This open-label exploratory study (n=23) examined inflammatory biomarkers in Veterans with PTSD during MDMA-assisted group therapy and found small changes in interleukin-6, tumour necrosis factor alpha and C-reactive protein. Higher baseline inflammation was linked with worse PTSD symptoms, and changes in interleukin-6 were related to symptom improvement.
Papers cited by this study that are also in Blossom
Background
Posttraumatic stress disorder (PTSD) is associated with elevated inflammation and risk for chronic illness, yet few studies have examined inflammatory biomarker outcomes of PTSD interventions. Rapid PTSD symptom reduction has been observed following 3,4-methylenedioxymethamphetamine (MDMA)-assisted therapy, which leverages MDMA as a prosocial adjunct to psychotherapy. No studies have evaluated inflammatory biomarker outcomes of MDMA-assisted therapy. This exploratory pilot study examined within-person changes in inflammatory biomarkers during MDMA-assisted group therapy for Veterans with PTSD.
Methods
Blood plasma samples were collected from 23 Veterans at baseline and end-of-intervention. Hedges' g effect sizes were calculated for interleukin-6 (IL-6), tumor necrosis factor alpha (TNF-α), and C-reactive protein (CRP). PTSD severity was assessed with the Clinician-Administered PTSD Scale for DSM-5 (CAPS-5) at baseline and 30-day follow-up. Spearman's rho correlations were calculated among biomarkers, PTSD symptoms, and change scores.
Results
Small increases were observed in IL-6 (g = 0.24; 95% CI -0.25, 0.72) and CRP (g = 0.23; 95% CI -0.30, 0.74), and a small decrease in TNF-α (g = -0.24; 95% CI -0.69, 0.23). Baseline IL-6 and TNF-α were positively associated with baseline CAPS-5 scores (ρ = 0.45, 0.32). Higher baseline IL-6 weakly predicted symptom improvement (ρ = -0.25), and IL-6 change correlated with symptom change (ρ = 0.41). CRP showed weak negative associations with PTSD symptoms (ρ = -0.26).
Conclusion
Findings suggest MDMA-assisted therapy may modulate inflammatory biomarkers and highlight biomarker-symptom relationships. Results are preliminary but may inform larger studies.
Posttraumatic stress disorder (PTSD) is associated not only with substantial psychological burden but also with poorer physical health and a higher risk of inflammatory conditions. The paper notes that prior research has repeatedly linked PTSD with elevated circulating inflammatory markers such as C-reactive protein (CRP), tumour necrosis factor alpha (TNF-α), and interleukin-6 (IL-6), but evidence on how PTSD treatments affect inflammation has been limited and inconsistent. Although MDMA-assisted therapy has shown rapid and robust effects on PTSD symptoms in earlier clinical trials, no previous study had examined inflammatory biomarker outcomes within this treatment context. The authors also frame the work against a broader background of HPA-axis dysregulation, chronic stress biology, and the possibility that symptom improvement may be accompanied by changes in immune signalling. The aim of this exploratory study was to examine within-person changes in inflammatory biomarkers during an MDMA-assisted group therapy intervention for Veterans with PTSD, and to explore whether biomarker levels were associated with PTSD symptom severity before and after treatment. The biomarkers selected were IL-6, TNF-α, and CRP, based on their prior relevance in PTSD and inflammation research. The authors expected MDMA-assisted group therapy to have anti-inflammatory effects, given the association between PTSD and inflammation and the symptom reductions seen in earlier MDMA-assisted therapy trials. This was an embedded biomarker substudy within a single-arm, open-label, pilot trial. The authors present it as the first study they are aware of to assess inflammatory biomarkers in the context of MDMA-assisted therapy for PTSD, including in a group format.
This study was embedded within a parent clinical trial-a single-arm, open-label, pilot study investigating the feasibility and safety of a 12-to 16-week MDMA-assisted group therapy protocol for PTSD (NCT05961527,. 1 This sub-study was approved by the Portland VA Institutional Review Board.
