Psilocybin with psychological support improves emotional face recognition in treatment-resistant depression

Psilocybin, a naturally occurring serotonergic psychedelic, has been investigated for a range of psychiatric indications and prior trials have reported improvements in depressive symptoms, anxiety associated with terminal illness, and substance use disorders when combined with psychological support. Like other classical psychedelics, psilocybin acts primarily via 5-HT2A receptor agonism; the authors note a complex and partly paradoxical literature on 5-HT2A pharmacology and antidepressant mechanisms. Earlier pilot work with psilocybin-assisted therapy produced promising results, but questions remain about durability of effect and reproducibility in treatment-resistant populations. This study extends a previously published pilot by increasing the sample from 12 to 20 patients with treatment-resistant unipolar major depressive disorder and lengthening follow-up to 6 months. The trial aimed to assess safety, tolerability and clinical effects of two supervised oral psilocybin sessions (10 mg then 25 mg, one week apart) delivered within a standardised psychological support model, with the primary clinical endpoint set at 5 weeks post high-dose session using the self-rated QIDS-SR16 as the primary outcome measure.

Regulatory approvals were obtained from UK research ethics and medicines agencies and psilocybin was sourced from a licensed manufacturer and encapsulated for investigational use. The investigators conducted an open-label feasibility trial of psilocybin-assisted psychological therapy in 20 patients meeting criteria for treatment-resistant, unipolar major depression. Key inclusion criteria included a HAM-D score ≥ 16 and failure to respond to at least two pharmacologically distinct antidepressant trials during the current episode; major psychotic disorder or a family history of psychosis were exclusion criteria. With one exception, participants taking antidepressants were tapered off under clinician supervision prior to dosing. Treatment comprised two oral dosing sessions separated by 7 days: 10 mg in the first session and 25 mg in the second. Psychological support was structured into three components (acronym PSI): preparation, acute/peri-acute support during sessions, and post-session integration. Safety monitoring included baseline physical and psychiatric screening, routine blood tests, ECG and drug/pregnancy screens, and systematic solicitation of side effects at every post-treatment visit. Clinical assessments featured the QIDS-SR16 as the primary outcome, collected at 1–3 and 5 weeks and at 3 and 6 months post-treatment, with the 5-week time point designated the primary endpoint. Secondary measures included the BDI, STAI-T, SHAPS, HAM-D and GAF at various time points. The 11-dimension Altered States of Consciousness (11D-ASC) questionnaire was administered after each dosing day with statistical contrasts between the 10-mg and 25-mg sessions. Analyses used two-tailed paired t tests for pre-vs post-treatment contrasts (Bonferroni correction for the six QIDS-SR16 contrasts, α = 0.0083), with 95% confidence intervals and Cohen's d for dependent data. The authors chose not to correct for multiple comparisons across the suite of secondary clinical measures because of high covariance among them; all relevant p values and effect sizes were reported.

