Pathological gambling, characterized by persistent and recurrent maladaptive patterns of gambling behavior, is associated with impaired functioning, reduced quality of life, and high rates of bankruptcy, divorce, and incarceration
(1). The past-year adult prevalence rate for pathological gambling is estimated to be 1%, similar to estimates for schizophrenia and bipolar disorder
(2). Because untreated pathological gambling symptoms can impair function in multiple domains
(3), empirically validated treatments for pathological gambling are needed.
Few randomized, controlled clinical trials have evaluated medication treatments for pathological gambling. Studies of serotonin reuptake inhibitors have shown mixed results, with only some studies demonstrating that the efficacy of the drug was superior to that of placebo
(4). In pathological gamblers with co-occurring bipolar disorder symptoms, lithium was superior to placebo in reducing gambling and manic symptoms
(5). Despite their promise, these studies have multiple limitations, including small numbers of patients and geographic homogeneity (i.e., generally performed at single sites), that may restrict the generalizability of the findings.
Given its efficacy in the treatment of alcohol and opiate dependence
(6–
9), the opioid receptor antagonist naltrexone was examined in the treatment of pathological gambling
(10). In a double-blind, placebo-controlled, single-site study of naltrexone, 75% of naltrexone-treated subjects were either “much improved” or “very much improved” according to Clinical Global Impression ratings, compared to 24% of those receiving placebo
(10). Despite the efficacy finding, the high dose of naltrexone (mean end-of-study dose was 188 mg/day) was associated with liver function test abnormalities in more than 20% of naltrexone-treated subjects, consistent with naltrexone’s dose-dependent hepatotoxicity.
Nalmefene hydrochloride, a long-acting opioid antagonist without associated liver toxicity, has demonstrated efficacy in the treatment of alcohol dependence
(11). On the basis of encouraging preliminary reports of the efficacy of naltrexone in pathological gambling and nalmefene’s lack of hepatotoxicity, we conducted a large double-blind, randomized, multicenter trial of nalmefene for pathological gambling. We hypothesized that nalmefene would reduce gambling symptoms (urges/thoughts and behaviors) in subjects with pathological gambling.
Method
Subjects
Men and women age 18 years or older with a primary DSM-IV diagnosis of pathological gambling were recruited through newspaper advertisements and referrals for medication treatment. Subjects were required to meet the DSM-IV criteria for pathological gambling as assessed with the clinician-administered Structured Clinical Interview for Pathological Gambling. A minimum score of 5 on the South Oaks Gambling Screen, at least moderate urges to gamble within the week before study entry (i.e., score ≥2 on the urge component of the Gambling Symptom Assessment Scale), and gambling behavior within 2 weeks before enrollment were required. Women subjects were required to have a negative result on the beta-human chorionic gonadotropin pregnancy test and to use a medically accepted form of contraception.
Exclusion criteria included 1) current axis I disorder determined with the Structured Clinical Interview for DSM-IV (SCID), except for nicotine dependence; 2) lifetime history of bipolar I disorder or bipolar II disorder, dementia, schizophrenia, or any psychotic disorder determined with the SCID; 3) current or recent (past 3 months) DSM-IV substance abuse or dependence; 4) treatment for pathological gambling (other than Gamblers Anonymous) within the last 6 months; 5) baseline score of >17 on either the 17-item Hamilton Depression Rating Scale (HAM-D) or the Hamilton Anxiety Rating Scale (HAM-A); 6) infrequent gambling (e.g., lottery and bingo) that did not meet the DSM-IV criteria for pathological gambling; 7) positive results on a urine drug screen (except for cannabis); 8) unstable medical condition; and 9) concomitant use of psychotropic medication.
The research was conducted at 15 outpatient psychiatric treatment centers in the United States from March 2002 through April 2003. Each treatment center’s institutional review board approved the study and the informed consent procedure. After complete description of the study, subjects provided written informed consent.
Study Design
Dose range selection was based on nalmefene’s clinical and pharmacokinetic data and on studies of naltrexone in the treatment of pathological gambling
(4,
10,
11). Studies with naltrexone in pathological gambling have suggested that relatively high doses (i.e., 3–4 times the recommended therapeutic dose approved for alcohol dependence) may be needed to elicit a therapeutic response
(4,
10). Thus, we selected nalmefene doses of 25 mg/day, 50 mg/day, and 100 mg/day, although findings in alcoholism studies suggested that doses above 20 mg/day or 40 mg/day may confer no additional therapeutic benefit
(11).