Participants were U.S. military Veterans (n = 23) recruited as part of the parent study. Eligibility criteria, further detailed in, required participants to have a charted PTSD diagnosis persisting for at least 3 months at the time of study enrollment and a concurrent score of ≥ 33 on the past-30-day PTSD Checklist for DSM-5 (PCL-5) at the time of screening.
The MDMA-assisted group therapy intervention was administered within the VA Portland Health Care System. Each of four cohorts consisted of five or six participants and a team of four co-facilitators. The first cohort consisted of women primarily with reported histories of military sexual trauma, the second and third cohorts were comprised of men with combatrelated trauma histories, and the fourth cohort was made up of Veterans with a combination of military sexual and combat-related trauma histories. The study intervention (Figure) consisted of: (1) four preparatory sessions (three group sessions and one individual), (2) one individual MDMA session, (3) four integration sessions (one individual and three group sessions), (4) one group MDMA session, and (5) four group integration sessions. The intervention and the dierent session types are further detailed in the parent study protocol, which was later revised for the final cohort-based on participant feedback from the first three cohorts-to include an additional treatment cycle (i.e., a group MDMA session, followed by four additional group integration sessions). To ensure consistency in the number of study visits and timing of data collection across all cohorts, variables of interest were measured after the same number of therapy sessions as the first three cohorts; for all cohorts, this timepoint is referred to as the end-of-intervention.
Create a free account to open full-text PDFs.
Feduccia, A. A., Mithoefer, M. C. · Progress in Neuro-Psychopharmacology and Biological Psychiatry (2018)
Illingworth, B. J. G., Lewis, D. J., Lambarth, A. T. et al. · Journal of Psychopharmacology (2020)
Lewis, C. R., Tafur, J., Spencer, S. et al. · Frontiers in Psychiatry (2023)
Mitchell, J., Bogenschutz, M. P., Lilienstein, A. et al. · Nature Medicine (2021)
Mitchell, J., Ot’alora G, M., van der Kolk, B. et al. · Nature Medicine (2023)
The biomarker analysis was nested within a parent clinical trial: a single-arm, open-label, pilot study of a 12- to 16-week MDMA-assisted group therapy protocol for PTSD. The substudy was approved by the Portland VA Institutional Review Board. Participants were U.S. military Veterans recruited into the parent study. Eligibility required a documented PTSD diagnosis for at least 3 months and a screening score of at least 33 on the past-30-day PTSD Checklist for DSM-5 (PCL-5). The extracted text reports 23 Veterans at recruitment, although one participant was later excluded because they did not complete the full parent protocol. The intervention was delivered at the VA Portland Health Care System in four cohorts, each with five or six participants and four co-facilitators. The cohorts differed somewhat in trauma history composition, including women with military sexual trauma, men with combat-related trauma, and a mixed cohort. The treatment sequence consisted of four preparatory sessions, one individual MDMA session, four integration sessions, one group MDMA session, and four group integration sessions. For the final cohort, the protocol was revised to add an additional treatment cycle, but biomarker variables were measured at the same session point as in the earlier cohorts so that timing would be standardised across groups. The paper does not provide the MDMA dose in the extracted text. Blood samples were collected at two timepoints: baseline, within 4 hours before the first therapy preparatory session, and end-of-intervention, timed before the final integration session for the earlier cohorts and the equivalent session for the later cohort. Plasma was prepared and stored before later immunoassay. Inflammatory biomarkers measured were IL-6, TNF-α, and CRP, assayed by ELISA. Duplicate measurements were required, and values were retained only if the coefficient of variation (CV) was 30% or lower; for some sensitivity analyses, a stricter CV threshold of 20% was used. The CAPS-5, a clinician-administered PTSD interview, was used to assess PTSD severity at baseline and again 30 days post-intervention, with interviews conducted by independent blinded raters. Analytically, the researchers excluded participants who did not complete the full parent protocol and log-transformed biomarker concentrations before analysis. Hedges’ g was used to estimate within-person effect sizes for biomarker change, with 95% confidence intervals reported because the study was exploratory and small. Spearman’s rho correlations were calculated to examine associations among biomarker concentrations, PTSD symptoms, and change scores. Sensitivity analyses were performed for CRP, including analyses restricted to participants with quantifiable values at both timepoints and analyses using the stricter CV threshold.
