Study Selection and Characteristics
Our search returned 43 RCTs that met our criteria [24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66]. A PRISMA flow diagram is presented in Fig. 1. Forty (93.0%) of these studies were double-blinded [24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45, 47,48,49,50,51,52,53,54,55, 57,58,59,60,61,62,63,64,65]. In total, 39 (90.7%) were placebo controlled, while the remaining four studies (9.3%) were designed with a treatment-as-usual or alternate treatment arm as the control [36, 49, 61, 66]. The study settings are described in Table 2. The studies were published between 1995 and 2019.
The 43 studies collectively randomised 4065 participants. Of the 43 studies, 38 (88%) reported on the total number of participants who completed the study, while five studies (12%) did not [36, 39, 57, 62, 64]. Of the 38 reporting on study completion rates, the total number of participants randomised was 3733 (92% of the total) and of these, 2298 participants completed the study (61.6%).
Of the 4065 participants reported on in the reviewed studies, 2858 (70.3%) were male. Nine of the 43 studies (21%) enrolled only males [24, 29, 30, 34, 46, 52, 55, 57, 58], however not all of these were by design. One study  did not report the sample by those randomised, only by completers (in a non-intention-to-treat analysis) and so the total randomised figure of men versus women was not able to be determined based on randomisation (authors did not respond to a request for further data).
Thirty of the 43 studies (69.8%) included participants dependent on MA only [24,25,26,27,28,29,30, 32,33,34,35, 38,39,40,41, 44, 45, 47,48,49,50, 53,54,55, 57, 59,60,61,62, 65]; four (9.3%) were on MA/AMPH [31, 37, 51, 64]; four (9.3%) were on AMPH only [42, 43, 46, 56]; three (7.0%) were on amphetamine-type stimulants (ATS) and opioids [52, 58, 63] and two (4.7%) were on MA/cocaine [36, 66].
The 43 studies examined 23 individual pharmacotherapies, most individually and some in combination. Table 3 lists the pharmacotherapies reviewed, and the proposed mechanisms of action related to their use in studies of MA/AMPH dependence.
A summary of the reviewed studies is presented in Table 4, and an extended version is available in Supplementary Table 1 (see ESM). In addition, the data collected by both reviewers can be located in its entirety in the Supplementary Data (see ESM).
Risk of Bias Within and Across Studies
Risk of bias in individual study methods and reporting are included in Supplementary Table 1 and Supplementary Data (see ESM) as considerations across a number of domains. Across all studies, allocation of participants was by random assignment, and all but three studies [46, 56, 66] were double-blind. Study completion rates were low, with studies reporting the proportion of the sample who did not complete the protocol as 38.4% of the total randomised. Eighty-three percent of studies analysed their results by intention-to-treat, while five (12%) [33, 46, 53, 57, 61] were unclear in this regard and two (5%) [24, 45] did not. Females were underrepresented in the data, being only 29.7% of the total participants. This comprises both studies that only enrolled males (nine studies, 21%) [24, 29, 30, 34, 46, 52, 55, 57, 58] and those enrolling both males and females but with higher male enrolments. Thirty-four (79.1%) of the studies we reviewed excluded participants with depression or psychotic disorders, or those taking an antidepressant or antipsychotic medication.
Some authors overstated conclusions; for example, recommending treatment uptake despite limited sample sizes, lack of placebo and/or low completion rates. The studies were overwhelmingly government or academic funded (65.1%, n = 28) [24,25,26,27,28, 30, 32, 33, 35,36,37,38,39,40,41,42, 44, 47, 50, 51, 54, 55, 57, 58, 60, 63, 64, 66]. Ten studies (23.3%) were funded by pharmaceutical companies, or the study drug(s) were provided by a pharmaceutical company, or a mix of funding and drugs were provided by a pharmaceutical company [29, 31, 45, 46, 48, 49, 56, 59, 61, 65]. Four studies (9.3%) did not state their funding source [43, 52, 53, 62], and one study (2.3%) received no funding .
Outcomes and Measures
In total, 55 primary outcome measures were used (inclusive of variations) 93 times (as some studies had multiple primary outcomes). The most common primary outcome measure reported was abstinence (51 times, 55%), followed by cravings (10 times, 11%). For abstinence, urine drug screens (UDS) were used 41 times (80%) and analysed or defined in 16 different ways. The most common method for analysing UDS was weekly proportion of AMPH/MA-free UDS, or overall proportion of AMPH/MA-free UDS. There were 75 distinct secondary outcomes inclusive of variations and often analysed differently to the primary outcomes of the same domain. These were used 158 times. The most common secondary outcome measure reported was craving (25 times), predominantly reported using the visual analogue scale (VAS) (16 times, 64% of the cravings measures). The frequency with which each measure was used is noted in Table 5.
