MDMA-assisted therapy significantly reduces eating disorder symptoms in a randomized placebo-controlled trial of adults with severe PTSD

Eating disorders (EDs) and posttraumatic stress disorder (PTSD) frequently co-occur and share risk factors such as female sex, trauma exposure, and limited social supports, leading to greater morbidity, treatment dropout and poorer outcomes in people with both conditions. Earlier research has established MDMA-assisted therapy (MDMA-AT) as an efficacious integrated treatment for treatment-resistant PTSD, but its effects on ED psychopathology—particularly ED symptoms that occur without low weight or active purging—have been little explored. Brewerton and colleagues used data from a Phase III randomized, double-blind, placebo-controlled trial of MDMA-AT for severe PTSD to examine pre-specified exploratory outcomes on disordered eating. They hypothesised that a substantial subset of trial participants would endorse elevated symptoms on the Eating Attitudes Test-26 (EAT-26) despite exclusions for active purging and low weight, and that MDMA-AT would reduce EAT-26 scores versus placebo-assisted therapy (PLAC-AT), with a particular effect in women. The primary aim reported here was to compare changes in EAT-26 scores between MDMA-AT and PLAC-AT participants.

This analysis used exploratory EAT-26 data collected within a Phase III, multicentre, double-blind, randomised controlled trial (trial ID NCT03537014) conducted at 15 sites across the United States, Canada and Israel. Ethics approvals were obtained at multiple institutional review boards. Full primary outcome methods have been published elsewhere; the present paper reports on pre-specified exploratory EAT-26 outcomes taken at baseline and study termination. Eligible participants were adults (≥18 years) meeting DSM-5 criteria for current PTSD of at least six months' duration. Individuals with an active ED involving purging or who were underweight were excluded for safety. After informed consent and washout of psychiatric medications (defined as 5 half-lives plus one week), participants were randomised to MDMA-AT or inactive placebo with identical therapy (PLAC-AT). The intervention comprised three 8-hour experimental sessions about 4 weeks apart. MDMA dosing was 80–180 mg followed by a supplemental half-dose of 40–60 mg; placebo sessions used inactive comparator. Each experimental session was preceded by fasting, drug and pregnancy screening, suicidality assessment and vital signs. Trained therapists delivered manualised psychotherapeutic support during sessions and three 90-minute integration follow-up sessions starting the morning after each experimental session. The EAT-26 self-report questionnaire was collected at baseline (visit 4) and at study termination (visit 20) to assess attitudes and symptoms related to eating; scores range 0–78, with cut-offs of ≥11 often used to indicate 'at risk' and ≥20 indicating probable clinical concern. The main trial’s primary outcome was CAPS-5 for PTSD severity, administered by blinded independent raters. Statistical analyses included descriptive comparisons, t-tests, ANCOVA adjusting for baseline EAT-26 scores to compare between-group change, and non-parametric tests where appropriate. Subset analyses stratified by gender/sex and baseline EAT-26 cut-offs (≥11 and ≥20) were prespecified. Pearson correlations assessed associations between change in EAT-26 and change in CAPS-5. Effect sizes were reported (Cohen’s d or Hedge’s g), and Reliable Change Index (RCI) and Jacobson-Truax methods were used to evaluate reliable and clinically meaningful change using an internal consistency α = .83 for EAT-26. Analyses were conducted in SAS 9.4. The extracted text does not clearly report any additional deviations from these planned analyses.

