Effect of Memantine on Cue-Induced Alcohol Craving in Recovering Alcohol-Dependent Patients

Thirty-eight male inpatients (mean age, 39.2 years [SD=9.0]) who met ICD-10 criteria for alcohol dependence on the basis of clinical interview and DSM-IV criteria according to the Structured Clinical Interview for DSM-IV Axis I Disorders—Patient Edition (7) were recruited from the St. Petersburg Regional Center of Addictions (St. Petersburg, Russia). Participants had been alcohol dependent for a mean of 11.2 years (SD=6.5). Eleven participants (29%) had a first-degree relative who met family history diagnostic criteria for alcohol dependence (8) . Exclusionary criteria included current psychoactive medications and being at risk for untoward side effects from memantine (i.e., liver enzyme levels more than three times higher than normal levels and evidence of active renal, hepatic, or cardiac disease on the basis of history, ECG, or other laboratory findings). Patients had been sober for a mean of 22.1 days (range=5–54 days; SD=13.4) and had not taken any psychoactive medications for at least 1 week before entering the study. Thirty-six participants (95%) smoked cigarettes and met criteria for nicotine dependence; participants had ad libitum access to cigarettes until the initiation of the test. Thirty-one participants (85%) identified vodka as their drink of choice, five (8%) preferred wine, and two (5%) chose beer.

Each subject participated in three daylong testing sessions separated by at least 3 three days. Each test day began at approximately 8:00 a.m. A number of clinical assessments were performed 60 minutes before the administration of study drug and repeated 180 minutes and 360 minutes afterward. At 240 minutes after drug administration—and 2 minutes after an assessment of level of craving—subjects were exposed to an alcohol cue with their drink of choice. They were instructed to smell and handle their drink, but not to consume it, and then to hold it for 3 minutes. At approximately 255 minutes, level of craving was once again assessed. At 390 minutes after drug administration, a relaxation script was reviewed with patients and a general clinical briefing was conducted.

On each test day, a single dose of 20 or 40 mg of memantine or placebo (Merz Pharmaceuticals, Inc., Frankfurt, Germany) was administered in a randomized order, under double-blind conditions. No order effects were observed in this study.

The Biphasic Alcohol Effects Scale (9) includes seven items that assess stimulant effects and seven that assess sedative effects. Visual analogue scales (0–100 mm) were used to assess similarity to ethanol, euphoria, and alcohol craving (10) . Intensity of alcohol effects was assessed with the “number of drinks scale”—the estimated number of standard ethanol drinks (10 g ethanol) needed to produce a given subjective state. The Brief Psychiatric Rating Scale (BPRS) was administered to evaluate a broad range of clinical responses (11, 12) ; in this report we highlight two BPRS subscales, one measuring positive symptoms (psychosis) and the other measuring negative symptoms associated with schizophrenia (11) . Two cognitive tests were used: a verbal fluency test (13) and the Hopkins Verbal Learning Test (14) . All instruments were administered 1 hour before and 3 and 6 hours after drug administration, except the cognitive tests; the verbal fluency test was administered 1 hour before and 3 hours after drug administration, and the Hopkins Verbal Learning Test was administered only at 3 hours.

At the end of each test day, subjects were debriefed about any craving that emerged over the course of the testing.

Each outcome was tested for normality using Kolmogorov-Smirnov test statistics and normal probability plots. The verbal fluency test and the immediate and delayed recall subtests of the Hopkins Verbal Learning Test were approximately normal; however, most other outcomes were positively skewed. Non-normal outcomes were analyzed using the ANOVA-type statistic (ATS) nonparametric approach for repeated-measures data by Brunner (15), and linear mixed models were used for normal outcomes. In these models, each outcome represented the dependent variable, and drug group (20 mg memantine, 40 mg memantine, and placebo) and time were included as within-subject explanatory factors. Subject was used as the clustering factor—that is, observations on the dependent variables were assumed to be correlated within subjects, and these correlations were accounted for in the repeated-measures analyses. Covariates considered in the models included age, family history of alcoholism, number of years dependent on alcohol, drinking pattern (binge or continuous), and history of concussion. The covariates were not significant and were dropped for parsimony, except family history, which contributed significantly to the model for craving visual analogue scale. All results were considered statistically significant at p<0.05 after Bonferroni adjustment for multiple comparisons within but not across hypotheses. Data analyses were conducted with SAS, version 9.1 (SAS Institute, Inc., Cary, N.C.).

Memantine produced modest but significant dose-related ethanol-like effects, as indicated by the visual analogue scale measuring similarity to ethanol, the number of drinks scale, the visual analogue scale measuring euphoria, and the stimulant and sedative subscales of the Biphasic Alcohol Effects Scale ( Table 1 ).

Memantine influenced the extent to which exposure to alcohol cues stimulated craving for alcohol, as measured by the visual analogue scale for craving ( Table 1 ). The presence of a family history of alcohol dependence was associated with higher levels of craving (family history main effect: ATS=4.0, df=1.0, p=0.05), but this history did not moderate the effects of memantine (memantine-by-family history interaction: ATS=0.47, df=1.7, n.s.). Memantine did not stimulate craving for alcohol significantly, and it reduced the cue-induced craving for alcohol. In a direct comparison of the levels of craving before and after cue exposure, the 40 mg memantine dose significantly reduced craving compared with the other two conditions (40 mg memantine versus placebo: t=3.2, df=36, p=0.008; 40 mg memantine versus 20 mg memantine: t=2.9, df=36, p=0.02).

No significant group differences were observed in the verbal fluency test or the immediate and delayed recall subtests of the Hopkins Verbal Learning Test, and memantine did not otherwise impair cognition or affect behavior. No clinically significant adverse events occurred in association with this study.