After excluding one participant who did not complete the parent study protocol, the final analytic sample was 22 Veterans. The authors state that none of the findings were statistically significant, which is unsurprising given the small exploratory sample. For biomarker change over the intervention period, the remaining IL-6 data came from 15 participants after excluding seven because of high CVs. IL-6 showed a small increase from baseline to end-of-intervention (Hedges’ g = 0.24; 95% CI -0.25 to 0.72). TNF-α was analysed in 17 participants and showed a small decrease (g = -0.24; 95% CI -0.69 to 0.23). CRP was analysed in 13 participants after excluding seven with values above the upper limit of quantification at both timepoints and two more for high CVs; CRP showed a small increase (g = 0.23; 95% CI -0.30 to 0.74). Sensitivity analyses changed the CRP estimate notably: restricting to participants with CRP values within the quantifiable range at both timepoints (n = 10) produced a larger positive effect (g = 0.61; 95% CI -0.04 to 1.22), but this appeared to be driven mainly by one participant. Removing that participant produced a negative effect (g = -0.57; 95% CI -1.12 to 0.03). Restricting to samples with CV ≤ 20% increased the IL-6 effect to a medium-sized increase (n = 9; g = 0.57; 95% CI -0.10 to 1.2), while CRP remained close to the primary finding in one analysis (n = 11; g = 0.28; 95% CI -0.29 to 0.83) and was somewhat larger when limited to within-range values (n = 9; g = 0.64; 95% CI -0.04 to 1.30). Correlational analyses found that baseline IL-6 and TNF-α were positively associated with baseline CAPS-5 scores, with higher biomarker levels corresponding to more severe PTSD symptoms (ρ = 0.45 and ρ = 0.32, respectively). CRP showed only a negligible association with baseline PTSD severity in the main analysis, but became weakly negative when restricted to quantifiable data at both timepoints (ρ = -0.26). At end-of-intervention, IL-6, TNF-α, and CRP all showed negligible associations with CAPS-5 scores measured 30 days later. For change scores, higher baseline IL-6 was weakly associated with greater symptom improvement (ρ = -0.25), whereas baseline TNF-α and CRP were mostly unrelated to symptom change; in the CRP sensitivity analysis, the association became moderately negative (ρ = -0.38), meaning higher baseline CRP was linked with greater PTSD symptom improvement. The change in IL-6 was moderately associated with change in CAPS-5 scores (ρ = 0.41), with increases in IL-6 corresponding to less symptom improvement. TNF-α change showed negligible association with symptom change, and CRP change was weakly negative (ρ = -0.26), suggesting that CRP increases were associated with symptom improvement; this remained similar in the CRP sensitivity analysis (ρ = -0.21).
The authors interpret the study as showing small, mixed biomarker shifts during MDMA-assisted group therapy, alongside several small-to-moderate associations between inflammatory markers and PTSD symptoms. They emphasise that the pattern was not statistically significant and should be viewed as preliminary, but they suggest the effect sizes may be useful for planning larger studies. Their primary interpretation is that, contrary to expectation, IL-6 and CRP increased slightly over treatment, whereas TNF-α decreased slightly. They propose that PTSD-focused therapies may sometimes transiently activate stress pathways through trauma recall and reprocessing, which could in turn influence inflammatory signalling. They also note that IL-6 and CRP often rise together biologically, which may help explain the concurrent direction of those findings. In positioning these results against earlier studies, the authors describe the literature as mixed. They say their IL-6 and CRP findings resemble some previous PTSD intervention studies that also found increases in these markers despite symptom improvement, while other studies have reported null findings or different patterns of association. For TNF-α, they note that their small decrease differs from several prior studies in which TNF-α increased during or after PTSD treatments, and they suggest this could indicate a distinct effect of MDMA-assisted therapy on TNF-α regulation, although they stress that this is speculative and requires further study. They also highlight that baseline IL-6 and TNF-α were associated with greater PTSD severity, consistent with much of the cross-sectional literature, whereas CRP relationships were weaker and in some analyses inverse. The authors discuss several reasons why biomarker levels at the end of treatment were not clearly related to PTSD symptoms measured 30 days later. One possibility is timing: biomarker samples were collected about 32 days before the follow-up CAPS-5, whereas baseline biomarker and symptom measures were much closer in time. They also suggest that physiological recovery may lag behind symptom improvement, and that total PTSD scores may obscure symptom-cluster-specific biology. They further note that stronger effects appeared in sensitivity analyses, but interpret these cautiously because the sample was very small and some findings were