Ninety participants were randomised and received treatment; 89 completed baseline EAT-26 and 82 (91.1%) completed both baseline and follow-up EAT-26 and were analysed as completers (3 MDMA, 4 placebo withdrew before follow-up). At baseline, 15.7% (15/89) met criteria for a current ED (BED n=5; OSFED n=9) and 14.6% (13/89) had a past ED history. The analysed sample was majority female (65.2%), mean age 41.0 (SD 12.0) years, and mean BMI was in the overweight/normal range overall; there were no significant baseline differences between treatment groups on demographics, ACE, BDI-II, CAPS-5 or suicidality measures according to the extracted text. On the EAT-26, 13/89 (15%) participants had baseline scores >20 (clinical range) and 28/89 (31.5%) had scores >11 (at-risk range). Among completers (n=82), 11 (13.4%) were >20 and 24 (29.3%) were >11 at baseline. In the total analysed sample, ANCOVA adjusted for baseline indicated a statistically significant between-group difference in EAT-26 change favouring MDMA-AT (reported adjusted difference = 4.68, p = .0335; Hedge's g = .33). Within-group changes showed the MDMA group had a mean reduction of -3.04 (SD 6.24) from baseline to follow-up (p = .02), versus a reduction of -0.68 (SD 8.04) in the placebo group. However, RCI analyses for the entire sample did not meet the threshold for reliable change (RCI placebo = -0.15; MDMA = -0.84), and the overall change was not judged clinically meaningful by the Jacobson-Truax criterion as reported. Gender-stratified analyses indicated women in the MDMA group (n=22) showed a within-subject mean reduction of -4.50 (SD 7.67) in EAT-26 scores (p = .02), though this change did not meet the RCI threshold for reliable change (RCI = -1.01) and the between-group difference in women only was not clearly significant in the primary gender-stratified comparison (the authors note limited power due to sample size). Participants with higher baseline EAT-26 scores improved more. In the baseline EAT-26 ≥11 subset (approximately 12 placebo and 12 MDMA participants), the MDMA group had a mean reduction of -9.58 (SD 7.59) (p = .0007), which was significantly greater than the placebo reduction (reported as approximately -3.58 in the extracted text). For women within the ≥11 subset the between-group difference was significant (F(2,14)=17.68; p = .0009; Hedge's g = .63). RCI indicated reliable change for the MDMA group in this ≥11 women subset (RCI = -2.90) but not for placebo (RCI ≈ -0.50); neither group's change was judged clinically meaningful by the authors' criteria in the broader ≥11 group. In the smaller baseline EAT-26 ≥20 subset, both placebo and MDMA groups showed large within-subject reductions (placebo mean change reported -14.08, SD 8.96; MDMA also substantial), and reductions were judged clinically meaningful for both groups. Only the MDMA group reached the RCI threshold for reliable change (RCI = -2.83), while the placebo group's RCI (-1.85) did not meet the conventional reliable-change cutoff. Among women with baseline EAT-26 ≥20, the between-group difference favoured MDMA (F(1,7)=5.75; p = .0478; Hedge's g = .60). The extracted text also reports an RCI value of -18.80 for women with baseline ≥20, but this value appears inconsistent with other RCI figures and is not clearly interpretable from the extraction. Change in EAT-26 scores was independent of change in PTSD severity: no significant correlations were found between change in CAPS-5 and change in EAT-26; the extracted text gives small, non-significant correlations (e.g. MDMA r = .07, p = .67; placebo r = .06, p = .72). Two participants identified as non-binary (both assigned female at birth and in the MDMA group), limiting gender subgroup analyses. Post-treatment BMI was available for 69 participants and showed no significant between-group difference in BMI change (MDMA mean change 0.036 [SD 1.59] kg/m2; placebo 0.305 [SD 1.2] kg/m2); the t-statistic is not clearly reported in the extracted text.

Brewerton and colleagues report this as the first exploratory analysis of MDMA-AT effects on ED symptomatology using data from a randomised placebo-controlled trial in individuals with severe PTSD. They emphasise that ED psychopathology was common in the PTSD sample despite exclusion of participants with active purging or underweight status: roughly 15% met criteria for a current ED and approximately 31.5% had EAT-26 scores in the at-risk range. These findings corroborate previous observations that ED symptoms and PTSD frequently co-occur even in the absence of frank low weight or purging behaviours. The authors interpret their main finding as evidence that MDMA-AT produced significantly greater reductions in EAT-26 scores than PLAC-AT in the total analysed sample, with the effect concentrated among participants with higher baseline EAT-26 symptom burden and, in many analyses, among women. They note that for participants with baseline EAT-26 ≥11 and ≥20, MDMA-AT achieved reliable change by RCI standards in several subgroup analyses, whereas PLAC-AT generally did not, although both groups sometimes showed clinically meaningful change in the small ≥20 subgroup. Mechanistic explanations are acknowledged as speculative within the paper: the investigators refer to MDMA's anxiolytic, prosocial and socioemotional processing effects as plausible contributors to improvement in ED symptoms, but they also point out that reductions in ED symptoms were not correlated with reductions in PTSD severity in their data, leaving the causal pathway unclear. Key strengths highlighted include the double-blind randomized design, use of validated structured instruments to diagnose PTSD, and a pre-specified, reliable self-report measure for ED symptoms. Limitations are described in detail: the trial excluded individuals with active purging or low weight so findings do not generalise to full-syndrome ED populations; the analysis was limited to completers of the EAT-26 (not an intention-to-treat analysis), which may introduce bias; subgroup sizes (particularly men and non-binary participants) were small and underpowered; some post-treatment BMI data were missing; and certain reported metrics in the extraction are unclear. The authors conclude that while MDMA-AT appears promising for reducing ED-related psychopathology in people with severe PTSD, further controlled trials specifically targeting ED populations (with and without comorbid PTSD) are needed to confirm efficacy and generalisability.