frequency was observed following both ALKS-33 and placebo treatment. There was no significant difference between treatment groups in binge eating episode frequency or any other measure of binge eating, body weight, or eating pathology.


In this preliminary proof-of-concept study in BED, ALKS-33 did not separate from placebo. Although a failed trial cannot be excluded, the finding is consistent with earlier observations in bulimia nervosa with other opioid antagonists and suggests ALKS-33, at least when administered daily for 6 weeks, may not be efficacious for BED. © 2013 by Wiley Periodicals, Inc. (Int J Eat Disord 2013)


Binge eating disorder (BED), defined by recurrent and distressing episodes of binge eating without the inappropriate compensatory weight loss behaviors of bulimia nervosa,1 is an important public health problem.2, 3 The lifetime prevalence of BED in the general population of the United States is estimated to be 3% and it is associated with psychiatric comorbidity, obesity, impaired quality of life, and disability.2–4

Despite the effectiveness of specialized psychotherapies and serotonin reuptake inhibitors (SSRIs), the treatment of BED remains a challenge.5, 6 Few therapies are associated with clinically significant weight loss, no medication is currently approved by a regulatory body for the treatment of BED, and many of the pharmacological agents examined to date have been associated with problematic side effects. Novel treatments that are well tolerated are therefore needed for BED.

ALKS-33, a new oral opioid modulator in clinical development for alcohol dependence, binds selectively to the opioid receptor.7, 8 The compound acts primarily as an antagonist at the μ opioid receptor. Unlike other drugs in this class, nonclinical studies indicate that ALKS-33 does not undergo extensive first-pass metabolism by the liver, and that it exhibits high oral bioavailability. These attributes may offer an improved safety margin compared with other opioid receptor antagonists, which carry safety warnings regarding hepatoxicity.

Several lines of evidence suggested that ALKS-33 would be a useful treatment for BED. First, the endogenous opioid system is involved in eating behavior, and preclinical data indicate that opioid agonists enhance, while opioid antagonists suppress, binge eating behavior in animal models.9–13 For example, the κ opioid agonist butophanol enhanced consumption of palatable food in a rodent model of binge eating, while the opioid receptor antagonist naloxone completely blocked this behavior.9 Nalmefene, a μ and κ opioid antagonist, significantly attenuated binge eating in rats conditioned to binge eat a high sugar diet.10 Moreover, in humans, the opioid antagonist naloxone selectively suppressed the consumption of sweet, high-fat foods in obese and lean subjects with bulimia nervosa, but not in control subjects;14 while the μ opioid receptor antagonist GSK1521498 significantly reduced attentional processing of food cues in obese subjects with binge eating.15 Second, although placebo-controlled studies of naltrexone at standard doses in patients with bulimia nervosa have largely been negative,16, 17 studies in bulimia nervosa and case reports in BED have suggested that supratherapeutic doses of the drug (e.g., 200 to 400 mg per day) may suppress binge eating.18–22 Compared to naltrexone, ALKS-33 has a substantially enhanced pharmacodynamic and pharmacokinetic profile with a 5-fold greater affinity at the μ opioid receptor and much greater bioavailability after oral ingestion (Alkermes data on file).

Third, preliminary safety data indicate that ALKS-33 is well-tolerated at the tested dose levels (Alkermes data on file). Compared with the pharmacological agents currently used to treat BED, ALKS-33 has the potential to be associated with lower rates of certain side effects (e.g. sexual dysfunction, cognitive impairment).

These observations led to the hypothesis that ALKS-33 would be an efficacious, safe, and well-tolerated treatment for BED, with sufficient potency and tolerability relative to commercially available opioid antagonists. The present study was a randomized, parallel-group, placebo-controlled, 6-week proof-of-concept study of ALKS-33 in 69 adult participants. In addition to evaluating the treatment effects of ALKS-33 on binge eating and other aspects of eating pathology, the effects of ALKS-33 on various metabolic measures, including weight, body mass index (BMI), and waist circumference, were assessed. The safety of ALKS-33 in participants with BED was also evaluated.