influenced by single participants or assay precision thresholds. The main limitations they acknowledge are the absence of statistical power, the small and variable biomarker subsamples, missing or imprecise assay data, exclusion of participants with CRP above the assay range, and the inability to account for potential confounders such as depressive symptoms, body mass index, smoking, and anti-inflammatory medication use. They also stress that the substudy had no control group and was embedded in a single-arm pilot trial, so the observed changes cannot be attributed specifically to MDMA-assisted therapy. In terms of implications, they argue that future work should include larger samples, comparator groups, longer biomarker follow-up, and more detailed modelling of biomarker-symptom relationships, including symptom clusters and relevant covariates. They also suggest that studies should examine people with comorbid inflammatory conditions, since PTSD commonly co-occurs with disorders that may alter baseline inflammatory state.
The authors conclude that this is, to their knowledge, the first study to examine inflammatory biomarker changes in the context of MDMA-assisted therapy. They state that the findings underscore the complexity and heterogeneity of PTSD-related immune and HPA-axis regulation, and that the modest biomarker changes observed could reflect random variation, measurement noise, or a true biological response to MDMA-assisted group therapy. They call for larger studies with comparator groups and more advanced analyses to clarify whether and how inflammatory processes relate to PTSD symptoms and treatment response.
Non-fasting blood sample collection occurred at the VA Portland Healthcare System via a single venipuncture into K 2 EDTA blood collection tubes (BD Vacutainer Systems, Franklin Lakes, NJ, USA) at two timepoints. For all cohorts, baseline blood samples were collected within 4 h (average: 1 h and 18 min) before the start Timeline of MDMA-assisted group therapy (MDMA-AGT) sessions and assessments. The fourth cohort (shaded) included an additional treatment cycle. To standardize timing across all cohorts, variables of interest were measured after the same number of sessions as the first three cohorts. of the first therapy preparatory session. For the first three cohorts, follow-up blood samples were collected within 4 h (average: 1 h and 27 min) before the start of the final integration therapy session. As one participant did not attend the final scheduled group integration, their sample was collected prior to the second-to-last group integration. In the fourth cohort, follow-up corresponded to before the same integration session, even though this group had an additional treatment cycle pending approximately 1 month later. Blood samples were centrifuged at 3,200 rpm for 10 min. Plasma samples were aliquoted and stored at -20 • C at the Portland VA Primary Care Laboratory and transferred within 2 weeks to a -80 • C freezer for later immunoassay.
Measured inflammatory biomarkers included the proinflammatory biomarkers interleukin-6 (IL-6), tumor necrosis factor alpha (TNF-α), and C-reactive protein (CRP). Selection of inflammatory biomarkers used in analyses was based on previously established associations with PTSD symptomsand their use in prior studies examining inflammation-PTSD intervention relationships. IL-6 and TNF-α plasma concentrations were individually assessed using high-sensitivity enzyme-linked immunosorbent assay (ELISA) kits; CRP was measured using a standard-sensitivity ELISA kit (R&D Systems, Inc., Minneapolis, MN, United States). For IL-6 and TNF-α, standard curves ranged from 0 to 10 pg/mL, both with a lower limit of quantification (LLOQ) of 0.156 pg/mL and upper limit of quantification (ULOQ) of 10 pg/mL. Reported sensitivities (lower limits of detection) for IL-6 and TNF-α were 0.09 and 0.049 pg/mL, respectively. The standard curve for CRP ranged from 0 to 50 ng/mL with a LLOQ of 0.78 ng/mL, a ULOQ of 50 ng/mL, and a reported sensitivity of 0.022 ng/mL. Immunoassays were performed per manufacturer protocols at the Psychoneuroimmunology Laboratory (VA Portland Health Care System). For all analytes, standards and plasma samples were assayed in duplicate. Absorbance was measured using a Bio-Rad Model 680 Microplate Reader (Bio-Rad Laboratories, Inc., Hercules, CA, United States). Plates were read at optical densities (OD) of 450 and 540 nm. A wavelength correction was applied by subtracting the 540 nm absorbance values from the 450 nm values, in accordance with manufacturer recommendations. Interpolating the wavelength-corrected OD values against a seven-point standard curve yielded final concentration values. Standard curves were fit using a sigmoidal four-parameter logistic (4PL) model using Prism GraphPad (San Diego, CA, United States). Curve fit quality was evaluated using the coeÿcient of determination (r 2 ), with all standard curves achieving r 2 ≥ 0.96. Following quality control procedures, plasma concentration data were retained for analysis if duplicate measurements were available for both timepoints and yielded a coeÿcient of variation (CV) ≤ 30%.