Adherence as an outcome was measured by participant self-report; pill count (i.e. total pills taken divided by total prescribed multiplied by 100 to provide a percentage of adherence); medication electronic monitoring systems (MEMS, e.g. an electronic monitor in bottle caps); proportion of study staff-administered doses received; or measures of metabolites/study drug in plasma. Eleven studies (25.6%) did not report adherence in methods/results. In studies reporting both self-report and another measure, there was low concordance between results. For example, in one study self-reported adherence was 93% but ad-hoc analysis of study drug/metabolite in urine results of participants randomised to the study drug group were presented in quartiles, with the top quartile achieving > 85% positive urines while the bottom quartile showed ≤ 40% positive urines . Adherence reported by both self-report and MEMS caps demonstrated non-concordance in the two studies reporting both—one study reported adherence assessed by MEMS caps was 42% as compared with 74% by self-report , another reported 48.5% versus 74.7% . No study reporting plasma metabolite/study drug reported a marker of adherence for placebo. Adherence rates ranged from 21%  to 100%  across studies. Full data for each study are available in the Supplementary Data (see ESM).
Results of Individual Studies
One study (2%) examined amineptine [300 mg oral (po) daily (OD)], an atypical tricyclic antidepressant, in inpatient participants for AMPH withdrawal over 14 days . Participants randomised to amineptine were significantly less depressed at Day 7 and had improved clinical global impression scores at Day 14 in the completer analysis (i.e. only those completing study protocol) compared with placebo. However, the sample size was small (29 analysed). In terms of feasibility, amineptine has never been approved by the US Federal Drug Administration (FDA) and has been suspended in other jurisdictions due to hepatotoxic effects and abuse liability.
Mirtazapine was examined in three studies (7%). Mirtazapine’s effects on withdrawal symptoms were reported on twice, yielding conflicting results. In a 2005 study examining mirtazapine (15–60 mg po OD) in AMPH withdrawal , Amphetamine Withdrawal Questionnaire (AWQ)  scores between baseline and Days 3 and 14 demonstrated significant differences favouring mirtazapine over placebo, but the study only enrolled 20 participants and the number analysed was unclear. In contrast, with a primary outcome of retention, a 2008 study of mirtazapine (30 mg po OD) for the treatment of MA withdrawal  demonstrated no difference in retention rates, or the secondary outcome of MA withdrawal symptoms. The third study  aimed to reduce MA use among MA-dependent sexually active men who have sex with men. The proportion of MA-positive UDS was significantly reduced in both study arms over time but was more pronounced and quicker in the mirtazapine (30 mg po OD) arm compared with the control arm. Participants randomised to the mirtazapine arm also reduced their high-risk sexual behaviours (based on a questionnaire), leading the authors to conclude that mirtazapine decreased both MA use and high-risk sexual behaviours in this population, despite fairly low adherence rates by MEMS caps and self-report (< 50%).
Bupropion was examined in six studies (14%) [26, 33, 39, 41, 60, 66]; four reported on AMPH/MA abstinence as the primary outcome, and two on reduction of AMPH/MA use. None of the six studies achieved a statistically significant difference in abstinence or reduction in use between the bupropion and placebo arm in planned primary outcome analyses.
In one study , a post hoc analysis found a statistically significant effect for bupropion (150 mg po twice daily [BD]) as compared with placebo when the sample was stratified by ‘baseline light-MA consumers’ (0–2 MA-positive UDS in 2-week baseline period) versus ‘baseline-heavy MA consumers’ (3–6 MA-positive UDS in 2-week baseline period). Among ‘baseline light-MA consumers’, the probability of achieving an MA-free week was significantly higher in the bupropion arm as compared with placebo (odds ratio [OR] of 2.8, p < 0.0001), but there was no statistically significant difference between bupropion and placebo in ‘baseline-heavy MA consumers’. A similar planned sub-group analysis in another study of bupropion (150 mg po BD) for the treatment of MA dependence  demonstrated that the sub-group of participants with ≤ 18 days’ MA use in the 30 days prior to baseline who were randomised to bupropion had an increase in weekly periods of MA abstinence as compared with placebo. In additional subgroup analysis, the male-only participants randomised to bupropion also demonstrated a higher proportion of MA-free weeks as compared to placebo . Further analysis determined that two subgroups were significantly more likely to have an MA-free week: male participants with low baseline use (OR 1.39 and OR 1.34; p ≤ 0.001) who were randomised to bupropion; and non-depressed female participants with low baseline use (OR 1.27; p = 0.02) who were randomised to bupropion.
