Effects of discontinuation of serotonergic antidepressants prior to psilocybin therapy versus escitalopram for major depression

Major depressive disorder (MDD) is a leading contributor to global disability and is commonly treated with selective serotonin reuptake inhibitors (SSRIs) and serotonin–norepinephrine reuptake inhibitors (SNRIs), but many patients do not respond adequately to these drugs. Psilocybin, a classic psychedelic that acts primarily at 5-HT2A receptors, has shown promise as an alternative treatment in recent trials, and acute subjective experiences during dosing are thought to help produce longer-term clinical benefits. In earlier work, including a double-blind randomised trial comparing psilocybin with escitalopram, participants who were taking serotonergic antidepressants at screening were required to taper off them before study entry because of limited safety data on co-administration and prior reports that chronic SRI treatment can attenuate psychedelic subjective effects. Erritzoe and colleagues set out to explore whether pre-trial discontinuation of SSRIs/SNRIs influenced outcomes in that psilocybin versus escitalopram trial. Using exploratory post hoc analyses, they asked whether people who had recently discontinued serotonergic medication ("discontinuers") differed from those who were unmedicated at trial entry on baseline symptom changes, antidepressant response (primary outcome QIDS-SR16 and several secondary measures), acute psychedelic experience measures, and expectation ratings, and whether any discontinuation effects differed between the two treatment arms.

Design and participants: The data derive from a double-blind randomised controlled trial that compared two oral doses of psilocybin (COMP360) plus 6 weeks of placebo to two placebo-like psilocybin doses plus 6 weeks of escitalopram in adults with moderate-to-severe unipolar MDD. Fifty-nine participants were randomised: 30 to the psilocybin condition (11 female) and 29 to escitalopram (9 female). Inclusion criteria included age 18–85, physical health and a diagnosis of unipolar MDD; key exclusions were psychotic disorder, family history of psychosis, pregnancy, substance misuse and MRI contraindications. Those taking serotonergic medication at screening were required to discontinue under GP and study psychiatrist supervision prior to baseline; tapering was typically linear over 2–4 weeks with a minimum 2-week (or ~5 half-lives) drug-free washout before baseline. Intervention and psychosocial support: Participants received approximately 20 hours of in-person therapeutic support (preparation, dosing support with reclining, eye mask and headphones, and integration) plus optional integration calls, following a standard psychedelic-therapy protocol. Psilocybin participants received two 25 mg doses separated by 3 weeks and a 6-week course of placebo; escitalopram participants received two 1 mg (placebo-like) psilocybin doses and a 6-week course of escitalopram. Outcomes: The primary outcome of the parent trial was the Quick Inventory of Depressive Symptomatology Self-Report (QIDS-SR16), administered weekly. Secondary and additional measures analysed here included the Beck Depression Inventory (BDI-1A, bi-weekly), clinician-rated HAM-D and MADRS (baseline and 6-week follow-up) and psychological well‑being (WEMWBS, bi-weekly). Acute subjective indices of the psychedelic experience were assessed with the Mystical Experience Questionnaire (MEQ), Ego Dissolution Inventory (EDI), Emotional Breakthrough Inventory (EBI) and Challenging Experience Questionnaire (CEQ). Expectation for treatment benefit was measured at baseline by asking participants to rate expected improvement (0–100) for 6 weeks of escitalopram and for two doses of psilocybin. Statistical approach: All analyses were exploratory and post hoc. The investigators used linear and linear mixed-effects models implemented in R (lme4, lmerTest) to test effects of Treatment (psilocybin vs escitalopram), Timepoint and Discontinuation status, and their interactions. Models compared nested specifications (referred to as M1, M2 and M3) via ANOVA or information criteria to determine whether Treatment or Discontinuation terms added significantly. Effect estimates, 95% confidence intervals and parameter effect sizes for interaction terms were reported; given the exploratory nature and high inter-correlation of measures, no correction for multiple comparisons was applied. Demographic group differences were tested with t-tests and chi-square tests. Acute-experience and expectation effects were tested with linear models and model selection using Akaike information criteria.