PTSD symptoms were assessed between the first and second preparation sessions (within 1 week of the initial blood draw; range: 1-6 days) and again 30 days post-intervention (approximately one to one-and-a-half months after the final blood draw; range: 25-47 days) (Figure) using the Clinician-Administered PTSD Scale for DSM-5 (CAPS-5), administered by independent, blinded raters. The CAPS-5 assesses PTSD diagnostic status and symptom severity using a structured interview format and demonstrates sound psychometric properties, including internal consistency (Cronbach's alpha = 0.91), test-retest reliability (r = 0.82), convergent validity (r = 0.59), and discriminant validity (r = 0.20-0.54).
Participants who did not complete the full parent study protocol were excluded from analyses. For all inflammatory biomarker concentration values, natural log transformations were applied prior to analyses. Participants with measurements exceeding the reliable range of quantification at both timepoints were excluded from analyses, as no meaningful change could be approximated using these values. For three participants with CRP values above the ULOQ at one timepoint, data were retained by imputing the ULOQ for the out-of-range value. While it likely underestimates variance, this conservative approach retains usable within-subject comparisons and limits bias from complete-case exclusion. Statistical analyses were performed using IBM SPSS Statistics version 30. Eect sizes were calculated using the Hedges' g statistic to correct for small sample size bias and interpreted using conventional thresholds (small ≈ 0.2, medium ≈ 0.5, large ≈ 0.8). This study emphasized eect size magnitude rather than null-hypothesis significance testing, given the exploratory nature and small sample size. To characterize estimation uncertainty, 95% confidence intervals are provided. Sensitivity analyses were conducted to assess: (1) CRP results when restricting analyses to participants with within-range concentrations at both timepoints and (2) outcomes when restricting to include samples with CV ≤ 20% at both timepoints. Spearman's rho correlations were calculated to evaluate relationships among inflammatory biomarker concentrations (baseline and end-of-intervention), CAPS-5 scores (baseline and 30 days post-intervention), and change scores for all variables, as this non-parametric measure is robust to outliers and appropriate for non-normally distributed data. Sensitivity analyses for correlations were conducted only for CRP, restricting to participants with within-range concentrations at both timepoints.
Data from one participant were excluded from all analyses due to not completing the full parent study protocol, leaving a final sample size of 22. Demographic data and clinical characteristics are reported in Table. No findings were statistically significant in this small exploratory pilot study. Tableshows descriptive statistics and eect sizes for inflammatory biomarker concentration changes.