Male and female individuals ≥18 years of age with a body mass index (BMI) ≥30 mg/kg2 who met the DSM-IV-TR criteria for BED, as determined by the Structured Clinical Intervention for DSM-IV-TR (SCID-P),23 were eligible. In addition, participants had to display ≥3 binge eating days per week as assessed by take-home binge diary during the 2 weeks of the screening period prior to baseline (randomization). A binge eating day was a day with at least one binge eating episode.

Individuals were excluded from participation if they had concurrent symptoms of bulimia nervosa or anorexia nervosa; suicidal ideation (defined as a score ≥2 on Item 9 of the Beck Depression Inventory, 2nd Ed [BDI]24); current major depressive disorder or a BDI score ≥17; lifetime bipolar or psychotic disorder; substance abuse or dependence (except nicotine or caffeine abuse or dependence) within 6 months prior to randomization; or any psychiatric disorder that might interfere with a diagnostic assessment or compliance with study procedures. Comorbid disorders were determined with the SCID-P.23 Individuals were also excluded if they had a positive urine toxicological screen at screening or randomization; participated in a psychological or weight loss intervention for BED that was initiated within the 3 months prior to screening; or had clinically unstable medical disease or clinically significant findings on ECG, urinalysis, or laboratory results. Individuals who had used any psychotropic medications (other than hypnotics) within 4 weeks prior to randomization as well as those who had a current or anticipated need for prescribed opioid medication during the study period were also ineligible. Women were excluded if they were pregnant, lactating, or if fertile, not using medically accepted contraception.

All participants signed approved consent forms prior to study procedures. The first participant was randomized 6/12/2010. The last was randomized 4/10/2011.

Study Design

This was a Phase 2, 6-week, randomized, double-blind, fixed-dose pilot study to evaluate the safety and efficacy of ALKS-33 compared with placebo for the treatment of BED conducted at 6 sites. One site, randomizing 29 participants, was an academic eating disorder program. The other 5 sites, which randomized the remaining 33 participants, were four private research groups and one academic eating disorder program.

The trial consisted of three phases: a 2- to 4-week screening period; a 6-week double-blind treatment period; and a 2-week follow-up period. Participants were evaluated at least twice during the screening period; after 1, 2, 3, 4, and 6 weeks during the treatment period; and after 2 weeks of medication washout during the follow-up period. At the last screening (baseline) visit, eligible participants were randomized in a 1:1 ratio to 10-mg ALKS-33 or matching placebo, given as a once daily nighttime dose, because of the incidence of somnolence observed in previous studies. At the discretion of the investigator, 1 dose decrease (from 10-mg to 5-mg ALKS-33 in the active group) was permitted for any participant who had poor tolerability to treatment. The investigator and all study personnel remained blinded to study treatment, except for the unblinded study pharmacist who was responsible for preparing all study drug and implementing any dose reduction orders. Participants were issued emergency treatment cards to notify emergency health professionals that they might be taking an opioid antagonist.

On an ongoing basis, participants completed take-home binge, study drug dosing, and noninvestigational medication diaries. The binge diary was used to record the number of binge eating episodes (binge episodes) and binge days, as well as snacks and meals consumed daily for the interval between visits. Adherence with study medication was further assessed with tablet count. Additionally, participants were counseled at each study visit about the importance of taking study medication as instructed.

Randomization was performed using a single master randomization schedule that was provided to the unblinded pharmacist at each study site; randomization was stratified by site. The pharmacist prepared all study drug in a double-blinded manner before providing it to the blinded study personnel for administration to participants. With the exception of the unblinded site pharmacist, treatment assignment was concealed from the participants, investigators, other study personnel, and other sponsor personnel throughout the study duration.

The intent-to-treat (ITT) population included all randomized patients who received at least one dose of study drug and who had at least one post-baseline efficacy binge assessment. The safety population included all randomized patients who received at least one dose of study drug.

