Long-term outcomes of single-dose psilocybin for U.S. military Veterans with severe treatment-resistant depression - 12-month data from an open-label pilot study

Depression is highly prevalent among US military Veterans and a substantial minority develop treatment-resistant depression (TRD), a condition associated with greater comorbidity, healthcare utilisation, and poorer long-term outcomes than non-TRD major depressive disorder. Previous clinical research has shown promising short-term antidepressant effects of psilocybin, a 5-HT2A receptor agonist derived from Psilocybe mushrooms, but few studies have reported outcomes beyond 12 weeks. Mechanistic work has emphasised potential neuroplasticity-related effects of psilocybin, yet the durability of antidepressant benefit—especially in people with severe, chronic, and treatment-refractory depression—remains poorly characterised. Ellis and colleagues designed this open-label pilot to address that gap by extending follow-up of Veterans with severe TRD who received a single 25 mg dose of pharmaceutical-grade psilocybin (COMP360) and manualised preparatory and integration therapy. The study aimed to assess clinician-rated depression severity (MADRS) out to 12 months post-dosing, with secondary measures of self-reported depression (QIDS), functional disability (SDS), acute psychedelic experience (5D-ASC), and safety, and to explore whether the acute experience predicted longer-term antidepressant outcomes.

This was a single-site, open-label pilot trial conducted at the VA Palo Alto Healthcare System and Stanford University. All participants received the active intervention and there was no masking. Fifteen US military Veterans aged 18–65 were enrolled between March 2022 and February 2023; 10 elected to take part in the extended long-term follow-up beyond the 12-week primary period. Severe TRD was defined as failure to respond to ≥5 prior treatments during the current episode or an episode lasting >2 years, with baseline Hamilton Rating Scale for Depression (HRSD) ≥18. Exclusion criteria included psychotic or bipolar disorders, personality disorder, current substance use disorders, unstable medical illness, and imminent suicidal risk. Prior to dosing, participants discontinued antidepressant or serotonergic medications at least five half-lives beforehand and completed preparatory psychotherapy (two 60-minute and one 90-minute sessions). The investigational agent was COMP360, a proprietary synthetic psilocybin formulation. Each participant received a single 25 mg oral dose administered by the study psychiatrist in a 6–8 hour supervised session with two licensed therapists present. Integration therapy included a 90-minute virtual session on Day 1 and at least two further sessions (Week 1 and Week 12), with optional additional integration during the 12-week period. Follow-up assessments occurred on Day 1, Weeks 1, 2, 3, 6, 9, 12, and Months 6, 9, and 12. The prespecified primary outcome was change in clinician-rated depression severity measured by the Montgomery–Åsberg Depression Rating Scale (MADRS) from Baseline to Month 12. Response was defined as ≥50% reduction in MADRS and remission as MADRS ≤10. Secondary outcomes were self-reported depression (QIDS) and functional disability (SDS). The Five-Dimensional Altered States of Consciousness scale (5D-ASC) assessed the acute psychedelic experience at the end of dosing. Safety was evaluated by adverse event monitoring and suicidality assessments (C-SSRS). Statistical analysis used mixed-effects models for repeated measures (MMRM) to handle small sample size and missing longitudinal data; the model included Time, Age, PTSD diagnosis, Gender, and a random intercept for Subject. An autocorrelation-moving average covariance structure was selected based on information criteria. SDS scores were log-transformed to meet model assumptions. Intention-to-treat (ITT) MMRM analyses included all available data from participants in the long-term follow-up, including those who restarted antidepressants. Pairwise comparisons used Bonferroni correction. Pearson and Spearman correlations examined associations between acute experience and MADRS change and between endpoints, and a logistic regression explored predictors of choosing to participate in the long-term follow-up.