For IL-6, data for seven participants were excluded due to high (= 30%) CVs. Analysis of the remaining data (n = 15) yielded a small increase in IL-6 (Hedges' g = 0.24; 95% CI = -0.25, 0.72) from baseline to end-of-intervention. For TNF-α, five participants had missing duplicates at one timepoint, and two of these also showed high CVs at the other timepoint. Findings for the remaining data (n = 17) demonstrated a small decrease in TNF-α (Hedges' g = -0.24; 95% CI = -0.69, 0.23). For CRP, seven participants were excluded from analyses for concentration values exceeding the ULOQ at both timepoints. Two more participants were excluded due to high CVs, yielding a final sample size of n = 13, which showed a small increase in CRP (Hedges' g = 0.23; 95% CI = -0.30, 0.74). When including only participants having CRP concentrations within-range values at both timepoints (n = 10) the eect became substantially larger in magnitude (Hedges' g = 0.61; 95% CI = -0.04, 1.22). Inspection of the data revealed that the large dierence in eect between the two sample sizes was driven primarily by a single participant; removing this participant from analyses yielded a Hedges' g eect size of -0.57 (95% CI = -1.12, 0.03). When analyses were restricted to samples with a CV ≤ 20% at both timepoints (Supplementary Table), the eect size for IL-6 (n = 9) increased, becoming medium in size (Hedges' g = 0.57; 95% CI = -0.10, 1.2). Restricting to CV ≤ 20% did not alter the TNF-α eect size, as sample size was unchanged. The eect size for CRP (n = 11) reflected primary findings (Hedges' g = 0.28; 95% CI = -0.29, 0.83), which also applied when including participants having within-range values at both timepoints (n = 9; Hedges' g = 0.64; 95% CI = -0.04, 1.30).
Spearman's rho correlations among variables of interest are shown in Table. Baseline IL-6 and TNF-α concentrations showed weak-to-moderate positive associations with baseline CAPS-5 scores (ρ = 0.45 and ρ = 0.32, respectively), where higher biomarker concentrations were associated with greater PTSD symptom severity. CRP demonstrated a negligible association with CAPS-5 scores. This association became negative and weak when including only participants for whom data was in the quantifiable CV values associated with imputed concentrations were excluded from mean CV calculations. ‡ Subset only includes participants for whom data was in the quantifiable range of the assay at both timepoints. Final N represents the number of participants with duplicate samples at both timepoints and CV ≤ 30%. For interpretability, descriptive statistics are reported as both raw data and natural log transformed values. Eect sizes and confidence intervals were calculated using natural log transformed data. range of the assay at both timepoints (ρ = -0.26), such that higher baseline CRP values were associated with lower PTSD symptom severity. End-of-intervention concentration values of IL-6, TNF-α, and CRP all yielded negligible associations with 30-day post-intervention CAPS-5 scores. When only using CRP concentrations for participants with data in the quantifiable range at both timepoints, the eect was negative and small (ρ = -0.27), where higher baseline CRP values were associated with lower PTSD symptom severity. Higher baseline IL-6 concentrations were weakly associated with PTSD symptom improvement (ρ = -0.25), whereas associations between baseline TNF-α and CRP concentrations and change in CAPS-5 scores were negligible. For CRP, the relationship became negative, approaching a moderate magnitude when limited to participants with quantifiable data at both timepoints (ρ = -0.38), where higher baseline CRP was associated with greater PTSD symptom improvement. Finally, a moderate association was found between the change in IL-6 concentration and change in CAPS-5 score (ρ = 0.41), in which increased IL-6 concentration over the intervention was associated with less PTSD symptom improvement. The association was negligible between the TNF-α and CAPS-5 change scores. For CRP, the relationship was weak and negative (ρ = -0.26), where increases in CRP over the intervention period were associated with PTSD symptom improvement. The finding was consistent when only using concentration values for participants with data in the quantifiable range at both timepoints (ρ = -0.21).