Outcome Measures

As this was a pilot study of ALKS-33 as a treatment for BED, a number of exploratory efficacy endpoints related to BED were analyzed descriptively in order to gain a better understanding of ALKS-33 in this patient population. Efficacy variables included: weekly binge frequency (WBF), defined as the total number of binges in the interval between visits, divided by the number of days with actual reported diary data, expressed as a 7-day average; weekly binge days (WBD); response, reflecting groups of participants who reduced their WBF by, respectively, at least 2, 3, or 4 binges; response, reflecting groups of participants who reduced their WBD by, respectively, at least 1, 2, or 3 binge days; remission, defined as no binges occurring in the last two study weeks or in the interval between an early termination visit and the previous visit; weight, BMI, and waist circumference; and scores from clinical assessment instruments. The latter included the Yale-Brown Obsessive-Compulsive Scale Modified for Binge Eating (YBOCS-BE),25 Clinical Global Impression Scale for Severity (CGI-S),26 Eating Inventory (EI),27 Food Craving Inventory (FCI),28 and BDI.24

The following safety measures were assessed: adverse events, clinical laboratory and ECG data, physical examination findings, vital signs, and the Columbia-Suicide Severity Rating Scale (C-SSRS).29

Statistical Methods

Baseline demographic and clinical characteristics, efficacy variables, participant disposition, and adverse events were compared between treatment groups using two-sample t-tests or Wilcoxon-Mann-Whitney tests for continuous variables, and chi-square or Fisher's exact tests for categorical variables. Wilcoxon-Mann-Whitney nonparametric tests were used to assess treatment differences in WBF and WBD changes from baseline at each visit. Additionally, longitudinal analyses were conducted for both the WBF and WBD outcomes. The difference in rates of change of these variables were estimated by random regression models, as described in Fitzmaurice et al.30 and Gibbons et al.31 and as used in previous pharmacotherapy studies of BED.32, 33 The models included treatment, time, and treatment by time interaction. Time was included as square root of days since baseline. The outcome variables were log transformed to correct for non-normality and to stabilize the variance. To simultaneously account for individual differences in baseline levels of the outcome, rate of change over time, and correlated within-subject observations, the models contained random coefficients for the intercept and slope, and a first-order antedependence structure for the residual correlation matrix.

The intent-to-treat (ITT) population was used for all analyses, except for the safety measures which were assessed on the safety population. Due to the small overall sample size, unbalanced sample sizes over sites, and preliminary nature of the study, site differences were not considered. The analyses were performed with SAS version 9.2 (SAS Institute, Inc., Cary, N.C.). All statistical tests were 2-sided with α = 0.05.


Of 155 individuals screened, 69 met entry criteria and were randomized to ALKS-33 (N = 32) or placebo (N = 37). Sixty-two of these individuals, 26 receiving ALKS-33 and 36 receiving placebo, completed at least one post-baseline binge efficacy assessment and thus comprised the ITT population. Of these 62 participants, their mean (SD) age was 45.2 (11.3) years, 56 (90%) were female, 50 (81%) were white, their mean (SD) baseline weekly binge day frequency was 3.6 (1.3), and their mean (SD) BMI was 38.95 (5.8) kg/m2. Baseline demographic and clinical characteristics are shown in Table 1. Since participants treated with ALKS-33 were significantly younger (mean [SD] age = 40.6 [11.2] years) than their placebo-treated counterparts (mean [SD] age = 48.6 [10.2] years, p = 0.005), age was added as a covariate in the longitudinal analyses. There were no other significant baseline differences between the treatment groups.

Table 1. Baseline demographic and clinical characteristics of 62 participants with BED randomly assigned to 6 weeks of double-blind treatment with ALKS-33 or placeboa
Characteristic/VariableALKS-33 (N = 26)Placebo (N = 36) p-Valueb
  • a

    Intent-to-treat population.

  • b

    Chi-square, Fishers Exact Test or t-Tests.

