Pilot study suggests DNA methylation of the glucocorticoid receptor gene (NR3C1) is associated with MDMA-assisted therapy treatment response for severe PTSD

Earlier research has implicated epigenetic alterations in hypothalamic-pituitary-adrenal (HPA) axis genes as markers or predictors of psychotherapy response in post-traumatic stress disorder (PTSD). Recent clinical trials have shown robust clinical benefit of MDMA-assisted therapy versus placebo with therapy for severe PTSD, but molecular mechanisms that might underlie treatment response remain unexplored. Epigenetic marks such as DNA methylation of HPA-related genes (notably NR3C1, FKBP5, and CRHR1) have previously been associated with trauma exposure, PTSD diagnosis, and psychotherapy outcomes, making them candidates for investigation in the context of MDMA-assisted therapy. This pilot sub-study, embedded within a recent Phase III randomized trial, aimed to test two hypotheses: (1) changes in DNA methylation of NR3C1, FKBP5, and CRHR1 would predict symptomatic reduction in PTSD (measured by CAPS-5) across both MDMA- and placebo-with-therapy groups; and (2) the MDMA-assisted therapy group would show greater methylation changes than the placebo group at sites that predicted treatment response. The study was exploratory and intended to generate loci and hypotheses for future, larger investigations.

This sub-study sampled participants from a multicentre Phase III trial of MDMA‑assisted therapy for severe PTSD. Of 90 participants treated in the parent trial, 33 consented to the epigenetics sub-study; due to COVID-19 interruptions 10 of these lacked post-treatment saliva, leaving 23 participants with paired pre- and post-treatment samples (MDMA n = 16; placebo n = 7). Participants had completed medication taper and preparatory therapy before randomization, which was stratified by site and allocated 1:1 to MDMA-assisted therapy or placebo with therapy. The intervention mirrored the parent trial: three 8‑hour experimental sessions about 4 weeks apart, each with a single divided dose of 80–180 mg MDMA or inactive placebo, followed by three 90‑minute integration sessions. Primary clinical outcome was CAPS-5 assessed at baseline and ~8 weeks after the third experimental session; saliva for methylation analysis was collected at the same baseline and outcome visits. Salivary DNA was extracted and bisulfite-converted, and genome-wide methylation was assayed using the Illumina Infinium MethylationEPIC BeadChip. Standard quality-control and preprocessing in R (minfi) were performed, probes on sex chromosomes were removed, and estimated epithelial cell proportion (derived with EpiDISH RPC) was used as a covariate. M-values were used for statistical modelling. All CpG probes annotated to NR3C1, FKBP5, and CRHR1 on the EPIC array were examined; sites differing between MDMA and placebo at baseline were excluded from longitudinal analyses. Statistical analyses compared the epigenetics subsample to the parent trial on demographics and CAPS-5 scores, and compared treatment groups within the sub-sample. Multivariate linear regression tested whether methylation change (independent variable) predicted CAPS-5 change (dependent variable), controlling for sex, age, and epithelial cell proportion. Because prior studies report heterogeneous directions of methylation change, the investigators tested both raw methylation difference scores and absolute value change scores (abs) to allow detection of predictive changes irrespective of direction. Given the pilot size, results are reported both uncorrected and after false discovery rate (FDR) correction (FDR threshold noted as < 0.1). For sites that predicted treatment response after FDR correction, group differences in methylation change were assessed using ANCOVA controlling for the same covariates.

