Results From a Long-Term Observational Follow-Up Study of a Single Dose of Psilocybin for a Treatment-Resistant Episode of Major Depressive Disorder

Earlier research has shown that psilocybin, a serotonin 2A receptor partial agonist, can produce rapid antidepressant effects in people with major depressive disorder and treatment-resistant depression (TRD). The largest Phase IIb randomised controlled trial to date (COMP 001, N = 233) found that a single 25 mg dose of COMP360 psilocybin plus psychological support produced higher short-term remission and sustained response rates than lower doses, and that psilocybin was generally well tolerated. However, evidence on longer-term safety and durability of effect is scarce, particularly from randomised trials in the harder-to-treat TRD population; most longer follow-ups have been small or open-label. This study reports COMP 004, a 52-week observational follow-up of participants drawn from COMP 001 and from a small open-label adjunctive study (COMP 003). The principal aim was to describe time to a prespecified “depressive event” over 52 weeks after COMP360 administration and to characterise adverse events (AEs) during this period. The analysis was descriptive and included a prespecified primary analysis of the full COMP 001 cohort and a post hoc supplementary analysis limited to those who enrolled into the COMP 004 follow-up study.

COMP 004 was a 40-week observational extension that, together with the 12-week lead-in from COMP 001 or COMP 003, produced a 52-week follow-up after COMP360 administration. Eligible participants were those who completed the Week 12 visit of COMP 001 (a Phase IIb RCT of single-dose COMP360 at 25 mg, 10 mg, or 1 mg) or participants from COMP 003 (an open-label study of 25 mg COMP360 given adjunctive to ongoing selective serotonin reuptake inhibitors). The extracted text does not provide the prespecified list of criteria used to define a “depressive event”; the authors state these were defined a priori but the specific items are not shown in the provided text. No further pharmacological treatment or psychological support was provided as part of COMP 004, although participants were permitted to initiate new treatments during follow-up and to continue any stable psychotherapy that predated entry to COMP 001. Blinding from COMP 001 was maintained through to Week 52. Assessments included clinician-rated Montgomery-Åsberg Depression Rating Scale (MADRS) administered by a blinded independent rater and were scheduled at Weeks 16, 20, 24, 28, 40, and 52; COMP 003 had additional early assessments. Visits were instructed to be remote with in-person visits allowed at investigator discretion. The protocol received ethics approval and all participants gave written informed consent. The prespecified primary efficacy endpoint for participants drawn from COMP 001 was time to first depressive event measured from baseline (one day prior to COMP360 administration in the lead-in study). Time-to-event estimates were obtained using Kaplan-Meier methods (a standard approach to estimate event-free probability over time). Safety endpoints were incidence and severity of adverse events, including treatment-emergent AEs (TEAEs), serious AEs (SAEs), treatment-emergent SAEs (TESAEs), and adverse drug reactions (ADRs) as judged by investigators. All analyses were performed by original randomised dose group for COMP 001 (25 mg, 10 mg, 1 mg) and as 25 mg open-label for COMP 003. Because COMP 004 enrollment was optional and fewer participants enrolled than were eligible, the study was not powered for formal hypothesis testing; no formal statistical tests comparing groups were performed. The primary (prespecified) analysis used the modified full analysis set (mFAS) including all COMP 001 participants who completed at least one efficacy assessment, incorporating depressive events that occurred during COMP 001. A post hoc supplementary analysis restricted the analysis to COMP 001 participants who actually enrolled into COMP 004 (FAS). Censoring rules included censoring participants who did not experience a depressive event by the time they discontinued or completed COMP 001 or COMP 004.