Demographic Characteristics   
Female, N (%)23 (88%)33 (92)0.69
Caucasian, N (%)22 (85%)28 (78%)0.50
Age (years), Mean (SD)40.6 (11.2)48.6 (10.2)0.005
Outcome variables   
Weekly binge days, Mean (SD)3.4 (1.0)3.7 (1.5)0.35
Weekly binge frequency, Mean (SD)4.4 (2.3)4.3 (2.3)0.87
Weight (kg), Mean (SD)106.0 (13.5)107.6 (17.7)0.70
Body mass index (kg/m2), Mean (SD)38.6 (4.8)39.2 (6.4)0.69
Waist Circumference (cm)114.1 (9.9)115.2 (12.1)0.70
Clinical Global Impression- Severity of Illness Scale, Mean (SD)4.2 (0.8)4.4 (0.7)0.41
Yale-Brown Obsessive-Compulsive Scale (modified for binge eating), total, Mean (SD)20.3 (4.0)20.6 (5.0)0.80
Three-Factor Eating Questionnaire, total, Mean (SD)29.3 (5.8)30.7 (4.8)0.30
 Cognitive Restraint Subscale6.0 (3.8)6.9 (3.7)0.35
 Disinhibition Subscale12.8 (2.8)13.4 (1.8)0.34
 Hunger Subscale10.6 (3.2)10.3 (2.9)0.70
Food Craving Inventory, total, Mean (SD)2.8 (0.7)2.9 (0.6)0.55
Beck Depression Inventory, total, Mean (SD)3.7 (2.8)4.7 (4.1)0.29

Sixteen (50%) of the 32 participants in the ALKS-33 group discontinued prematurely (adverse events, N = 12; lost to follow-up, N = 3; participant withdrawal, N = 1); and 4 (11%) of the 37 participants in the placebo group discontinued prematurely (lost to follow-up, N = 4). Individuals in the ALKS-33 group discontinued at a significantly higher rate than those in the placebo group (50% vs. 11%, p < 0.001). This difference was also seen when only considering discontinuation rate from the time of randomization to finalization of the ITT group (19% vs. 3%, p = 0.04) and discontinuations due to adverse events (37% vs. 0%, p < 0.001). Forty nine (71%) participants completed the 6-week treatment phase. At endpoint evaluation for the 26 participants receiving active study drug, 22 were receiving 10 mg daily and 4, 5 mg daily.

Weekly binge frequency (WBF) decreased a modest and stable degree in both treatment groups over the course of the trial (Fig. 1). As2 shown in Table 2, ALKS-33 was not significantly superior to placebo in reducing WBF or WBD, body weight, BMI, waist circumference, obsessive-compulsive features of binge eating symptoms, other measures of eating pathology, or overall severity of illness. ALKS-33 was not associated with a significantly higher level of categorical response, whether response was defined as reduction in WBD by ≥ 1, 2, or 3 binge days, or WBF by ≥ 2, 3, or 4 binges, compared with baseline. Results from the longitudinal analyses showed an overall significant reduction in WBF (p < 0.001), but no significant ALKS-33 effect over time as measured by the treatment by time interaction (p = 0.35). Similarly, an overall significant reduction was seen in WBD (p < 0.001), but with no treatment effect over time (p = 0.22). At the last post-baseline assessment, remission rates were not significantly different (p = 0.16) between the ALKS-33 recipients (9 of 26 [35%]) and the placebo recipients (19 of 36 [53%]).

Figure 1.

Weekly binge frequency with ALKS-33 versus placebo in intent-to-treat population (mean [SD]).

Table 2. Change in outcome variables (from endpoint to baseline) in 62 participants with BED randomly assigned to ALKS-33 or placeboa
 ALKS-33 (N = 26)Placebo (N = 36) p-valueb
  • a

    Intent-to-treat population.

  • b

    Chi-square, Fishers Exact, Wilcoxon-Mann-Whitney or t-Tests.

  • b

    Response = reduction in weekly binge day frequency by ≥ 1, 2, or 3 binge days, or weekly binge frequency by ≥ 2, 3, or 4 binges, compared with baseline.

  • d

    Remission = no binges in last 2 study weeks or in interval between an early termination and the previous visit.