The analysed sample (n = 23; MDMA n = 16; placebo n = 7) did not differ from the full trial sample on age, sex, baseline CAPS-5, or outcome CAPS-5. Within the sub-study, MDMA and placebo groups were similar on age and baseline CAPS-5 but differed in sex composition; sex was therefore included as a covariate. Clinically, the MDMA group showed greater reduction in CAPS-5 scores than placebo, consistent with the parent trial. Across the three candidate genes, 259 CpG sites on the EPIC array were evaluated. Baseline comparisons identified five CRHR1, two FKBP5, and three NR3C1 sites that differed between treatment groups; these were removed from subsequent predictive analyses. Correlations between salivary beta values and publicly available brain methylation beta values were strong for all three genes (NR3C1 r = 0.73; CRHR1 r = 0.88; FKBP5 r = 0.83; all p < 0.00001), indicating saliva measurements related to brain methylation patterns at these loci. When testing whether methylation change predicted PTSD symptom change across groups, 37 of 259 CpG sites showed uncorrected significant associations. Broken down by gene, before FDR correction 17 CRHR1 sites, nine FKBP5 sites (15.8% of FKBP5 sites examined, 9 of 57), and 11 NR3C1 sites (10% of NR3C1 sites examined, 11 of 111) predicted CAPS-5 change at p ≤ 0.05. After FDR correction, two CpG sites remained significant: CRHR1 cg08276280 (open sea; pre-FDR B = 42.12, p = 0.005) and NR3C1 cg01391283 (absolute change score; open sea). For FKBP5 no sites survived FDR correction; the largest FKBP5 effect cited was cg16012111 (island site; B = 32.52, p = 0.012, pre-FDR). For NR3C1, the largest raw-effect site reported was cg6222722 (open sea; B = 19.32, p = 0.003). Of the two FDR-significant predictive sites, NR3C1 cg01391283 also exhibited a significant group difference in methylation change: the MDMA group showed greater absolute methylation change than placebo at this site (ANCOVA controlling for sex, age, epithelial proportion; F(1,18) = 4.78, p = 0.042). The extracted text does not report effect-size estimates for this group comparison beyond the F statistic and p-value.

Hendricks and colleagues report the first pilot evidence linking DNA methylation changes at HPA‑axis genes with treatment response to MDMA‑assisted therapy for severe PTSD. In this small sample, methylation change at multiple CpG sites across CRHR1, FKBP5, and NR3C1 was associated with PTSD symptom reduction; after FDR correction one CRHR1 site (cg08276280) and one NR3C1 site (cg01391283, assessed as an absolute change) remained significant. Notably, the NR3C1 cg01391283 site showed a larger magnitude of methylation change in the MDMA group than in the placebo group. The authors situate these findings within the extensive literature on NR3C1 and stress biology: NR3C1 encodes the glucocorticoid receptor that mediates HPA negative feedback, and prior studies have linked trauma and PTSD to differential methylation of this gene. The investigators note that most prior work has focused on CpG islands in the NR3C1 promoter, whereas the significant sites in this pilot were in open sea regions where functional consequences are less well understood. Because methylation at different positions can have different effects (for example, promoter methylation often represses transcription while gene-body methylation may not), the authors caution against simple directional interpretations and justify their use of absolute change scores to capture predictive changes regardless of sign. Several limitations are acknowledged. The pilot sample was small (n = 23 with paired samples), limiting statistical power and the ability to control for potential confounders such as race, smoking, genetic ancestry, or detailed trauma history. Use of the Infinium EPIC array meant some previously studied CpG loci (for example, certain FKBP5 promoter/intron sites) were not assayed. Peripheral saliva samples were used, which cannot definitively represent brain tissue methylation, although saliva–brain correlations for these genes were reported as strong. Finally, the study was candidate‑gene focused because of limited power; larger, better-powered studies are needed to validate the suggested loci and to determine whether greater clinical efficacy of MDMA‑assisted therapy is mechanistically linked to larger epigenetic changes. Overall, the authors interpret the data cautiously: results support the idea that psychotherapy-related symptom improvements may be accompanied by methylation changes in HPA genes and suggest specific CpG sites for hypothesis-driven follow-up in larger cohorts. They propose a plausible model in which MDMA produces a physiological state (including transient increases in cortisol and monoamines) that could render the epigenome more labile while eliciting a psychological state conducive to therapeutic processing, but they emphasise that mechanistic links remain speculative and require further investigation.