After screening, eligible subjects were randomly assigned (in blocks of eight by using computer-generated randomization with no clinical information) to one of the following four conditions: 25 mg/day, 50 mg/day, and 100 mg/day of nalmefene or placebo. Treatment was initiated at 25 mg/day of nalmefene or the placebo equivalent during week 1. At week 2, the subjects randomly assigned to receive 50 mg/day or 100 mg/day of nalmefene began receiving the higher doses. After week 2, the subjects continued to take the doses to which they were randomly assigned. Subjects were free to withdraw from the study at any time. Any subject who was significantly nonadherent to the study procedures could be discontinued from the study. The subjects were assessed during outpatient visits at weeks 1, 2, 4, 6, 8, 10, 12, and 16 of the study.
Screening Assessments
At screening, the subjects were evaluated with the Structured Clinical Interview for Pathological Gambling, a reliable and valid diagnostic instrument that is based on the DSM-IV criteria for pathological gambling. Psychiatric comorbidity was assessed with the SCID. The subjects were assessed for medical history, and a physical examination, electrocardiogram, and routine laboratory testing were completed. The investigators rated subjects’ pathological gambling symptoms using the Yale-Brown Obsessive Compulsive Scale Modified for Pathological Gambling. The subjects reported on the severity of their pathological gambling using the South Oaks Gambling Screen and the self-rated Gambling Symptom Assessment Scale. Subjects’ psychosocial functioning was assessed with the self-report Sheehan Disability Scale. Although subjects with a lifetime alcohol use disorder were excluded, alcohol intake was assessed with the self-report Alcohol Use Disorders Identification Test.
Efficacy and Safety Assessments
The primary outcome measure was the Yale-Brown Obsessive Compulsive Scale Modified for Pathological Gambling total score. Investigators who were blind to subjects’ group assignment administered the scale at every outpatient visit. The Yale-Brown Obsessive Compulsive Scale Modified for Pathological Gambling is a reliable and valid 10-item clinician-administered scale used to rate gambling symptoms within the last 7 days. The first five items of the scale constitute the gambling urge/thought subscale, which measures time occupied with urges/thoughts, interference and distress due to urges/thoughts, and resistance against and control over urges/thoughts. Items 6–10 constitute the gambling behavior subscale, which measures time spent gambling, amount of gambling, interference and distress due to gambling, and ability to resist and control gambling. Items are rated from 0 to 4, with higher scores reflecting greater severity; total scores range from 0 to 40. Each subscale was used as a secondary efficacy measure. Other secondary outcome measures were the Gambling Symptom Assessment Scale, the Clinical Global Impression (CGI) improvement scale, and the Sheehan Disability Scale.
The Gambling Symptom Assessment Scale, a reliable, valid 12-item self-rated scale, is used to assess gambling urges, thoughts, and behaviors during the previous 7 days. Each item is rated 0 to 4, with higher scores reflecting greater pathological gambling severity.
The CGI improvement scale is a reliable, valid seven-item scale that was used to evaluate change in pathological gambling symptoms since the baseline visit. The scale ranges from 1 (very much improved) to 7 (very much worse). Clinicians completed the CGI at every outpatient visit.
The Sheehan Disability Scale is a reliable, valid three-item self-report scale used to assess functioning in work, social, or leisure activities and in home and family life. Each item is rated on an 11-point Likert-type scale ranging from 0 (no impairment) to 10 (extreme impairment). The mean of the three item values was used as a secondary outcome assessment.
Safety assessments (sitting blood pressure, heart rate, adverse effects, and use of concomitant medications) were documented at each visit. Laboratory assessments, including clinical chemistry measures, hematology measures, liver function tests, and urinalysis, were performed at screening and at week 16; liver function tests were also performed at weeks 4 and 8. The 17-item HAM-D, the HAM-A, and urine pregnancy tests were completed at screening and at weeks 8 and 16. Medication adherence was monitored by pill count.