Weekly binge days, Mean (SD)−2.4 (1.6)−2.7 (1.5)0.50
Responder, ≥ 1 dayb, n (%)20 (77%)32 (89%)0.30
Responder, ≥ 2 day, n (%)18 (69%)24 (67%)0.83
Responder, ≥ 3 day, n (%)11 (42%)18 (50%)0.55
Weekly binge frequency, Mean (SD)−3.3 (2.4)−3.2 (1.8)0.99
Responder, ≥ 2 bingesb, n (%)19 (73%)26 (72%)0.94
Responder, ≥ 3 binges, n (%)16 (62%)20 (56%)0.64
Responder, ≥ 4 binges, n (%)10 (38%)13 (36%)0.85
Remissiond9 (35%)19 (53%)0.16
Weight (kg), Mean (SD)−0.03 (2.02)−0.23 (3.16)0.76
Body mass index (kg/m2), Mean (SD)0.01 (0.69)−0.07 (1.27)0.75
Waist Circumference (cm), Mean (SD)−0.95 (4.14)−0.57 (3.54)0.74
Clinical Global Impression–Severity, Mean (SD)−1.6 (1.6)−2.1 (1.3)0.19
Yale–Brown Obsessive–Compulsive Scale (modified for binge eating), total, Mean (SD)−10.2 (8.1)−12.3 (8.2)0.32
Three–Factor Eating Questionnaire, total, Mean (SD)−2.6 (5.3)−4.6 (7.2)0.24
Cognitive Restraint Subscale1.2 (3.3)1.3 (3.3)0.91
Disinhibition Subscale−1.7 (2.5)−2.9 (3.6)0.15
Hunger Subscale−2.0 (3.3)−3.1 (3.9)0.25
Food Craving Inventory, total, Mean (SD)−0.54 (0.44)−0.72 (0.64)0.22
Beck Depression Inventory, total, Mean (SD)−0.7 (3.5)−2.4 (4.1)0.09

Of the 68 participants receiving at least one dose of study medication, significantly more ALKS-33 recipients [29 (93.5%)] reported adverse events than placebo recipients [24 (65%); p = 0.004]. The most common adverse events reported by ALKS-33-treated participants were nausea, dizziness, headache, insomnia, dry mouth, fatigue, somnolence, vomiting, and sedation (Table 3). Of the 12 participants who stopped ALKS-33 prematurely, single adverse events led to drug discontinuation in 6 participants. These were dizziness (N = 2), nausea (N = 1), somnolence (N = 1), hallucination (N = 1), and palpitations (N = 1). Multiple adverse events lead to discontinuation in the other six participants and included nausea and/or vomiting (N = 5), photopsia or visual hallucinations (N = 2), dizziness (N = 2), lethargy or somnolence (N = 2), insomnia (N = 3), agitation (N = 1), headache (N = 1), tremor (N = 1), oropharyngeal pain (N = 1), fatigue (N = 1), diarrhea (N = 1), dry mouth (N = 1), hot flush (N = 1), abnormal dreams (N = 1), vertigo (N = 1), irritability (N = 1), and akathisia (N = 1). There were no serious adverse events.

Table 3. Side effects in 68 patients with BED receiving at least 1 dose of ALKS-33 or placeboa
 ALKS-33 (N = 31) N (%)Placebo (N = 37) N (%)
  • a

    Safety population.

Nausea10 (32.3)4 (10.8)
Dizziness10 (32.3)0
Headache9 (29.0)6 (16.2)
Insomnia7 (22.6)1 (2.7)
Dry Mouth5 (16.1)1 (2.7)
Fatigue4 (12.9)4 (10.8)
Somnolence4 (12.9)3 (8.1)
Vomiting4 (12.9)1 (2.7)
Sedation3 (9.7)1 (2.7)
Diarrhea2 (6.5)2 (5.4)
Muscle spasms2 (6.5)2 (5.4)
Sleep disorder2 (6.5)2 (5.4)
Myalgia1 (3.2)3 (8.1)

There were no significant changes in physical examination or ECG findings, vital signs, or clinical laboratory values. There was no evidence of withdrawal symptoms upon discontinuation of ALKS-33. There was no emergence of suicidal ideation or behavior.


In this randomized, double-blind, 6-week trial in 62 patients with BED, the novel opioid antagonist ALKS-33 was not superior to placebo in reducing measures of binge frequency, binge eating response or remission, eating pathology, or body weight. It was associated with a significantly higher rate of adverse events, and a higher discontinuation rate due to adverse events.