In this exploratory study, we examined the concentrations of inflammatory biomarkers across the intervention period of a single-arm, open-label, pilot trial of a 12-16-week MDMAassisted group therapy protocol for PTSD. Primary analyses yielded small increases in IL-6 and CRP, and a small decrease in TNFα, from before to after MDMA-assisted group therapy for PTSD. Sensitivity analyses produced outcomes generally consistent with, and sometimes larger than, primary analyses. We also found several small-to-medium-sized associations of varied directionality among biomarker concentrations and PTSD symptoms. These included associations observed at baseline, as well as links between baseline biomarker levels and symptom change, and between changes in biomarkers and changes in symptoms. No findings reached statistical significance in this small exploratory pilot study; however, these eect sizes may contribute to future research designs with larger samples. Contrary to our expectations, IL-6 and CRP demonstrated small increases over the intervention period, which further increased when stricter criteria were applied during sensitivity analyses. As PTSD-focused interventions generally involve exposure to and reprocessing of memories or reminders of traumatic events, they may confer a paradoxical eect on inflammation through repeated activation of the stress response over the course of treatment. The concurrent elevations we observed in IL-6 and CRP concentrations align with the well-known relationship in which IL-6 stimulates CRP synthesis, thus the two are often observed as increasing together (Del Giudice and Gangestad, 2018). Though we anticipated decreases in IL-6 and CRP, our findings were similar to, who tested a yoga intervention for PTSD and observed greater IL-6 increases with yoga relative to cognitive processing therapy, as well as CRP elevations in both groups 3 months post-intervention, all alongside PTSD symptom improvements.also reported significantly increased IL-6 and CRP in response to both pharmacotherapy treatment with sertraline and interpersonal psychotherapy adapted to PTSD at 1 year follow-up, despite PTSD symptom improvements. In contrast, another study reported no overall dierences in IL-6 and CRP before and after both mindfulness-based stress reduction and active control interventions for PTSD, though IL-6 and CRP increases were both significantly associated with decreased PTSD symptom severity. We found diverging patterns between changes in IL-6 and CRP concentrations relative to PTSD symptom severity. Increases in IL-6 were moderately associated with smaller reductions in PTSD symptoms, whereas increases in CRP were weakly associated with greater PTSD symptom reduction. The mixed findings among our study and others may reflect dierences in study design, sample characteristics, and intervention type, as well as the complex mechanisms regulating immune and stress responses. Stronger eects for both IL-6 and CRP emerged in sensitivity analyses. For IL-6, the eect size increased when restricting analyses to samples with ≤ 20% CV values, indicating that the eect strengthened with greater assay precision. For CRP, the eect size increased when only including participants with concentrations in the quantifiable range at both timepoints (n = 10), as opposed to including those for whom the high standard was imputed at one of the two timepoints (n = 13). The dierence in eect between the two sample sets appears to have been influenced disproportionately by a single participant whose baseline CRP value was imputed with the high standard. This individual showed the greatest reduction in CRP concentration across the entire sample, despite only a onepoint change in CAPS-5 score. Excluding this individual from analyses (n = 12) produced an eect size consistent with the n = 10 sensitivity analysis. While the eect size increases for IL-6 and CRP are notable, interpretation is limited by small sample size. Contrasted with the increases in IL-6 and CRP, and aligned with our expectations, we observed a small decrease in TNFα concentration from before to after MDMA-assisted group therapy. This decrease was directionally consistent with an earlier preliminary analysis on a subset of 10 participants (Hedges' g = -0.53; 95% CI = -1.16, 0.13), prior to completion of full assays. Interestingly, this finding diers from those of other PTSD intervention studies where TNF-α concentrations increased during psychotherapyand 1 year after either pharmacotherapy or psychotherapy. Although this contrast should be interpreted cautiously, it raises the possibility that MDMA-assisted therapy may exert distinct eects on TNF-α regulation, warranting further study. Consistent with numerous cross-sectional studies, elevations in IL-6 and TNF-α were associated with greater PTSD symptom severity; however, these associations were only observed at baseline. End-of-intervention biomarker levels showed negligible associations with 30-day CAPS-5 scores, possibly attributable to a ∼32-day gap between end-ofintervention biomarker data collection and follow-up assessments. Conversely, there was an average of approximately 4 days between baseline biomarker data collection and baseline CAPS-5 assessment. Physiological recovery and immune re-regulation may occur more slowly than PTSD symptom reduction, and symptom improvements may have continued during the 32 days post-intervention as participants integrated therapeutic gains. Additionally, total