Given bupropion’s licensed indication as a smoking cessation aid, unsurprisingly in one study examining the effects of bupropion on both smoking and stimulant use, participants randomised to bupropion were more likely to reduce their smoking compared with placebo .
One study (2%) examined sertraline (50 mg po BD), along with CM for the treatment of MA dependence over 14 weeks . The four study arms were sertraline only, sertraline and CM, placebo only, placebo and CM. There was no statistically significant main or interaction effect of either sertraline or CM for measures of MA use. In fact, those in the sertraline-only arm were significantly less likely to achieve 3-week abstinence and significantly more likely to have an MA-positive UDS throughout the study compared with other study arms.
One study examined imipramine (150 mg po OD) for the treatment of cocaine and MA dependence . The primary outcome was retention in care, and the survival analysis indicated that higher doses of imipramine were associated with enhanced retention in treatment which in turn was associated with more frequent clinical contacts. There were no statistically significant differences between arms on secondary outcomes of time since last MA use, cravings or depression. Notably, 32 of the 183 participants (17.4%) were MA dependent, the rest were cocaine dependent. This study was not placebo controlled, control participants were provided with a reduced dose of imipramine (10 mg versus 150 mg daily), primarily to increase blinding efficacy and acceptability by staff and participants.
A study of atomoxetine (80 mg po OD), a selective norepinephrine (noradrenaline) reuptake inhibitor (SNRI), randomised 69 opioid and ATS-dependent participants to 16 weeks of treatment, assessing ATS use as the primary outcome . The proportion of ATS-negative UDS was higher in the atomoxetine arm compared with placebo, but achieved only a small effect size, while there was no statistically significant difference in days abstinent. For secondary outcomes, the proportion of morphine-negative UDS was lower for the atomoxetine arm, while the depression scores were significantly reduced in the atomoxetine arm compared with placebo .
Aripiprazole was assessed in two studies on its own [28, 62], and once in a study that also had a second active agent arm—methylphenidate . Aripiprazole (20 mg po OD) for the treatment of MA dependence was no more effective than placebo in reducing MA consumption or reduction in MA-positive UDS . In another study of the treatment of MA dependence and associated psychoses, participants randomised to aripiprazole (5–10 mg po OD) demonstrated no significant difference in abstinence compared with placebo . However, those randomised to the aripiprazole arm were more likely to be retained in treatment, and to demonstrate a decrease on symptom scores for psychopathology on the Positive and Negative Symptoms Scale (PANSS) .
A study investigating aripiprazole versus methylphenidate for AMPH dependence (primary outcome AMPH use, secondary outcomes change in use and retention) enrolled three study arms: aripiprazole 15 mg daily, methylphenidate 54 mg daily or placebo . The aripiprazole study arm had significantly more AMPH-positive UDS compared with either the methylphenidate or placebo arms. The study was ceased early due to this finding at interim analysis. The analysis also demonstrated that participants randomised to the methylphenidate arm returned significantly fewer AMPH-positive UDS than placebo. In addition, two participants (4% of randomised participants) in the aripiprazole arm discontinued the study due to adverse events.
Topiramate was investigated in two studies reviewed here. One study examined topiramate (200 mg po OD) in MA-dependent adults (n = 140 randomised, 77 completed study) with the primary outcome being MA abstinence at Week 12 . While there was no statistically significant difference between topiramate and placebo on the primary outcome, a higher proportion of participants randomised to topiramate reduced their MA use compared with placebo.
In a separate study, topiramate (200 mg po OD) for MA dependence was examined in 62 males and analysed in 57, who were all on prescribed methadone for opiate replacement therapy , with the outcomes of interest being dependence severity, cravings, depression and MA use. There was a statistically significant difference between groups on the Addiction Severity Index (drug use severity and drug need domains) that favoured the topiramate arm; however, there was no statistically significant difference in cravings or depression symptoms between the study groups. Participants randomised to topiramate returned significantly fewer MA-positive UDS at Week 6, but this result was not sustained throughout the final 4 weeks of the treatment period .
