Psilocybin-assisted psychotherapy for treatment resistant depression: A randomized clinical trial evaluating repeated doses of psilocybin

Over the past two decades, interest in serotonergic psychedelics has resurged, with psilocybin showing promise for rapid and robust antidepressant effects. Previous clinical trials of psilocybin-assisted psychotherapy (PAP) have reported preliminary efficacy in major depressive disorder (MDD), treatment-resistant depression (TRD) and in patients with life-limiting illness, but have commonly used small samples, intensive psychotherapy protocols, and strict eligibility criteria that exclude many real-world, complex patients. These design features raise concerns about generalisability, potential expectancy effects and functional unblinding, and leave open questions about the durability of benefit and whether repeated dosing is helpful when relapse occurs. This randomised trial set out to address those gaps by evaluating feasibility, preliminary efficacy and safety of a briefer PAP model in a more clinically complex TRD sample. Specifically, Rosenblat and colleagues conducted a waiting-list randomised clinical trial that broadened eligibility (including both MDD and bipolar II disorder, comorbid personality disorders, baseline suicidality and no upper limit on prior failed treatments) and permitted repeat psilocybin doses for participants showing signs of relapse. The study therefore aimed to test whether a condensed psychotherapy “dose” and flexible repeat psilocybin administration could be delivered safely and produce antidepressant effects in a real-world refractory population.

This was an open-label, randomised, waiting-list controlled clinical trial conducted at a single community clinic in Mississauga, Ontario (Braxia Health). Blocked randomisation (permuted blocks of ten) allocated participants to immediate treatment or a two-week delayed-treatment waitlist; there was no blinding of participants, therapists or investigators. The trial was registered (NCT05029466) and approved by a community IRB. Rosenblat and colleagues describe the design as prioritising real-world applicability by abbreviating psychotherapy and allowing repeat dosing when clinically indicated. Eligible participants were adults aged 18–75 years with a primary diagnosis of MDD or bipolar II disorder and a current major depressive episode of at least three months’ duration. Treatment resistance was required (failure of at least two adequate pharmacological trials for the current episode) with no upper limit on the number of failed trials. Key exclusions included recent moderate-severe substance use disorder, recent use of classic hallucinogens, uncontrolled medical conditions and pregnancy; bipolar I disorder was excluded after regulatory feedback. Participants were required to be off antidepressants, antipsychotics and other interfering psychotropic medications for at least five half-lives prior to screening and throughout follow-up, though BDII participants could remain on conventional mood stabilisers at clinical discretion. Intervention sessions used synthetic psilocybin, 25 mg dissolved in water for each dosing session. Each dose was paired with a brief preparatory psychotherapy session (1–2 hours), an on-site supervised dosing session (6–8 hours) and two integration sessions (1–2 hours each), amounting to approximately 4.5 hours of psychotherapy per dose outside the dosing session. Therapists worked in licensed multidisciplinary dyads and used a transtheoretical model adapted from prior psilocybin trials; a professional chaplain was included to support spiritual care when relevant. Participants could receive one, two or three dosing sessions: eligibility for repeat dosing required prior clinical benefit, adequate tolerability and at least two weeks of depressive relapse, as judged by the study clinician. Primary outcomes were feasibility criteria prespecified for the trial: (1) <30% all-cause dropout before the week-2 primary endpoint, (2) zero or minimal worsening of baseline suicidality post-dose, (3) <15% experiencing serious adverse events (SAEs), and (4) >80% of adverse events resolving within 48 hours. A secondary, preliminary efficacy comparison used the clinician-rated Montgomery–Åsberg Depression Rating Scale (MADRS), comparing mean change from baseline to week 2 in the immediate treatment arm versus the delayed arm during their waitlist period. Statistical analysis was descriptive for feasibility; efficacy comparisons used two-tailed t-tests with significance at p<0.05 and effect sizes reported as Hedges’ g, which adjusts for small samples. No formal power calculation was performed; the planned recruitment target was 15 participants per arm to achieve approximately 12 completers per arm for feasibility assessment. The authors note that raw data are not publicly available due to ethics constraints, although summary statistics may be provided on request.