Fifteen participants were enrolled; ten completed assessments out to 12 months. Baseline mean MADRS was 35.3 (SD 7.3), indicating severe depressive symptoms. Clinician-rated depression (MADRS) decreased significantly across time (ANOVA F(5,36) = 15.67, p < 0.0001). Mean reductions relative to baseline remained statistically significant at Month 6 (mean change M = -24.59, SE = 3.75), Month 9 (M = -20.84, SE = 3.75), and Month 12 (M = -16.71, SE = 3.75). On average, MADRS scores were 54.23% (SE = 9.46%) lower at Month 12 compared with Baseline. Response and remission rates among the 10 long-term participants were 80% (n = 8) and 50% (n = 5) at Month 6, and declined to 40% (n = 4) and 30% (n = 3) at Month 12, respectively. In ITT MMRM analyses including available data, there were significant overall effects of time (F(5,45) = 14.45, p < 0.0001), age (F(1,3) = 15.87, p = 0.03), and comorbid PTSD (F(4,3) = 14.87, p = 0.02) on MADRS; gender did not have a significant effect. Four participants did not continue past the 12-week visit; among these, three met response criteria and two met remission at Week 12. Self-reported depression (QIDS) also showed significant reductions from Baseline at all long-term endpoints (F(5,34) = 21.56, p < 0.0001), with an average 54.68% (SE = 12.05%) decrease to Month 12. Functional disability (SDS) showed a small overall change (F(5,40) = 4.71, p = 0.002), but SDS scores at Month 6, 9, and 12 were not significantly different from Baseline after multiple-comparison correction; a small increase in SDS between Week 3 and Month 12 was statistically significant (M = 1.22, SE = 0.36, p = 0.03). Exploratory analyses found no significant Pearson correlations between the Day 0 5D-ASC composite score and change in MADRS at Month 6 (R = 0.03, p = 0.94) or Month 12 (R = -0.04, p = 0.92), and no subscale of the 5D-ASC correlated significantly with MADRS change. Logistic regression examining bias in long-term follow-up participation suggested that greater immediate improvement (Baseline to Day 1 MADRS decrease) was associated with increased log-odds of participating in long-term follow-up by approximately 0.16, a relationship that approached but did not reach conventional statistical significance. Spearman correlations indicated that MADRS and QIDS scores were correlated across multiple endpoints. Regarding safety, no serious adverse events related to psilocybin or study participation were reported. Acute adverse events on or immediately after dosing were mild-to-moderate and transient (headache, nausea, brief back pain, transient tinnitus worsening, and temporary psychological distress) and resolved within hours to 24 hours; no increase in suicidality was observed in the immediate period. Five of the ten long-term participants had been tapered from concomitant psychiatric medications prior to dosing; two of those five restarted at least one antidepressant after dosing (one before Week 1, one at Week 9). The extracted text reports no additional adverse events after the originally published 12-week follow-up.

Ellis and colleagues interpret the findings as evidence that a single 25 mg dose of psilocybin, administered with preparatory and integration psychotherapy, can produce clinically meaningful reductions in depressive symptoms that persist to 12 months in a sample of Veterans with severe TRD. They highlight that response and remission rates were substantial at 6 months (80% and 50%) and attenuated by 12 months (40% and 30%), suggesting that antidepressant effects began to wane after about 6 months and more notably after 9 months, although these increases in depression scores did not reach statistical significance in this small sample. The authors position their results alongside prior long-term psilocybin studies in civilian samples, noting differences in dosing regimens, baseline illness severity, and sample sizes that may account for variability in response and remission rates. They also note the absence of a relationship between the intensity of the acute psychedelic experience (5D-ASC) and longer-term MADRS change in this dataset, which aligns with some prior studies but contrasts with others that have reported predictive associations. Mechanistic speculation in the paper references emerging neuroplasticity findings as a plausible explanation for enduring effects, while acknowledging that mechanisms remain to be clarified. Key limitations acknowledged by the investigators include the open-label design, small sample size, lack of a control group and blinding, and potential selection bias among those who opted into the extended follow-up. The authors note a trend—not reaching conventional significance—toward greater early improvement among those who continued into long-term follow-up, raising the possibility that the long-term sample over-represents early responders. Expectancy effects driven by media attention and unmeasured participant expectations are also cited as potential confounds. Finally, the authors recommend that future research use larger, randomised controlled designs, investigate whether multiple dosing or alternative schedules prolong benefit, systematically measure and manage expectations, and further explore predictors and mechanisms of sustained response.

This first-in-kind open-label pilot in Veterans with severe TRD found that a single 25 mg psilocybin session delivered with preparatory and integration therapy was associated with reduced depressive symptoms persisting to 12 months in many participants, though clinical benefits declined for some after 6–9 months. The authors conclude that larger, controlled, and blinded trials are needed to confirm durability, optimise dosing schedules, and determine generalisability to the broader Veteran TRD population.