CAPS-5 scores may have obscured nuanced associations, as biomarker changes may dier by symptom cluster (e.g., hyperarousal may align more closely with inflammation via HPA-axis reactivity). Although many studies report positive associations between CRP and PTSD symptoms, our findings of a negligible association between CRP and PTSD symptoms at baseline were consistent with those of. Of note, a small negative correlation emerged following sensitivity analysis, with higher CRP concentrations weakly associated with lower PTSD symptoms. The relationship was maintained when examining the association between endof-intervention CRP concentration and 30-day post-intervention PTSD symptoms. We hope our findings will inform further research investigating changes in inflammatory biomarkers as potential physiological mechanisms of change for MDMA-assisted therapies. If these mixed findings in pro-inflammatory biomarker concentration changes are replicated across future studies, they may suggest complex modulation of inflammatory processes across treatment. Although we anticipated decreases in the pro-inflammatory biomarkers IL-6, TNF-α, and CRP across the intervention, it is possible that increases in IL-6 and CRP may correspond to trauma re-processing and integration, as well as a period of recalibration for nervous system regulation. To better understand the trajectory of inflammation in the context of MDMA-assisted therapy, future studies should measure inflammatory biomarker levels over an extended period, following treatment completion. Given the higher rates of inflammatory diseases observed in individuals with PTSD, it will also be important to examine inflammatory biomarker concentration trajectories in individuals with comorbid PTSD and inflammatory conditions (e.g., autoimmune disorders), as well as in those with conditions associated with chronic inflammation (e.g., substance use disorders). Our novel exploratory study has several limitations. First, the study was not powered to detect significant eects, and biomarker research generally requires larger sample sizes; therefore, all findings should be interpreted with caution. Additionally, the limited sample size precluded the incorporation of factors associated with systemic inflammationsuch as depressive symptoms, body mass index (BMI), smoking status, and anti-inflammatory medication use)-into analytic models, limiting mechanistic inference. Furthermore, we were unable to use data from several participants for whom CV values exceeded 30%. Sample sizes were therefore smaller and varied across inflammatory biomarkers, further limiting statistical power. For CRP, concentration values for seven participants surpassed the upper limit of quantification at both timepoints. This is consistent with prior evidence that individuals with PTSD are more likely to exhibit clinically elevated CRP levels than those without the disorder. These participants were excluded from analyses, which limits interpretation, as individuals with values outside the quantifiable range may exhibit dierent patterns of change than those with in-range data at both timepoints. Exclusion of these data further limits statistical power, as well as the generalizability of findings to individuals with higher concentration values. Finally, as this was an exploratory sub-study that occurred within a single-arm pilot trial, there was no control group against which to compare outcomes. As such, findings cannot be attributed specifically to the MDMA-assisted therapy intervention.
To our knowledge, this is the first study examining changes in inflammatory biomarkers in the context of MDMA-assisted therapy. Taken together, our findings highlight the complexity of PTSD-inflammation relationships, which are shaped by substantial heterogeneity in both HPA-axis and immune modulation. PTSD symptoms themselves are also heterogeneous; for example, diagnostic subtypes have been associated with dierential eects on neuroendocrine function. Further compounding complexity, intervention studies vary in design, modality, and sampling strategy, often yielding mixed results. Additional research employing comparator groups and larger samples will be needed to determine whether these modest changes reflect random variation, measurement noise, or a true biological response to MDMAassisted group therapy. A larger study sample would also permit more advanced modeling, including examination of biomarkersymptom cluster relationships while controlling for demographics and covariates such as depression symptoms, body mass index, and use of anti-inflammatory medications. These approaches could contribute to a clearer understanding of how inflammatory processes relate to PTSD symptoms and whether they are influenced by MDMA-assisted therapy. (#I01 CX002668), and the Tartar Trust Fund (OHSU School of Medicine), which contributed to immunoassay materials. The parent study was funded by the Steven and Alexandra Cohen Foundation, the Bronner Family Foundation, and the Lars Reierson Psychedelic Medicine Innovation Fund. Study drug was provided by Lykos Therapeutics. This study was the result of work supported with resources and facilities at the VA Portland Health Care System (VAPORHCS) and Oregon Health and Science University (OHSU), Portland, OR. JL, and CS were employed as Research Scientists at VAPORHCS. The contents do not represent the views of the U.S. Department of Veterans Aairs or the United States Government.