Demographics and discontinuation: There were no significant demographic differences between discontinuers and unmedicated patients in age, sex, ethnicity or employment. In total, 11 psilocybin-arm and 9 escitalopram-arm participants had discontinued SSRIs/SNRIs (venlafaxine, paroxetine, sertraline, duloxetine and citalopram) prior to the trial; a small number who had discontinued other psychiatric medications were excluded from further analyses. All discontinued participants had been off medication for at least 2 weeks before baseline. Baseline effects of discontinuation: Regression and mixed‑model analyses indicated a worsening of depressive symptoms from screening to baseline among discontinuers versus unmedicated patients. For QIDS-SR16, discontinuation predicted higher baseline scores with a mixed-model estimate of β = 2.21 (95% CI 0.21, 4.23), and for BDI the increase was β = 3.66 (95% CI 0.85, 6.48). Simple regressions on baseline scores showed trends in the same direction for QIDS-SR16, HAM-D, MADRS and BDI, but some of those CIs crossed zero in the extracted text. Primary outcome (QIDS-SR16): Analyses stratified by discontinuation status produced contrasting results. Among discontinuers, model comparison showed that Treatment (psilocybin vs escitalopram) did not significantly add to the model, indicating no consistent difference between arms. Conversely, among unmedicated patients Treatment did add significantly, with the psilocybin arm showing greater decreases from baseline at almost all timepoints except week 2. Reported additional decreases for psilocybin versus escitalopram in unmedicated patients included Week 1 β = -3.73 (95% CI -6.40, -1.06), Week 3 β = -3.21 (95% CI -5.81, -0.61), Week 4 β = -3.89 (95% CI -6.56, -1.22), Week 5 β = -4.08 (95% CI -6.73, -1.44), Week 6 β = -3.39 (95% CI -6.01, -0.76) and Follow-up β = -3.00 (95% CI -5.60, -0.40). The authors state that these amounted to an additional three to four QIDS-SR16 points in the psilocybin group, corresponding to an additional antidepressant effect of roughly 50%–120% over escitalopram in this subgroup. Other depression and well‑being measures: The pattern seen for QIDS-SR16 was mirrored on HAM-D, MADRS, BDI and WEMWBS. For discontinuers, Treatment did not significantly improve model fit for these measures. Among unmedicated participants, psilocybin produced substantially greater improvements than escitalopram: for HAM-D the additional drop was reported as 9.52 points (95% CI -12.43, -6.62), and for MADRS an additional drop of 12.08 points was reported (95% CI not fully clear in the extraction). BDI showed significant additional drops at Week 2 (β = -9.25, 95% CI -15.52, -2.99), Week 4 (β = -13.68; CI not fully reported) and Follow-up (β = -9.68; CI not fully reported). WEMWBS scores increased more with psilocybin among unmedicated patients, with added increases of about 13.4–14.2 points at several timepoints. Full-model interactions (M3): When modelling all subjects together, discontinuation moderated responses differently by treatment arm. In the escitalopram arm, discontinuers showed greater decreases in QIDS-SR16 than unmedicated patients at weekly timepoints (e.g. Week 1 β = -3.64, 95% CI -7.04, -0.24; Week 3 β = -5.91, 95% CI -9.18, -2.63). By contrast, in the psilocybin arm discontinuation was associated with a reduced antidepressant response versus unmedicated patients, with significant positive Time:Treatment:Discontinuation interactions at most timepoints (e.g. Week 2 β = 5.99, 95% CI 1.33, 10.66; Week 3 β = 6.47, 95% CI 1.91, 11.02; Week 5 β = 8.31, 95% CI 3.67, 12.94; Follow-up β = 4.79, 95% CI 0.23, 9.35), indicating that discontinuers in the psilocybin condition improved less than unmedicated participants. Acute subjective experience and expectation: Regression analyses showed that discontinuation status did not significantly predict acute experience scores on MEQ, EDI, EBI or CEQ for either dosing session (examples: MEQ dosing 1 β = 0.04, 95% CI -0.11, 0.18; CEQ dosing 1 β = 0.002, 95% CI -0.089, 0.093). By contrast, Treatment condition strongly predicted these acute measures, with clear psilocybin > escitalopram effects on MEQ, EDI and EBI. Baseline expectation analyses indicated no significant relationship between discontinuation and expected efficacy of psilocybin, though there was a trend for higher expectancy for escitalopram among discontinuers (β = 10.9, 95% CI -0.09, 21.88).

Erritzoe and colleagues interpret these exploratory post hoc findings to mean that recent discontinuation of SSRIs/SNRIs may have attenuated the antidepressant response to psilocybin in this trial, while paradoxically those who discontinued appeared to respond better to escitalopram. The investigators suggest that this discontinuation effect could have contributed to the absence of a statistically significant difference on the trial's primary outcome when analysed without accounting for discontinuation, and therefore that discontinuation status is an important potential confound to consider in psychedelic trials. Two principal mechanisms are discussed. First, chronic SSRI/SNRI treatment can induce downregulation or desensitisation of serotonin receptors, notably 5-HT2A receptors, which could in theory blunt psilocybin's efficacy; however, the present analyses found no evidence that discontinuation altered the intensity of acute subjective psychedelic experiences, which complicates a simple receptor-downregulation explanation and raises the possibility of a dissociation between acute subjective effects and antidepressant outcome. Second, the process of tapering and short-term withdrawal may have produced a discontinuation syndrome or temporary worsening of depressive symptoms: discontinuers showed increased symptom severity between screening and baseline, and withdrawal-related somatic and affective symptoms could have inflated baseline severity and interact with subsequent treatment response. Expectation effects may also have played a role, as there was a trend for discontinuers to have higher expectation for escitalopram, which could partly explain their relatively larger response in that arm. The authors emphasise that these analyses are exploratory and post hoc, with small and unequal subgroup sizes, and therefore cannot establish causation. Important limitations they acknowledge include the absence of formal measurement of withdrawal symptoms, lack of detailed taper-duration data, potential baseline differences between discontinuers and unmedicated patients (for example, greater prior treatment exposure among discontinuers), and the inability to disentangle drug effects from the considerable psychological support provided to all participants. They note that prior small studies have produced mixed findings on concomitant SSRI use and psychedelic effects, and that safety evidence for combining SSRIs/SNRIs with classic psychedelics is still limited. Finally, the researchers outline possible clinical and research implications discussed in the paper: if concomitant SSRIs/SNRIs reduce psilocybin efficacy, potential strategies for future study include longer or partial tapering regimens, investigating higher psychedelic doses for those on serotonergic medication, or formally testing continuation versus discontinuation in controlled trials. They also caution about cardiac considerations (QTc prolongation) when combining high-dose psilocybin with certain SSRIs/SNRIs and call for confirmatory research to inform clinical practice and policy.

The investigators conclude that recent discontinuation of SSRIs/SNRIs may be associated with a reduced antidepressant response to psilocybin. They recommend caution when deciding to taper serotonergic antidepressants prior to psychedelic treatment and suggest that longer or partial tapering strategies could be considered, but stress that confirmatory, prospectively designed research comparing continuation versus discontinuation is urgently needed. The authors also note that a requirement to discontinue commonly prescribed antidepressants could complicate broader implementation of psilocybin therapy, and that further work is required to determine whether discontinuation is a necessary prerequisite for effective treatment.