This preprint reports development and application of an ancestry-calibrated polygenic risk score (AC-PRS) for posttraumatic stress disorder (PTSD) in a large longitudinal cohort of recent trauma survivors (N = 1,801). The analysis tested whether the AC-PRS predicted PTSD symptom trajectories and whether neighborhood-level resources or disadvantage modified genetic effects in gene-by-environment (GxE) models.
Participants provided a blood specimen for genotyping and completed the PTSD Checklist for DSM-5 (PCL-5) at four post-trauma timepoints: 2 weeks, 8 weeks, 3 months, and 6 months. From these repeated assessments, six distinct PTSD symptom trajectories had been previously identified in this cohort. The sample is described as admixed trauma survivors, and the AC-PRS was designed to perform across ethnoracial groups.
Residential addresses were used to derive two neighborhood-level measures. Socioeconomic disadvantage was measured using the area deprivation index (ADI). Greenspace exposure was assessed via normalized difference vegetation index (NDVI). These variables served as socioenvironmental moderators in the GxE analyses.
The investigators constructed an ancestry-calibrated PTSD polygenic score intended to address ancestry-related differences in PRS scale and variance that limit cross-group comparisons. According to the report, the AC-PRS performed well across ethnoracial groups and captured genetic risk for PTSD in this admixed sample. In models predicting PTSD trajectories, the AC-PRS explained a measurable portion of variance (R2 = .053) in trajectory assignment.
Primary analyses used logistic regression to examine interactions between AC-PRS and neighborhood measures (ADI, NDVI) on assignment to PTSD trajectories. Models adjusted for sociodemographic and trauma-related covariates. Secondary analyses used linear models to consider AC-PRS × neighborhood interactions on continuous 6-month PCL-5 scores.
The AC-PRS significantly discriminated among PTSD courses, showing good discrimination between nonremitting and resilient trajectories. The reported explained variance for PTSD trajectories attributable to the AC-PRS was R2 = .053. This performance was observed across ethnoracial groups in the sample, addressing a common limitation of PRS transferability in admixed populations.
Area-level socioeconomic disadvantage (ADI) moderated the association between AC-PRS and PTSD outcomes. Specifically, ADI moderated both the likelihood of assignment to a high nonremitting PTSD trajectory and the severity of symptoms at 6 months in models that included the AC-PRS (reported p values < .05). The source reports these moderation effects but does not provide detailed effect estimates, magnitude, or stratified results in the abstract.
No significant interactions were found between NDVI (greenspace) and AC-PRS in any of the logistic or linear models reported in the preprint abstract.
The authors conclude that an ancestry-calibrated PRS can capture genetic risk for PTSD in an admixed cohort and may help discriminate chronic, nonremitting symptom courses from resilient outcomes. However, neighborhood socioeconomic disadvantage appears to modify the association between genetic risk and PTSD trajectories, indicating that context matters when interpreting PRS. The findings suggest careful consideration before applying PRS for PTSD across different social and environmental contexts.
This work is posted as a medRxiv preprint and has not undergone peer review; the source explicitly notes it should not yet guide clinical practice. The abstract provides principal results (sample size, timepoints, R2, and significance of ADI interactions) but does not report detailed statistical estimates, confidence intervals, or full model coefficients in the abstract. Replication, peer review, and examination of subgroup and sensitivity analyses in the full manuscript are needed to evaluate robustness and generalizability. Future work should also report effect sizes and consider additional socioenvironmental measures and clinical applicability.
Posted on medRxiv: July 20, 2026.
(Competing interests and funding disclosures are reported in the source preprint.)