A Comparison of Aripiprazole, Methylphenidate, and Placebo for Amphetamine Dependence

It was estimated that a group size of 70 individuals was sufficient to detect an effect of medium magnitude (alpha=0.05, beta=0.80). Therefore, we aimed to randomly assign 70 subjects to each group, resulting in a total number of 210 subjects. However, after obtaining unexpected results from interim analysis of the first 53 patients (one active medication arm being significantly worse than placebo arm) and contacting the local ethical committee and the National Agency of Medicines, the enrollment was discontinued.

All study subjects gave written informed consent. The inclusion criteria were amphetamine (or methamphetamine) dependence (per DSM-IV), age between 18 and 65, recent and accustomed intravenous amphetamine/methamphetamine use, confirmed by urinalysis. The study protocol was approved by the local ethical committee. Details of the subject progression through the trial are included in the supplement that accompanies the online version of this article. The study was registered by Current Controlled Trials Ltd (http://www.controlled-trials.com).

The eligibility of the potential study subjects was assessed during a 2-week screening period, after which subjects included in the study started to receive oral regimens of aripiprazole (15 mg/day), methylphenidate (18 mg/day for the first week, 36 mg/day for the second week, and 54 mg/day thereafter) or placebo in identical gelatin capsules for 20 weeks. The medication was given daily under staff supervision (the patient had to swallow the capsule and then drink a sufficient amount of water). The urine samples were obtained twice a week under supervision. All patients received unstructured psychosocial treatment with elements of cognitive therapy and psycho-education, counseling, and support.

The primary outcome measure was the proportion of amphetamine-positive urine samples during pharmacological treatment. An intention-to-treat analysis was used, and all missing samples were considered to be amphetamine positive. The data were analyzed by a logistic regression model. Change in amphetamine use (proportion of positive urine samples) as a function of time in each treatment group was tested with cumulative sums method. Cumulative sums method tests existence of any change as a function of time in proportion of positive urine samples, and if change is observed, detects the change point (6). The retention in treatment was analyzed with Cox’s proportional hazard regression analysis. All data analyses were carried out using R software (7).

The study groups differed concerning their mean age, but not in baseline severity of dependence or other clinical or demographic characteristics ( Table 1 ). None of the subjects were found to be using methamphetamine. No significant difference among the three treatment arms was observed in study retention (p=0.32, age-adjusted p=0.47). Fourteen patients gave at least one amphetamine-negative urine sample. No significant differences were observed in the baseline characteristics between these patients versus those 39 patients who gave no amphetamine-negative samples. The mean proportion of amphetamine-positive urine samples was 90.7% in the aripiprazole group, 67.3% in the methylphenidate group, and 82.0% in the placebo group in observed cases (missing samples ignored). These values were 99.1% (aripiprazole), 93.5% (methylphenidate), and 97.2% (placebo) when all missing urine samples (considered as positive) were included in the intention-to-treat analysis (number of samples for each patient: 40). During the second half of the follow-up period, i.e., during the last 10 weeks, the mean proportion of amphetamine-positive urine samples was 100% in the aripiprazole arm, 46.3% in the methylphenidate arm, and 79.1% in the placebo arm in the observed samples analysis, and 100% in the aripiprazole arm, 91.5% in the methylphenidate arm, and 97.4% in the placebo arm in the intention-to-treat analysis.

Two patients in the aripiprazole group were removed from the trial because of a worsening in their physical condition: one patient developed a transient increase of liver enzymes, attributed to recently started HIV medication, and the other incident was a transient ischemic attack that was attributed to continued amphetamine use. All other recorded adverse events were mild and did not lead to premature discontinuation of the medication. The frequency of reported side effects did not differ between the treatment groups.

In the intention-to-treat analysis, patients allocated to aripiprazole had a higher proportion of amphetamine-positive urine samples than patients who had received placebo (odds ratio=3.09, 95% CI=1.29–7.40; z=2.53, p=0.01). The patients allocated to methylphenidate had fewer positive urine samples than patients in the placebo group (odds ratio=0.42, 95% CI=0.24–0.72; z=–3.14, p=0.002). Since the treatment groups differed in mean age, the age-adjusted odds ratios were also calculated. Adjusted odds ratios were 3.77 (95% CI=1.55–9.18; z=2.92, p=0.003) for aripiprazole and 0.46 (95% CI=0.26–0.81; z=–2.67, p=0.008) for methylphenidate. The fluctuation in the proportion of amphetamine-positive samples (only the analyzed samples included) showed that the significant reduction in amphetamine use was observed after 18 weeks in the methylphenidate arm (p value for change from the baseline was 0.01 for methylphenidate, 0.71 for aripiprazole, and 0.79 for placebo).

