Psilocybin-assisted therapy for relapse prevention in alcohol use disorder: a phase 2 randomized clinical trial

Harmful alcohol use remains a leading global health risk, and alcohol use disorder (AUD) is typified by repeated cycles of abstinence, withdrawal and relapse. Existing pharmacotherapies such as naltrexone and acamprosate reduce drinking in some patients but relapse rates remain high, with at least 60% of treated individuals returning to drinking within six months in typical samples. Psychedelic compounds, notably psilocybin, have re-emerged in psychiatric research because of early clinical signals and more recent trials suggesting reductions in heavy drinking and improvements in mood when combined with psychotherapy, but no randomized controlled trial (RCT) at current methodological standards had tested psilocybin specifically for relapse prevention after withdrawal treatment in AUD. Rieser and colleagues designed this Phase II RCT to test whether a single oral 25 mg dose of psilocybin, delivered with a brief psychotherapeutic regimen, could prolong abstinence and reduce alcohol consumption in patients who had completed withdrawal treatment. The primary outcomes were duration of abstinence and mean alcohol use during the 4-week period after dosing; secondary outcomes included craving, self-efficacy, psychopathology measures and safety. The trial therefore addresses a distinct clinical question: can a single psilocybin-assisted treatment session prevent relapse in recently withdrawn, moderately to severely affected AUD patients?

This single-centre, double-blind, parallel-group RCT was conducted at the Psychiatric University Hospital Zurich. Eligible German-speaking participants aged 18–60 met DSM-5 criteria for AUD (assessed with the MINI) and had completed inpatient, outpatient or autonomous withdrawal treatment within six weeks before enrolment. The extracted text contains an inconsistency in the allocation numbers: one section reports 18 participants receiving psilocybin and 19 placebo, while the Results report 19 psilocybin and 18 placebo; the precise randomised allocation at enrolment is therefore unclear from the provided extraction. Exclusion criteria included major psychiatric or other substance use disorders (except tobacco), substantial recent hallucinogen use, suicidal history, unstable medical conditions and family history of certain psychotic or bipolar disorders; full criteria are said to be in the protocol supplement. Participants were randomised 1:1 by the hospital pharmacy to receive either a single oral dose of 25 mg psilocybin or placebo (mannitol) in identical capsules. Stratified randomisation accounted for age (≤45 v >45 years), sex and AUD severity. Blinding integrity was assessed by asking participants and treatment providers to guess allocation at the 1-day follow-up. The investigational drug was administered during an 8 h outpatient session under two staff members (one psychiatrist), with an MRI scan performed 1.5 h post-dose for 30 min (imaging results reported elsewhere). Psychotherapy comprised two preparation visits, the dosing session embedded in semi-structured sessions based on the BRENDA approach (incorporating motivational interviewing and cognitive behavioural principles, adapted for psychedelic-assisted therapy), and three integration visits; total psychotherapeutic contact was approximately 5.5 hours, plus two online surveys at 3 and 6 months. Primary alcohol outcomes used the timeline followback (TLFB) to quantify abstinent days and mean standard alcohol units (SU) per day; relapse was defined as drinking >1 SU/day. Self-report was corroborated by urine ethylglucuronide and blood biomarkers (ASAT/GOT, ALAT/GPT, Gamma-GT). Secondary measures included craving (PACS), alcohol self-efficacy (AASE), depression (BDI), hopelessness (HS), emotion regulation (ERQ), affect (PANAS), quality of life, anhedonia (SHAPS, TEPS), psychopathology (SCL-90-R) and subjective drug effects (5D-ASC). Vital signs were monitored intensively during dosing and (serious) adverse events ((S)AEs) were recorded at each visit. The target sample size was 60 (power 95%, two-tailed alpha 0.05) based on prior effect estimates, but recruitment was curtailed by the Covid-19 pandemic and the study stopped early. Statistical analyses used R: normality checks guided choice of tests, Kaplan–Meier survival analysis and Cox regression evaluated time to relapse, Mann–Whitney U or t-tests compared groups for primary outcomes, Wilcoxon tests compared blood markers, and linear mixed models (random intercepts) examined longitudinal secondary outcomes. For craving, a dummy-coded time variable separated the 1-day post-dose assessment from other timepoints. Model selection used the Bayesian Information Criterion (BIC). An intention-to-treat (ITT) analysis was also reported for primary outcomes.

