Time-varying mediation analysis for incomplete data with application to DNA methylation study for PTSD

Area of Science:

• Epigenetics
• Psychiatry
• Biostatistics

Background:

• Traumatic experiences can causally influence post-traumatic stress disorder (PTSD) risk.
• DNA methylation (DNAm) is a key epigenetic mechanism mediating this relationship.
• The temporal dynamics of DNAm mediation in PTSD development remain largely unexplored.

Purpose of the Study:

• To develop and validate time-varying structural equation models for assessing dynamic DNAm mediation effects.
• To identify specific DNA methylation sites (CpG sites) involved in the trauma-PTSD pathway over time.
• To investigate the longitudinal interplay between trauma exposure, DNAm, and PTSD risk.

Main Methods:

• Development of time-varying structural equation models to capture dynamic mediation.
• Introduction of a novel Longitudinal Multiple Imputation (LMI) method to handle non-monotone missing DNAm data.
• Application of the generalized method of moments for integrating multiple imputations in longitudinal analyses.

Main Results:

• Simulations demonstrated superior performance of the proposed LMI method compared to existing approaches in longitudinal settings.
• Analysis of the Detroit Neighborhood Health Study (DNHS) data identified CpG sites with dynamic DNAm mediation effects.
• Several identified genes showed previously reported associations with PTSD, highlighting their potential role in time-varying epigenetic regulation.

Conclusions:

• The developed methodology effectively identifies dynamic DNAm mediation effects in longitudinal data.
• Findings suggest that DNAm plays a time-varying role in the causal pathway from trauma exposure to PTSD.
• This research deepens the understanding of epigenetic mechanisms underlying PTSD development and offers potential targets for intervention.