Participant flow and baseline characteristics: Of 126 COMP 001 completers who were offered COMP 004 participation, 58 (46%) consented and 45 completed the final COMP 004 visit; a further 8 participants from COMP 003 enrolled (of 19 original COMP 003 participants). From the entire COMP 001 population (N = 233), the numbers who had already experienced a depressive event by Week 12 were 25 mg: n = 37; 10 mg: n = 38; 1 mg: n = 44. Demographics and baseline clinical features were broadly similar across dose groups, though the subset who entered COMP 004 differed in some respects from the full COMP 001 cohort (for example, a greater proportion recruited from North America and the 10 mg subgroup who entered COMP 004 had greater MADRS improvement at Week 12 than the equivalent subgroup in COMP 001 overall). Primary time-to-event outcomes (prespecified analysis of the full COMP 001 cohort): When all COMP 001 participants who completed at least one efficacy assessment were included (mFAS, n = 233), the proportion experiencing any depression-related event during follow-up was similar across dose groups: 25 mg 54.4% (n = 43), 10 mg 54.7% (n = 41), and 1 mg 57.0% (n = 45). Median time to the first depressive event was longer in the 25 mg group compared with the 1 mg group: 25 mg median 92 days (95% CI 49–199), 10 mg median 83 days (95% CI 39–142), and 1 mg median 62 days (95% CI 28–NE). The most frequent event types were initiation of a new antidepressant (25 mg 48.1%, 10 mg 38.7%, 1 mg 43.0%) and worsening MADRS score (25 mg 21.5%, 10 mg 18.7%, 1 mg 22.8%). Post hoc supplementary analysis restricted to COMP 004 enrollees: Among the subset of COMP 001 participants who actually entered COMP 004 (n = 58), the difference in median time to depressive event was more pronounced: 25 mg median 189 days (95% CI 24, -), 10 mg median 43 days (95% CI 19–142), and 1 mg median 21 days (95% CI 3–78). Only 10 participants from COMP 001 experienced a depressive event after enrolling into COMP 004 (25 mg n = 6, 10 mg n = 3, 1 mg n = 1). Two COMP 003 participants experienced depressive events (both MADRS worsening); quartile estimates and CIs were not calculated for COMP 003 because of small numbers. New treatments and timing: By Week 52, initiation of new antidepressant treatment was common; among those entering COMP 004 the 1 mg group had the highest proportion initiating new treatment by Week 52 (over 75% in the 1 mg group), and initiated earlier post-COMP360 than the 25 mg group. At entry into COMP 004 (Week 12), 58.8% of 1 mg participants had already started a new antidepressant compared with 27.3% in the 25 mg group and 47.4% in the 10 mg group. By Week 52 the proportions initiating new antidepressant treatment were similar between the 25 mg and 10 mg groups (25 mg 54.5% [n = 12] vs 10 mg 57.9% [n = 11]). Safety: Across COMP 004, there were 61 ongoing or new TEAEs reported in 27 participants. The most common TEAE was COVID-19 infection, reported in 13.8% (n = 8); all other TEAEs occurred in two or fewer participants. No AEs led to participant withdrawal or death. Nine ADRs were reported in four participants: for example, one 25 mg participant had decreased appetite, altered mood and anxiety beginning within days of administration and ongoing; one 10 mg participant had affect lability and irritability; in the 1 mg group two participants reported insomnia, altered time perception and an episode of suicidal ideation (the suicidal ideation onset at Day 96 post-administration and resolved by Day 108). Four TESAEs occurred after entry into COMP 004 among three participants and were judged by investigators to be unrelated to study drug; these included one 25 mg participant hospitalised for suicidal ideation on Study Day 307, and in the 1 mg group one participant with breast cancer and one participant with two hospitalisations related to therapy change for moderate suicidal ideation. A nonserious mild suicidal ideation event in the 1 mg group was judged possibly related to study drug; this episode began during COMP 001 and resolved within two weeks after COMP 004 enrolment. No cases of hallucinogen persisting perception disorder (HPPD) or persisting hallucinations were observed in COMP 004 or reported in COMP 001.

Goodwin and colleagues describe these analyses as descriptive and emphasise caution in interpretation because COMP 004 was optional, underpowered, and initiated after the lead-in study, leading to selective enrollment. In the prespecified analysis of the full COMP 001 cohort, median time to a depressive event was longer for the 25 mg dose compared with 1 mg, but overall proportions experiencing any depressive event were similar across dose groups. The post hoc analysis restricted to those who enrolled into COMP 004 showed a larger separation between doses in time to event; the authors note, however, that this subsample may be unrepresentative of the full study population and subject to selection bias. The investigators place their long-term safety observations in context with prior small open-label studies that reported durable antidepressant effects out to 3–12 months, noting that the current findings of clinical benefits out to approximately 6 months are generally consistent with those reports. They also highlight that adverse events, including suicidal ideation and behaviour, occurred across dose groups and that the majority of serious events occurring after six months were judged unrelated to the study drug by investigators. Absence of persisting perceptual disturbances (HPPD) during follow-up is reported. Key limitations acknowledged by the authors include the modest sample size of COMP 004 driven by late study start relative to COMP 001 and the COVID-19 pandemic, low consent/enrolment rate (46% of those offered), resulting selection bias, and a high censoring rate (31.8% of COMP 001 participants were censored at COMP 001 completion). Because of small numbers and the optional design, formal statistical comparisons between treatment groups were not performed. The authors conclude that these limitations restrict firm conclusions about the long-term comparative efficacy of dose levels. Finally, the investigators state that while the single administration of COMP360 appeared safe over 52 weeks and clinical benefits were observed to about 6 months, larger and longer-term controlled studies are necessary to clarify the durability of effect and longer-term safety in TRD and to inform future trial design.

The authors conclude that a single administration of COMP360 psilocybin in participants with treatment-resistant depression was associated with an acceptable safety profile over 52 weeks and that clinical benefits were observable to roughly 6 months post-administration. They stress that the findings are descriptive and limited by sample size and selection bias, and they call for larger long-term studies to confirm durability of antidepressant effect and further characterise safety in this population.