MDMA-assisted therapy for severe PTSD: a randomized, double-blind, placebo-controlled phase 3 study

This was a randomised, double-blind, placebo-controlled Phase III trial testing MDMA-assisted therapy for people with moderate-to-severe PTSD. Participants were recruited via advertisements and referrals and had to meet DSM-5 criteria for current PTSD of at least 6 months’ duration and have a baseline CAPS-5 total severity score of ≥35. Key exclusions included primary psychotic disorder, bipolar I disorder, dissociative identity disorder, current alcohol or substance use disorders, pregnancy or lactation, and medical conditions that would make a sympathomimetic drug hazardous (for example uncontrolled hypertension or arrhythmia). Participants were required to discontinue many psychiatric medications before baseline according to a supervised taper schedule. Randomisation was 1:1 to MDMA-assisted therapy or therapy with an inactive placebo, stratified by site and managed centrally to maintain blinding. To limit bias in efficacy measurement, outcome assessments (CAPS-5 and SDS) were conducted by an independent, observer‑blind pool of raters who did not repeatedly see the same participant and were blind to treatment allocation and visit number. The treatment model comprised three 90‑minute preparatory sessions, three 8‑hour experimental sessions spaced approximately 4 weeks apart, and nine 90‑minute integration sessions, with independent-rater efficacy assessments after each experimental session and a primary endpoint at ~18 weeks after baseline (about 8 weeks after the third experimental session). Experimental-session dosing used a circadian-adjusted morning schedule: session 1 typically 80 mg followed by a supplemental 40 mg 1.5–2.5 h later; sessions 2 and 3 typically 120 mg followed by a supplemental 60 mg. Primary efficacy was change in CAPS-5 total severity score from baseline to 18 weeks after baseline; the secondary efficacy endpoint was change in clinician-rated functional impairment as measured by the Sheehan Disability Scale (SDS). Exploratory outcomes included depression (BDI-II), alcohol and drug use screening measures, and adverse childhood experiences. Safety monitoring emphasised treatment-emergent adverse events (TEAEs) and specified adverse events of special interest (AESIs) related to cardiac function, abuse liability and suicidality; suicidality was tracked throughout with the Columbia Suicide Severity Rating Scale (C-SSRS). Statistical analysis used mixed model repeated measures (MMRM) comparing change from baseline in CAPS-5 and SDS between groups. The statistical analysis plan implemented the use of estimands: a de jure estimand (efficacy if the drug is taken as directed) as the primary analysis and a de facto estimand (effectiveness if taken as assigned regardless of adherence) as a sensitivity analysis. The modified intent-to-treat (mITT) set included participants who completed at least one blinded experimental session and one post-treatment assessment; a per-protocol completer set comprised participants who completed all three experimental sessions and assessments. Missing outcome data were not imputed. Effect sizes were reported with Cohen’s d and two‑sided hypothesis tests controlled type I error using a hierarchical testing strategy.