Between June 2020 and August 2023, 65 participants consented and 46 were enrolled; after exclusions and dropouts 37 participants provided TLFB data up to the 4-week primary endpoint and were included in the main analysis. The available 6-month analysis included 35 participants. As noted above, the extraction contains conflicting statements about whether 18 or 19 participants received psilocybin; the Results present analyses with psilocybin and placebo groups of roughly 18–19 participants each. Primary outcomes: Using TLFB data for the 28 days after dosing (N = 37), no significant difference in duration of abstinence was observed between groups (Kaplan–Meier p = 0.55). Mean abstinent days were reported as psilocybin mean = 16.8 days (95% CI: 14.31–19.29) versus placebo mean = 13.8 days (95% CI: 10.97–16.63). Cox regression yielded a hazard ratio of 0.78 (95% CI: 0.34–1.76) for psilocybin versus placebo, which the authors interpret as a 22% lower likelihood of relapse in the psilocybin group, but they caution that the estimate is imprecise due to limited sample size. The percentage remaining abstinent over 4 weeks was similar: 39% in the psilocybin group and 37% in the placebo group. Total abstinent days and other abstinence metrics also showed no significant group differences (e.g. duration of abstinence p = 0.37, Cohen's d = 0.151; total abstinent days p = 0.76, Cohen's d = 0.05). Mean alcohol consumption per day between dosing and 4-week follow-up did not differ significantly (Mann–Whitney p = 0.51, Cohen's d = 0.111). Median SU/day at 4 weeks was 0.48 (range 0–3.99) in the psilocybin group and 0.54 (range 0–5.96) in the placebo group. Analyses extended to 6 months likewise found no group differences in alcohol use measures. An ITT analysis including all randomised participants (n = 38 with available TLFB to 2 weeks for one additional participant) produced consistent null findings for duration of abstinence, total abstinent days and mean alcohol use per day. Secondary and exploratory outcomes: Craving (PACS) showed a strong, acute reduction at 1 day post-dose in both groups (random coefficient model β = -4.12, 95% CI -6.36 to -1.87, p = 0.00037) and an additional significant reduction attributable to psilocybin (β = -5.40, 95% CI -8.62 to -2.19, p = 0.0011). AASE temptation scores declined in both groups (β = -7.16, 95% CI -10.66 to -3.67, p < 0.0001) with a further decrease in the psilocybin group (β = -7.71, 95% CI -11.71 to -1.70, p = 0.0089); confidence to abstain did not change significantly pre to post dosing. Participants receiving psilocybin also reported improvements on several clinical measures including depressive symptoms (BDI), hopelessness, negative affect, emotion suppression and quality of life; the extraction notes a decline in depressive symptoms below remission thresholds in the psilocybin group but does not provide exact scores or effect sizes in the main text. Blood biomarkers of alcohol use (ASAT/GOT, ALAT/GPT, Gamma-GT) did not differ between groups at screening or 4 weeks. Safety, blinding and other analyses: The 25 mg psilocybin dose was described as well tolerated; comprehensive (S)AE data are reported in the supplement. Blinding was strongly compromised: participants correctly guessed their treatment in 91.89% of cases and therapists in 97.30% of cases. An exploratory subgroup analysis suggested that, among participants without prior psychedelic experience, those allocated to psilocybin (n = 8) had longer abstinence and lower mean alcohol use at 4 weeks than those with prior psychedelic use and, when compared with the placebo subgroup without prior psychedelic use (n = 10), showed significant differences; the authors emphasise that these subgroup analyses were underpowered and exploratory. Overall, the extracted results indicate no clinically meaningful impacts of a single-dose psilocybin regimen plus brief psychotherapy on abstinence or alcohol consumption after withdrawal in this sample.

Rieser and colleagues interpret their findings as a null result for the trial's primary question: a single 25 mg dose of psilocybin combined with approximately 5.5 hours of psychotherapy did not significantly prolong abstinence or reduce alcohol use compared with placebo in patients who had recently completed withdrawal treatment. Both treatment arms experienced acute reductions in craving and temptation immediately after the dosing visit, with an additional short-term benefit on craving and temptation in the psilocybin group. The psilocybin group also showed improvements on several secondary clinical measures, including depressive symptoms, hopelessness, negative affect and quality of life, but these changes did not translate into reduced drinking. The investigators situate their results against prior work that found reductions in heavy drinking after two psilocybin doses plus a substantially larger psychotherapeutic package; they note important design differences that may explain discrepant outcomes. Key distinctions include the study population (this trial required abstinence at baseline and included many participants who had undergone medically supported withdrawal), the single-dose versus two-dose schedules, the briefer psychotherapy provided here, and higher baseline alcohol use in this sample. The authors also reference preclinical and animal data suggesting dose- and severity-dependent effects, and they interpret their findings as consistent with studies showing limited relapse-prevention effects of psychedelics after an initial abstinence phase in some animal models. Several limitations acknowledged by the investigators temper the conclusions. Recruitment fell short of the planned sample size because of Covid-19 disruptions, leaving the trial underpowered to detect smaller effects and producing imprecise effect estimates. Blinding was largely ineffective, with most participants and therapists correctly guessing allocation, which may have biased subjective outcomes. Primary alcohol outcomes relied on self-report, although urine ethylglucuronide and blood markers provided objective corroboration. The sample was predominantly Caucasian and therefore may not generalise widely. The adjunct 30-minute MRI during the acute drug session could have interfered with therapeutic processes, although participants still reported substantial subjective effects. The exploratory finding that prior psychedelic experience may moderate response is noted but treated cautiously because of small subgroup sizes. In terms of implications, the authors suggest that a single psilocybin administration with limited psychotherapeutic support may be insufficient for relapse prevention in moderately to severely affected AUD patients. They propose that future studies should examine alternative strategies such as repeated dosing, more intensive or longer-term psychotherapy, or adjunctive techniques (for example neurofeedback or virtual reality), and that larger, better-powered trials with diverse populations are needed to evaluate efficacy and safety more definitively.