Central Nervous System Stimulants
Two studies reviewed examined dexamphetamine as stimulant agonist treatment. The first study reviewed 49 participants with MA dependence and prescribed 110 mg daily sustained-release oral dexamphetamine over 16 weeks. It measured MA use by self-report and analysis of hair, severity of dependence over time and treatment retention—finding no statistically significant difference between the study groups on planned analysis. Post-hoc analysis demonstrated a reduction in MA dependence symptoms in the dexamphetamine arm compared with placebo using the Leeds Dependence Questionnaire . Secondary analysis included withdrawal symptoms. The participants randomised to dexamphetamine demonstrated a greater reduction in withdrawal severity compared with placebo ; however, participants were outpatients and continued MA use complicates the interpretation of withdrawal scores.
Another study examined sustained-release oral dexamphetamine (30 mg po BD) for 60 MA-dependent participants . The primary outcomes included safety and efficacy defined as abstinence from MA—measured by a new MA-positive UDS (measured twice weekly) and self-reported MA consumption. There was no significant difference between study groups on measures of MA consumption; however, the participants randomised to dexamphetamine reported significantly reduced MA withdrawal severity and cravings compared with placebo , although these were secondary outcome measures and again in the context of continued MA use by outpatient participants.
Three studies examined sustained/extended-release oral methylphenidate in addition to the study reporting methylphenidate versus aripiprazole discussed earlier. All three were in the outpatient setting and used the same dose (54 mg po OD).
The first randomised 79 MA/AMPH-dependent participants for 22 weeks to methylphenidate or placebo, with abstinence (measured by twice-weekly UDS, and defined as the weekly percentage of AMPH/MA-positive results) as the primary outcome . Twenty-seven participants (34%) completed the study. In intention-to-treat analysis there were no differences in abstinence or study retention rates (defined by number of doses collected), although the methylphenidate arm achieved higher study retention from Week 6. The sample was heterogeneous, as participants were enrolled in both Finland, where all participants took intravenous AMPH, and New Zealand, where all participants smoked MA, but the results were analysed in aggregate. There was no concomitant psychosocial therapy.
Another study enrolled 110 MA-dependent participants in the USA with active study drug for 10 weeks followed by 4 weeks of blinded placebo treatment to encourage follow-up . Participants received weekly CBT and CM. There was no difference between study groups in self-reported MA use in planned analysis of the final 30 days of treatment; however, a secondary analysis of data from baseline to Week 10 found there were significantly fewer self-reported MA use episodes in the methylphenidate arm than placebo.
The final study enrolled 56 Iranian MA-dependent participants for 10 weeks of treatment examining craving as the primary outcome . At Week 10 of the study there was a reduction in craving in the treatment arm, and the treatment arm demonstrated fewer positive UDS and reduced depressive symptoms at Week 10 compared with the placebo arm.
Other Central Nervous System Agents
Modafinil was examined in four studies reviewed here, in doses of 200–400 mg daily. Three were conducted in outpatient settings [25, 38, 59] and one in an inpatient withdrawal setting . One pilot withdrawal study examined feasibility and withdrawal symptoms in 19 participants prescribed 7 days of modafinil (200 mg po OD Days 1–5 and 100 mg po OD Days 6–7) versus placebo. There were no differences between study arms in retention or withdrawal symptoms . None of the other three studies demonstrated a difference in MA use, adherence or retention between study arms. One study analysed a subset of participants with the greatest adherence (> 85%) to study treatment compared with other study participants randomised to modafinil and observed greater abstinence from MA in the > 85% adherent participants; however, the comparison did not include the placebo group and external confounders were not identified or controlled for .
GABA Agonist/GABAergic Agents
One 16-week outpatient study of 88 MA-dependent participants examined baclofen (20 mg OD three times daily [TDS]) and gabapentin (800 mg OD TDS) for their effects on MA use . Secondary outcomes included treatment retention, depression, cravings and adverse events. No differences were observed between study arms. Post-hoc analysis demonstrated higher probability of MA-negative urines was associated with medication adherence in all arms, higher in the treatment arm.