The modified intent-to-treat set comprised 90 participants (46 randomised to MDMA, 44 to placebo with therapy); the de jure efficacy analysis included 89 participants after excluding one participant with no post‑dose data. The per-protocol completer set included 42 participants in the MDMA arm and 37 in the placebo arm. The primary outcome showed a significant and clinically meaningful greater reduction in CAPS-5 scores for the MDMA-assisted therapy group versus placebo with therapy. MMRM analysis of the de jure estimand reported a between-group difference of 11.9 points on the CAPS-5 from baseline to 18 weeks (95% CI = 6.3–17.4, P < 0.0001, d.f. = 71). Among completers, mean CAPS-5 change was −24.4 (s.d. 11.6; n = 42) in the MDMA group and −13.9 (s.d. 11.5; n = 37) in the placebo group. Between-group effect size was d = 0.91 (95% CI = 0.44–1.37) for the de jure estimand and d = 0.97 (95% CI = 0.51–1.42) for the de facto estimand. When considering within‑group change (which includes the effect of the supportive therapy common to both arms), the reported within‑group effect sizes were larger in the MDMA group (d = 2.1) than in the placebo group (d = 1.2). Clinician-rated functional impairment improved significantly more in the MDMA group. MMRM of the de jure estimand for SDS showed a significant between‑group difference (n = 89, P = 0.0116) with an effect size of 0.43 (95% CI = −0.01–0.88); completer mean SDS change was −3.1 (s.d. 2.6; n = 42) for MDMA and −2.0 (s.d. 2.4; n = 37) for placebo. Exploratory analysis of depressive symptoms (BDI‑II) found greater reduction in the MDMA arm: mean change −19.7 (s.d. 14.0; n = 42) versus −10.8 (s.d. 11.3; n = 39) in placebo (t = −3.11, P = 0.0026), with an effect size of 0.67 (95% CI = 0.22–1.12). Diagnostic outcomes favoured MDMA-assisted therapy. At the primary endpoint (18 weeks), 28 of 42 participants (67%) in the MDMA group no longer met criteria for PTSD versus 12 of 37 (32%) in the placebo group. Remission, defined by the study as loss of diagnosis plus CAPS-5 total score ≤11, occurred in 14 of 42 participants (33%) in the MDMA arm and 2 of 37 participants (5%) in the placebo arm. Subgroup and site analyses indicated consistency across study sites (no significant site effect, P = 0.1003). MDMA benefit was also observed in participants with comorbidities often associated with poorer outcomes. For example, participants with the dissociative subtype showed substantial reductions in CAPS-5 with MDMA (mean Δ = −30.8, s.d. 9.0) compared with placebo (mean Δ = −12.8, s.d. 12.8); reductions in non‑dissociative participants were similar (MDMA mean Δ = −23.6, s.d. 11.7; placebo mean Δ = −14.3, s.d. 11.2). The extracted text reports no obvious modulation of treatment effect by history of SSRI use, history of alcohol or substance use disorder, overnight stay, or severe childhood trauma. Safety findings: TEAEs that were more common in the MDMA arm tended to be transient and of mild–moderate severity and included muscle tightness, decreased appetite, nausea, hyperhidrosis and feeling cold. Expected transient increases in systolic and diastolic blood pressure and heart rate occurred during MDMA sessions; two MDMA participants had transient body temperature rises to 38.1 °C. Two participants in the placebo arm reported three serious adverse events (SAEs) related to suicidal behaviour or ideation; no SAEs were reported in the MDMA arm in the extracted text. AESIs of suicidality/self‑harm were reported by five participants in the placebo group and three in the MDMA group. One participant in the placebo group reported two cardiovascular AESIs where cardiac aetiology could not be ruled out. One MDMA‑randomised participant discontinued because the CAPS-5 assessments were triggering and because of depressed mood following an experimental session; this participant was classified as a non‑responder. Lifetime suicidal ideation was common (>90% reported lifetime ideation), and baseline past‑week ideation was 37% in MDMA and 32% in placebo; the prevalence of suicidal ideation during the study never exceeded baseline levels and was not exacerbated in the MDMA group. Serious suicidal ideation during the study was minimal and occurred almost entirely in the placebo arm.

Mitchell and colleagues interpret their findings as showing that three doses of MDMA administered alongside manualised supportive therapy produced significant and robust reductions in PTSD symptoms and clinician‑rated functional impairment over an 18‑week period. The investigators also report significant reduction in depressive symptoms and emphasise that MDMA did not increase suicidality in this trial. They note that the between‑group effect size (d ≈ 0.91) for CAPS-5 change exceeds effect sizes reported for FDA‑approved SSRIs in pivotal trials and that participants receiving MDMA plus supportive therapy showed larger within‑group improvement than those receiving supportive therapy with placebo, suggesting MDMA potentiated the therapeutic model used in the study. The discussion places the clinical findings in a mechanistic context, noting prior preclinical and human imaging evidence implicating serotonergic modulation of amygdala activity, brain‑derived neurotrophic factor–mediated plasticity and oxytocin‑related prosocial effects as plausible contributors to therapeutic change. The authors propose that MDMA may open a transient window of tolerance or heightened neuroplasticity that facilitates processing of traumatic memories with less overwhelming fear or dissociation and that the acute prosocial effects may improve therapeutic alliance and adherence. Safety and generalisability are discussed candidly. The authors report no major safety signals specific to MDMA in this dataset, and consider the observed transient cardiovascular effects expected based on prior studies. They acknowledge limitations: the achieved sample size was reduced by the COVID‑19 pandemic though the data retained adequate statistical power; the sample was relatively homogeneous and lacked racial and ethnic diversity; the primary outcome is a short‑term endpoint (≈8 weeks after the final experimental session) with longer‑term durability to be reported later; and blinding was challenging given the subjective effects of MDMA, which may introduce expectation effects despite the use of blinded independent raters for efficacy assessments. The authors also note that safety data collection by site therapists could have affected blinding of some safety assessments, but that efficacy data were collected by blinded raters to minimise bias. Finally, the investigators highlight clinical implications they draw from the data: MDMA-assisted therapy showed efficacy in participants with chronic, severe and comorbid PTSD presentations, including the dissociative subtype, suggesting a potential role for this model in treatment‑resistant or complex cases. They recommend further studies to replicate findings, evaluate specific comorbid populations, and ultimately consider head‑to‑head comparisons with existing pharmacotherapies and longer‑term follow‑up to establish durability and broader safety.