Statistical Analysis
To calculate the number of subjects needed to detect a mean difference in scores on the Yale-Brown Obsessive Compulsive Scale Modified for Pathological Gambling, we used total scores reported in a previous study (mean=14.6, SD=7.1)
(12). For the current study, we assumed 30% and 60% decreases in scores for the placebo group and all nalmefene groups, respectively, by week 16, leading to mean scores of 10.2 and 5.8. Normal distribution was assumed. To detect a mean difference of 4.4 with 80% power and a 5% significance level in a two-sided test, 43 subjects per group were needed. To account for expected dropouts, we chose 50 as the number of subjects needed per group.
All subjects who were randomly assigned to study groups were included in the intent-to-treat analyses of baseline demographic characteristics and safety. In all efficacy analyses, only subjects with at least two postrandomization observations were included (to guarantee that a slope for a linear regression line could be calculated), except for the analysis of CGI improvement, where only one evaluation was required. No imputation was undertaken for missing outcome data. All tests of hypotheses were performed by using a two-sided significance level of 0.05.
The statistical model for the primary variable (Yale-Brown Obsessive Compulsive Scale Modified for Pathological Gambling total score) was a linear mixed-effects model that included terms for treatment group, time, site, treatment-by-time interaction, and treatment-by-site interaction. Each subject’s outcome profile (i) between the week-1 and week-16 visits was summarized with a linear regression line defined by a subject-specific intercept αi and slope βi. The subject-specific intercept αi was assumed to depend on the subject’s Yale-Brown Obsessive Compulsive Scale Modified for Pathological Gambling total score at baseline and on study site and treatment group, while the subject-specific slope βi was assumed to depend on the treatment group. The number of study sites in the analysis was reduced from 15 to six by pooling the data from the sites with small enrollment (N<16). The longitudinal factor of time was defined as weeks since randomization minus 1. Thus, the groupwise intercepts represent model-based mean scores at the scheduled week-1 visit (time=0), while the groupwise slopes represent the evolution of mean scores during weeks 2–16. The model seems to appropriately describe the observed biphasic shape of the response curves.
The statistical test for the null hypothesis “no treatment effect” was performed by testing simultaneously the differences in groupwise intercepts and slopes, with both being clinically meaningful. Linear contrasts were programmed for each nalmefene group versus the placebo group and for all nalmefene groups versus the placebo group. This same statistical model was applied to the Gambling Symptom Assessment Scale total score and Yale-Brown Obsessive Compulsive Scale Modified for Pathological Gambling urge/thought and behavior subscale scores.
The mean scores on the Sheehan Disability Scale had a Poisson-type distribution, with a large proportion of values close to zero. The model was therefore converted to a log-linear model to account for this distribution. Because the first postrandomization Sheehan Disability Scale score was obtained at week 4, the intercepts represent model-based mean scores at week 4.
The rates of CGI improvement were evaluated by fitting an ordinal logistic regression model in which the distributions of the last-observed CGI ratings are associated with the indicators of treatment group and gender.
Cox proportional hazards regression analysis was performed to examine time to discontinuation. Group differences in the HAM-A and the HAM-D scores were tested cross-sectionally with Kruskal-Wallis tests, as linear models were considered inappropriate because of the strongly skewed distributions. Descriptive statistics were used to evaluate changes in laboratory values, blood pressure, and heart rate.
All statistical analyses were carried out by using SAS 8.2
(13). The linear mixed-effects models, log-linear mixed-effects models, logistic regression models, and survival analysis were fitted with the procedures MIXED, NLMIXED, LOGISTIC, and PHREG, respectively
(13).
Discussion
In this multicenter, randomized, double-blind clinical trial, we found that nalmefene was superior to placebo in the treatment of pathological gambling across a spectrum of illness-specific and global outcome measures. The results demonstrate that nalmefene treatment reduces the symptoms associated with pathological gambling. Of the three fixed doses evaluated, the 25 mg/day and 50 mg/day doses demonstrated superior efficacy, compared to placebo, on the primary efficacy measure (Yale-Brown Obsessive Compulsive Scale Modified for Pathological Gambling total score) and secondary efficacy variables, including the Yale-Brown Obsessive Compulsive Scale Modified for Pathological Gambling behavior and urge/thought subscale scores and the Gambling Symptom Assessment Scale total score. Only the 25 mg/day dose demonstrated efficacy superior to placebo in terms of the overall response to treatment (measured by the CGI). The 100 mg/day dose seemed to confer no additional benefit, compared to the lower dose levels, on any efficacy measure, and the upper limit of the dose range we selected thus seems to have been unnecessarily high.
