Psilocybin-Assisted Psychotherapy for Treatment-Resistant Depression in Bipolar II Disorder

Between 1 November 2021 and 1 February 2023, 135 patients were assessed and 31 were enrolled and randomised; one withdrew before receiving the intervention because of difficulties tapering antidepressants, leaving 16 allocated to immediate treatment and 14 to delayed treatment. One further participant dropped out prior to the week-2 primary endpoint, so 29 participants were included in the primary outcome analysis. Over the 6-month follow-up, eight additional participants discontinued to begin other treatments, leaving 21 participants who completed the full follow-up. The sample was clinically severe and highly treatment resistant. Mean age was 44.4 years (SD 13.7) and mean duration of depression was 18.3 years (SD 11.5). The mean number of failed medication trials was 11.27 (SD 5.59), and most participants had failed multiple psychotherapies; 12/30 (40%) had previously failed intensive interventions such as electroconvulsive therapy or ketamine. Baseline mean MADRS was 30.5 (severe range), with the immediate treatment arm more severely affected at baseline by an average of 5.4 MADRS points. Primary diagnoses were mainly MDD (n = 26), with four participants diagnosed with BDII. All participants met criteria for at least one comorbid psychiatric condition; nine participants had comorbid personality disorders. All four predefined feasibility criteria were met. Dropout before the week-2 primary endpoint was 1/30 (3%), two participants experienced transient worsening of suicidality for 24–48 hours post-dosing without requiring further intervention, zero participants experienced SAEs, and all but one adverse event resolved within 48 hours. The single persistent treatment-emergent adverse event was ongoing genital arousal that persisted through the 6-month follow-up and, according to the authors, has not been previously reported with psilocybin. For the primary clinical comparison, change in MADRS from baseline to week 2 differed between groups: the waitlist group had a mean change of −3 (SE 0.9) versus −9.6 (SE 1.9) in the immediate treatment group, yielding a between-group difference of 6.6 MADRS points. This corresponded to a Hedges' g of 1.1 (95% CI 0.3–1.8) and a p value of 0.005. Open-label longitudinal data over the 6-month follow-up showed sustained reductions in MADRS scores; 17 participants received a second psilocybin dose and 5 participants received a third dose. The cumulative reduction in mean MADRS at two weeks after the most recent dose increased with each subsequent dose. Self-reported depressive symptoms (QIDS-SR) declined by a magnitude similar to the MADRS, whereas self-reported anxiety (GAD-7) showed only small, short-lived improvements and largely returned to baseline by two months. There was no statistically significant change on the MADRS suicidal ideation item between baseline and the primary endpoint in either group. Adverse events reported were mostly mild to moderate and transient, consistent with expectations for PAP, aside from the single persistent adverse event described above.

Rosenblat and colleagues conclude that a brief PAP protocol is feasible and can be safely evaluated in a more complex, treatment-refractory clinical population that included both MDD and BDII, substantial comorbidity and baseline suicidality. Feasibility was shown both for broadened eligibility criteria and for an abbreviated psychotherapy model consisting of one brief preparatory session and two integration sessions per psilocybin dose, amounting to roughly 4.5 hours of non-dosing psychotherapy per dose. The investigators observed clinically meaningful reductions in depressive symptoms that were consistent across clinician-rated (MADRS) and self-reported (QIDS-SR) measures, while anxiety improvements were limited and transient. The trial also demonstrated feasibility for offering flexible repeat doses upon signs of relapse, and exploratory data suggested incremental antidepressant benefit with subsequent doses, although the authors caution that indication bias limits causal interpretation of that finding. When positioned against prior trials, the authors note that the between-group MADRS difference of 6.6 points at the primary endpoint mirrors that reported in a large phase 2b RCT (Goodwin et al.), but absolute reductions were smaller than in some prior studies. Rosenblat et al. attribute this partly to the greater complexity and higher degree of treatment resistance in their sample—29/30 participants would have been excluded from that previous TRD RCT—while acknowledging that abbreviated psychotherapy could also have contributed to smaller effect sizes. Inclusion of BDII (n = 4) and participants with personality disorders (n = 9) was feasible; no treatment-emergent mania, hypomania or psychosis was observed, although the authors stress that the small subsamples preclude firm safety conclusions for these diagnoses. Key limitations are openly acknowledged: the open-label design, small sample size, use of a waitlist rather than placebo control, sample heterogeneity, lack of expectancy measurement, and deliberate modifications to trial procedures compared to prior PAP studies. These limitations impair causal inference and preclude contribution toward regulatory approval. The authors therefore recommend adequately powered, controlled RCTs to evaluate PAP in complex clinical populations, to determine optimal numbers of psilocybin doses for acute and maintenance phases, and to clarify the necessary intensity of adjunctive psychotherapy for safe, effective and scalable implementation.