Our results indicate that methylphenidate treatment is associated with a statistically significant reduction in intravenous amphetamine use when compared with placebo, providing the first controlled evidence of an effective pharmacological treatment for amphetamine dependence. On the contrary, aripiprazole treatment was associated with a higher proportion of amphetamine-positive urine samples than placebo.

While aripiprazole (in amphetamine dependence) and naltrexone (in opioid dependence) may be good treatments in theory, it seems that effective pharmacological maintenance treatments for intravenous drug dependence are substances that induce at least some euphoria, such as methadone and buprenorphine in opioid dependence (3), or methylphenidate in amphetamine dependence. Slow-release methylphenidate may be superior to usual short-acting formulation, since the patient may start experiencing cravings for amphetamine as soon as the effect of the substitute drug disappears. It is likely that methylphenidate should be dispensed mostly on a daily basis under supervision because of its abuse potential. Aripiprazole (15 mg/day) was not effective in this trial (an abstinence facilitation trial), but we cannot draw any conclusions on its potential efficacy in a relapse prevention study among detoxified patients.

These results show that amphetamine use began to decrease substantially as a function of time after 10 weeks of methylphenidate treatment reaching statistical significance at 18 weeks, which indicates that it may take an even longer period of time than 20 weeks to achieve full benefit from this treatment. The exclusion criteria in the present study were mainly due to safety issues, and the patient population included in the study, having a severe dependence and long history of intravenous drug use, was practically similar to that found in real-life clinical practice. Therefore, this trial was more of an effectiveness study, rather than an efficacy study, and these results can be generalized to real-life settings.

Urinalysis may underestimate the decrease in drug intake, since decline in drug use from several times a day to once a day would indicate no change (8). In this study, the difference between methylphenidate and placebo in the proportions of amphetamine-positive urine samples was 6 percentage units in the intention-to-treat analysis and 33 percentage units in observed samples analysis during the second half of the follow-up phase. It is likely that a large proportion of those patients who were removed from the trial because they stopped visiting the clinic for a time period longer than 7 days would continue to be treated in usual practice. Thus, it can be estimated that absolute risk reduction attributable to methylphenidate treatment measured with urinalysis in a real-life setting might be between 6 and 33 percentage units, and that the figures for the decrease in actual drug use would be probably even greater. An effect of this magnitude should result in a significant reduction of the collateral medical and social problems related to intravenous amphetamine use.

Received Jan. 12, 2006; revisions received Feb. 28 and April 19, 2006; accepted May 11, 2006. From the Department of Forensic Psychiatry, University of Kuopio; the Department of Psychiatry, Institute of Clinical Medicine, University of Helsinki Faculty of Medicine, Helsinki; the Department of Mental Health and Alcohol Research, National Public Health Institute, Helsinki; Substance Abuse Welfare Work and Mental Health Care, Helsinki Deaconess Institute, Helsinki; and the Department of Psychiatry, Helsinki University Hospital, Helsinki. Address correspondence and reprint requests to Dr. Tiihonen, Department of Forensic Psychiatry, University of Kuopio, Niuvanniemi Hospital, FI-70240 Kuopio, Finland; [email protected] (e-mail). All authors report no competing interests. No support was provided by any pharmaceutical company. The study was supported by Helsinki Deaconess Institute; National Public Institute, Helsinki; Helsinki University Hospital; University of Helsinki; and Annual EVO Financing (special government subsidies) from Niuvanniemi Hospital, Kuopio, Finland.The authors thank Erkki Vuori, Ulla Knuuti, Anna-Maarit Penttilä, Petteri Sokero, Hannele Kanerva, Heikki Katila, Aarno Palotie, Sinikka Lassila, Riitta Päivärinta, Sirpa Tamminen, Salli Lesonen, Tuuli Bernard, Pia Vironen, Anna Salo, Markku Nyman, Jouko Mattila, Aija Räsänen, and Tarja Koskela for their collaboration.Current Controlled Trials Ltd. number: 1SRCTN54619615 (www.controlled-trials.com).