The efficacy of nalmefene lends support to the hypothesis that pharmacological manipulation of the opiate system may target core symptoms of pathological gambling
(10). Opioid antagonists have been effective in treating other addictive disorders involving alcohol, heroin, and cocaine use
(6–
9). It has been proposed that the efficacy of opioid antagonists in the treatment of addictive disorders involves opioidergic modulation of mesolimbic dopamine circuitry
(14). The behavioral effects of opioid antagonist administration include diminished urges to engage in the addictive behavior and longer periods of abstinence
(6–
11), consistent with a mechanism of action involving ventral striatal dopamine systems
(15–
17). Further work to define the precise manner in which nalmefene and other opioid antagonists mediate their beneficial effects could enhance treatment strategies for pathological gambling, other impulse control disorders, and substance use disorders.
Nalmefene has been extensively studied, and its lack of potential hepatotoxicity may present a marked advantage over other opioid receptor antagonists, such as naltrexone
(11). Adverse events reported in this study were consistent with nalmefene’s previously reported safety profile
(11,
18,
19). In contrast to naltrexone, nalmefene has not resulted in hepatotoxicity, regardless of dose. Although there has been concern that opioid antagonists may engender depression
(20), there were no increases in the depression scores (HAM-D) or the anxiety scores (HAM-A) during treatment in the subjects in the current study.
To our knowledge this study represents the largest randomized pharmacotherapy trial involving subjects with pathological gambling performed to date, but several limitations exist. First, pathological gambling is a chronic disease that may require long-term therapy. By design, this study did not assess treatment effects beyond the acute 16-week treatment period, and longer-term effects thus require further evaluation. It is possible that a longer course of therapy could result in continued and even greater reductions in gambling symptoms. Alternatively, nalmefene’s therapeutic effects in pathological gambling might not endure beyond 16 weeks. Second, we enrolled subjects seeking pharmacological treatment, not psychotherapy, and we recruited only subjects without current comorbidities. Given these stringent exclusion criteria (e.g., no comorbidity with substance use disorders or bipolar disorder), these results may not generalize completely to the larger population of people with pathological gambling. Third, approximately two-thirds of the subjects discontinued treatment. Although rates of treatment discontinuation in studies of pathological gambling are generally high (up to 49%)
(4,
10), discontinuation in this study was most likely a result of poor management of medication side effects. It is likely that the initial dose was too high and that the dose titration at week 2 was too abrupt. Flexible dosing strategies may have allowed for improved tolerability. The relatively high discontinuation rate may compromise the conclusions drawn from this study. Perhaps only those subjects truly committed to stopping gambling tolerated the side effects of nalmefene and stayed in the study. In addition, the elevated rates of nausea for subjects who received nalmefene, compared to those who received placebo, may have jeopardized the blind. Evaluation of lower doses and a slower titration on initiation warrant consideration. Fourth, although subjects were excluded if they had lifetime bipolar I disorder or bipolar II disorder, it is possible that some may have had histories of subsyndromal mania or hypomania. The presence of these subsyndromal symptoms may have led to discontinuation of some subjects taking nalmefene, because the medication could have induced subtle mood destabilization. More detailed assessments of subsyndromal mood symptoms are needed for future studies. Fifth, the subjects assigned to receive placebo demonstrated improvement over time. Although this placebo effect is a confounder, examination of the relative pattern over time demonstrated that the treatment signal appeared to outweigh the placebo effect. Sixth, this study did not include behavioral therapy. Effective behavioral treatments for pathological gambling are emerging
(21) and should be considered in conjunction with pharmacotherapies.
This investigation suggests that nalmefene may be effective in the acute treatment of pathological gambling. Although optimal dosing and titration of nalmefene cannot be determined from this study, lower doses and a slower titration should be considered for future studies. As effective treatments for pathological gambling emerge, it becomes increasingly important that physicians and mental health care providers screen for pathological gambling in order to provide timely treatment.
Editor"s note: Namelfene in tablet form is an investigational drug in the United States and not yet approved for general clinical use.