Potential Mitochondria-Related Key Genes in Post-Traumatic Stress Disorder Analyzed by Machine Learning Methods

Post-traumatic stress disorder (PTSD) is a debilitating psychiatric disorder triggered by exposure to traumatic events. Emerging evidence suggests that mitochondrial dysfunction may contribute to PTSD pathogenesis by disrupting cellular energy metabolism, increasing oxidative stress, and impairing neuroplasticity. This study investigates mitochondrial dysfunction-associated biomarkers, potentially opening new avenues for targeted therapeutic approaches.

Gene expression matrices from datasets GSE199841 and GSE81761 were derived from peripheral blood samples, used to identify differentially expressed genes (DEGs) between PTSD patients and healthy controls. Functional annotation and enrichment analysis were carried out using Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses. Next, least absolute shrinkage and selection operator (LASSO), support vector machine recursive feature elimination (SVM-RFE), and random forest were employed to screen and prioritize potential biomarkers, followed by validation through receiver operating characteristic (ROC) analysis and independent cohort from human verification. We also examined the expression of mitoDEGs in peripheral blood from PTSD-related mouse models by RT-qPCR. Given the close interplay between mitochondrial metabolism and immune function, we investigated the relationship between key MitoDEGs and immune infiltration proportions.

DEGs in PTSD were enriched in mitochondrial pathways, including mitochondrial ribosome function and nitric oxide synthase regulation. Machine learning identified UCP2, CISD1, NADK2 and IDE as key MitoDEGs. Then we assessed the diagnostic performance of four key genes through ROC curve analysis. We evaluated four key genes using ROC analysis, showing good diagnostic performance in the discovery cohort (AUC=0.871). Results were replicated in validation cohorts (GSE81761 AUC=0.745; GSE97356 AUC=0.638). These genes correlated with immune cell proportions (regulatory T cells, naïve B cells, CD4+/CD8+ T cells) and showed conserved dysregulation in murine blood, aligning with human data.

Mitochondrial-related genes UCP2, CISD1, NADK2 and IDE may serve as cross-species diagnostic biomarkers for PTSD, with potential links to neuroimmune mechanisms.