Recruitment began with 120 enquiries; 74 proceeded to telephone screen, 29 were invited for in-person screening and 20 were enrolled, of whom 19 completed all assessment time points. At baseline 18 of 20 patients met criteria for severe or very severe depression (QIDS-SR16 ≥ 16). The sample had a chronic course (mean reported duration of current illness 17.7 ± 8.4 years) and a high treatment history burden (median lifetime failed medications = 4; mean 4.6 ± 2.6). Relative to baseline, QIDS-SR16 scores were significantly reduced at all six post-treatment time points (all p < 0.001). The largest mean change occurred at 5 weeks: mean change −9.2 (95% CI −11.8 to −6.6), t = −7.2, p < 0.001, Cohen's d = 2.3. All 19 completers showed some reduction in depression severity at 1 week, and reductions were sustained in the majority through weeks 3–5. The BDI showed significant reductions at 1 week (mean reduction −22.7, 95% CI −17.6 to −27.8, p < 0.001), 3 months (−15.3, 95% CI −8.7 to −21.9, p < 0.001) and 6 months (−14.9, 95% CI −8.7 to −21.1, p < 0.001). Trait anxiety (STAI-T) decreased significantly at 1 week (mean reduction −23.8, 95% CI −16.5 to −31.1, p < 0.001), 3 months (−12.2, 95% CI −6.1 to −18.3, p < 0.001) and 6 months (−14.8, 95% CI −8.1 to −21.6, p < 0.001). SHAPS anhedonia scores and clinician-rated HAM-D and GAF also showed significant improvement at the reported time points (SHAPS at 1 week and 3 months; HAM-D and GAF at 1 week). Safety and tolerability were acceptable in this sample; there were no serious adverse events. The most common transient effects were anxiety during the acute session (n = 15) and short-lived headaches (n = 8); five patients reported transient nausea and three reported transient paranoid ideas during the acute experience. One participant became uncommunicative during the 25-mg session but returned to baseline as drug effects subsided; that patient later described the experience as “blissful and beneficial but also overwhelming” and subsequently withdrew from completing most follow-up measures. Fourteen patients reported autobiographical-type visions, generally experienced as insightful. Suicidality scores on the QIDS-SR16 and the HAM-D suicide item were significantly reduced at early post-treatment time points (e.g. QIDS suicidality mean reductions at week 1 = −0.9, 95% CI −0.4 to −1.4, p < 0.002). The 11D-ASC revealed that several dimensions were significantly higher after 25 mg versus 10 mg following Bonferroni correction: experience of unity, spiritual experience, blissful state, insightfulness and complex imagery (all p ≤ 0.001 for reported contrasts). Because unity, spiritual experience and blissful state were highly intercorrelated (r > 0.92), the authors combined them into a single 'USB' factor. Scores on USB and on the insightfulness factor during the 25-mg session correlated with greater reductions in QIDS-SR16 at 5 weeks (USB r = −0.49, p = 0.03; insight r = −0.57, p = 0.01). At 6 months some participants had received additional treatments: one participant remained on venlafaxine throughout, six began new antidepressants after 3 months and five received psychotherapy around the 3-month point. Five participants independently obtained psilocybin outside the trial between the 3- and 6-month assessments; excluding these five did not substantially change the 3- and 6-month results (3 months effect size ≈ 1.6, p < 0.001; 6 months effect size ≈ 1.7, p = 0.018). Of the nine participants who met response criteria at 5 weeks, three met conservative relapse criteria by 6 months while six maintained response. The authors caution that these follow-up observations are confounded by additional treatments and small subgroup sizes.

Carhart-Harris and colleagues present extended data from an open-label feasibility trial of psilocybin plus psychological support in treatment-resistant depression, finding rapid and often sustained reductions in depressive symptoms following two supervised doses (10 mg then 25 mg). Improvements were evident in both self-report and clinician-rated measures, with peak nominal effects at 5 weeks, and some participants retained clinically meaningful benefit to 3 and 6 months. The pattern of acute subjective effects—particularly dimensions labelled as unity, spiritual experience, blissful state and insightfulness—was associated with subsequent clinical improvement, a finding the authors view as supportive of a psychological-mechanism model linking acute experience quality to therapeutic outcome. The investigators emphasise several important limitations. The open-label design and lack of a control condition prevent causal attribution and make placebo, expectation and contextual effects plausible contributors to observed outcomes. Later follow-up was confounded by additional, non-study treatments some participants received, and the small sample size—especially the subgroup of responders—limits inferences about durability of effect and relapse risk. The final eight enrolees being all male is noted as a limitation for generalisability. Measurement issues were also acknowledged, for example variability in how patients estimated duration of their current depressive episode. For future work, the authors call for double-blind randomised controlled trials to better establish efficacy, for more systematic characterisation and measurement of the psychological components of the treatment model (preparation, in-session support and integration), and for potential manualisation of psychotherapeutic support. They note the ethical and safety implications of experimenting with context and psychological variables in vulnerable participants and argue that certain standards of care will likely be non-negotiable. Finally, the authors suggest that comparative efficacy trials, refinement of mechanistic hypotheses (including using neuroimaging data collected in this programme) and assessment of cost-effectiveness are important next steps if psilocybin-assisted therapy is to be evaluated for broader clinical use.