Two Iranian studies reviewed examined opioid agonists, one buprenorphine  and one buprenorphine and methadone . Both studies were in MA-dependent inpatient males with no co-occurring substance use disorder. In one 16-week study of 40 participants with concomitant psychosocial therapy (Matrix model), reduction in MA cravings and fewer MA-positive UDS were demonstrated among the buprenorphine (6 mg sublingual [SL, i.e. applied under the tongue] OD) arm as compared with the placebo arm during the treatment phase, trending back to baseline following cessation of medication . The second study examined buprenorphine (8 mg SL OD) versus methadone (40 mg po OD) over 17 days, with 20 participants in each study arm. There was a reduction in MA craving compared with placebo, and no participants produced MA-positive UDS in the study period, but the setting was a controlled inpatient environment .
Five studies examined the opioid antagonist naltrexone, including two that used an extended-release formulation [29, 56] and one that used an implant . An additional study reported on naltrexone and n-acetyl cysteine (see below).
Results of the studies are conflicting. There was no difference in MA use by UDS in the treatment arm compared with placebo in the extended-release studies [29, 56]. One study of naltrexone (a single 4-week injection) reported on 37 of 52 randomised participants and found a reduction in past 30-day MA use, but relied entirely on self-report , and there was a crossover in primary outcome measures given the past 30-day questionnaires were administered within 3 weeks of each other. One outpatient study of AMPH-dependent participants in Sweden reported fewer AMPH-positive UDS in the naltrexone (50 mg po OD) arm compared with placebo , a result shared by the study examining naltrexone implants (1000 mg subcutaneously) administered to Russian participants with AMPH dependence .
A single study has investigated ondansetron for the treatment of MA dependence . This four-arm trial assessed different doses of ondansetron (0.5 mg, 2 mg, 8 mg po OD) against placebo in measures of abstinence, use, severity of dependence, withdrawal, craving and retention in treatment. There was no observable difference in any outcome measure between doses or against placebo. The authors suggest that the nil result may be due to the short half-life of ondansetron (approximately 5 h) and suggest a sustained-release formulation or more aggressive dosing may give more efficacious results. At the time of this review no follow-up studies had been conducted.
Partial Cholinergic Nicotinic Agonists
A single, recent American study assessed varenicline (1 mg po BD) as a pharmacotherapy for MA dependence . There were no differences between treatment and placebo arms for any measures of dependence; however, there was a reduction in cigarettes smoked in the treatment arm (consistent with its licensed indication as a smoking cessation medication).
Glutamatergic agents have been assessed, as either riluzole , N-acetyl cysteine (NAC) , or a combination of NAC and naltrexone . A recent trial of Iranian men found that riluzole (50 mg po OD) was associated with higher rates of retention in treatment and abstinence at Week 12, as well as overall improvements in a range of secondary outcomes . Efficacy for NAC is conflicted, however. When trialled as a combination of NAC (escalating dose to 2400 mg po OD) with naltrexone (escalating dose to 200 mg po OD), there was no difference between arms in measures of craving, use or psychological scales . Mousavi et al., however, found that NAC treatment (escalating dose to 1200 mg po OD) was associated with a reduction in craving. This paper did not report on secondary outcomes . Both studies had low participant numbers (n = 31 and n = 32, respectively). Mousavi et al.  allowed weekly ‘matrix model’ psychological therapy to all participants, while Grant et al.  did not provide any psychosocial support, and this may explain the discrepancies in results.
We reviewed pexacerfont in one study, a 3-week trial of 51 Iranian men within residential treatment camps where treatment is not normally provided . Dosing was tapered from 300 mg po OD for the first week, to 200 mg po OD in Week 2 and 100 mg po OD in Week 3. While measures of craving reduced significantly more in the treatment arm than placebo, there was no difference in end-of-treatment abstinence between groups. Additionally, levels of temptation and depression, but not anxiety, withdrawal severity, or treatment effectiveness, improved favouring treatment.
Benzodiazepine Antagonist/GABA Agonist/H1 Histamine Receptor
The combination therapy of flumazenil (2 mg intravenous Days 1, 2, 3, 21, 22), gabapentin (titrated up to 1200 mg po OD) and hydroxyzine (50 mg po pre-intravenous medication and as required for sleep), marketed and trademarked as the ‘PROMETA protocol’, has been assessed twice in RCTs for MA dependence. Both trials were similar in terms of participant numbers and followed an identical medication protocol; however, results were conflicting. A 30-day trial found significantly improved craving scores, but no difference in use (missing UDS imputed as positive) . However, a 40-day trial conducted the same year found no differences in any measures, including craving .