The lack of efficacy found for ALKS-33 for BED is inconsistent with preclinical data suggesting opioid antagonists might be therapeutic agents for binge eating.9–15 It is, however, consistent with extant clinical trial data finding other opioid antagonists ineffective for the treatment of the related conditions bulimia nervosa and obesity, at least when administered as monotherapy at usual therapeutic doses.16, 34 Thus, our findings do not exclude the possibility that ALKS-33 might be efficacious for BED if used adjunctively (e.g., with an SSRI22 or bupropion, as has been successfully done with the latter agent for obesity35, 36), in higher doses (though adverse events might limit or preclude this option), or if administered by a different route (i.e., parenteral or intranasal) or preparation (i.e., via an extended release formulation).37, 38 Nor does it preclude the possibility that other novel opioid modulating agents with different mechanisms of action might have beneficial effects on binge eating in BED.13, 15

The frequency of adverse events observed in this group of BED patients was higher than what has been observed in other ALKS-33 studies and they appeared more likely to discontinue from the study due to an adverse event. Of 101 participants with alcohol dependence treated with 10 mg of ALKS-33 for 12 weeks in the ALK33-005 study (ClinicalTrials.gov Identifier NCT00981617), 84% and 15% had an adverse event and an adverse event leading to study discontinuation, respectively (Alkermes data on file). These rates are substantially lower than those observed in this study despite the longer duration of treatment in the alcohol dependence trial. Reasons for this potential difference are unknown, but could reflect unique illness-related sensitivity factors in the BED population studied.

This study had several limitations that should be considered. First, because of the small sample size and high placebo response rate, the study may have been inadequately powered to detect clinically important treatment effects of moderate size. Therefore, it could be argued that the study allowed for a high likelihood of a type II error, indicating a nonsignificant result to be inconclusive (i.e., the study was a failed trial rather than a negative trial). However, this possibility appears remote given the observed data. Using the baseline-to-endpoint change in WBF, the effect size seen in our study was 0.03 (95% CI = -0.48, 0.53), with the ALKS-33 group (-3.3) having a slightly greater reduction in WBF than the placebo group (-3.2). A moderate effect size is traditionally considered to be 0.50,39 which in our study corresponds to a decrease of more than 1 binge episode per week in the ALKS-33 group than in the placebo group. Given our observed data, the probability that the actual effect size is 0.50 or larger is 0.035. Additionally, it is possible to simulate additional observations, assuming optimistically that the true difference in WBF reduction between the groups is 1 day (0.50 effect size). This approach to futility monitoring will yield the probability that continuing the study will lead to a significant result.40 Two hundred datasets were simulated for an additional 62 participants (26 in placebo group; 36 in ALKS-33 group). The probability that the combined sample (n=124) would produce a significant result is 0.28. To summarize, the observed data argue against a clinically meaningful effect size of 0.50 for ALKS-33, as evidenced by a less than 4% probability that the true effect size is 0.50 or larger; and even assuming this unlikely event is true, doubling the sample size would have been unlikely to produce a significant ALKS-33 effect on binge eating. Nevertheless, the possibility that the large placebo effect observed for binge episode frequency obscured the study's ability to detect a significant treatment effect cannot be excluded.

A second limitation is a high overall attrition rate, rendering results heavily dependent on assumptions regarding missing data. Additionally, the greater discontinuation rate with ALKS-33 prior to finalizing the ITT sample may have compromised randomization. Third, is that ALKS-33 may not have been dosed properly or given for an adequate period of time as used in this trial, though the duration was reasonable in light of extant positive BED trials. A fourth limitation is that persons with substance use disorders were excluded and it is possible that ALKS-33 might be efficacious in BED when it co-occurrs with addictive disorders.41

In summary, in a 6-week trial in outpatients with BED, the opioid antagonist ALKS-33 was not superior to placebo in reducing binge episode or binge day frequency, improving eating disorder psychopathology related to BED, or reducing body weight. It was also not well tolerated when begun and continued at a daily dose of 10 mg in BED.