Between November 1, 2021 and February 1, 2023, 135 patients were assessed for eligibility; 31 were enrolled and randomised. One participant withdrew prior to receiving the intervention due to difficulties tapering antidepressant medication, resulting in 16 participants assigned to immediate treatment and 14 to delayed treatment. One further participant dropped out before the week-2 primary endpoint, leaving 29 participants in the primary outcome analysis. Over the 6-month follow-up period, eight additional participants left to pursue other treatments, and 21 participants completed the entire follow-up. The sample had chronic, highly treatment-resistant depression: mean age 44.4 years (SD 13.7), mean depressive episode duration 18.3 years (SD 11.5), and a mean of 11.27 failed medication trials (SD 5.59). Baseline mean MADRS score was 30.5 (severe range), with the immediate treatment group more severe by about 5.4 points. Diagnostic composition was 26 participants with MDD and four with bipolar II disorder. All participants had at least one psychiatric comorbidity; 12 of 30 (40%) had previously failed more intensive interventions such as electroconvulsive therapy or ketamine. All four prespecified feasibility criteria were met: only 1/30 (3%) dropout prior to the week-2 primary endpoint, two participants experienced transient worsening of suicidality lasting 24–48 hours post-dose without need for further intervention, zero SAEs were reported, and all but one adverse event resolved within 48 hours. The one persistent treatment-emergent adverse event was persistent genital arousal that lasted throughout the 6-month follow-up period and, according to the authors, has not been previously reported with psilocybin. On the primary clinical comparison, mean (SE) change in MADRS during the waitlist period was −3.0 (0.9) versus −9.6 (1.9) in the immediate treatment arm, a between-group difference of 6.6 points. This corresponded to a Hedges' g of 1.1 (95% CI 0.3–1.8; p = 0.005). Open-label longitudinal data across both arms showed sustained reductions in MADRS scores over time; 17 participants received a second psilocybin dose and five received a third dose. The authors report that cumulative reductions in mean MADRS (measured 2 weeks after the most recent dose compared to baseline) were greater after successive doses. Self-reported depressive symptoms (QIDS-SR) decreased in parallel with MADRS findings, whereas self-reported anxiety (GAD-7) showed only small, short-lived improvement returning close to baseline by two months. There was no statistically significant change on the MADRS suicidal ideation item from baseline to the primary endpoint. Adverse events were numerous but largely mild to moderate and transient, resolving within 48 hours except for the single persistent event noted above.

Rosenblat and colleagues interpret their findings as demonstrating feasibility for conducting PAP in a more clinically complex TRD population and for using a briefer psychotherapy protocol. The brief model comprised one preparatory session and two integration sessions per psilocybin dose, and the study also established feasibility for providing repeat doses when clinical relapse occurred. Although prespecified as a feasibility study, the investigators observed descriptive antidepressant effects: clinician-rated (MADRS) and self-reported (QIDS-SR) measures showed aligned improvements, while anxiety reductions were modest and not durable. The authors situate their results in relation to prior trials. Two earlier trials have specifically addressed PAP in TRD: an open-label single-arm study (n = 20) with large effect sizes after two dosing sessions, and a larger phase 2b RCT (Goodwin et al.) in which a single 25 mg dose produced a 6.6-point greater reduction in MADRS versus 1 mg at the 3-week endpoint. The present trial produced a between-group MADRS difference of 6.6 points as well, but absolute reductions were smaller than in Goodwin et al., which the authors attribute plausibly to the greater treatment resistance and diagnostic complexity of their sample. Indeed, the investigators note that 29 of 30 participants would have been excluded from the Goodwin et al. TRD trial, emphasising the present study’s greater real-world representativeness. Safety observations included no treatment-emergent mania, hypomania or psychosis among the four BDII participants, and no additional safety signals identified in participants with comorbid personality disorders (n = 9); however, the authors caution that these subgroup sizes are too small to draw firm safety conclusions. Repeat dosing appeared to yield incremental reductions in depressive symptoms without new safety concerns, but the authors acknowledge indication bias because participants selected for additional doses were those judged clinically likely to benefit. Key limitations are enumerated: the open-label design, small sample size, use of a waitlist rather than a placebo control, sample heterogeneity and lack of measurement of participant expectancy. The authors also highlight that three intentional design modifications—broader inclusion criteria, abbreviated psychotherapy and flexible repeat dosing—make it difficult to attribute observed effects to any single element. Given these constraints, the study cannot support regulatory claims about efficacy, but the authors argue that the data justify further adequately powered RCTs to evaluate safety, optimal dosing frequency and the generalisability of PAP to complex, real-world patient populations.