Abstract The men and women who worked in rescue and recovery operations at the 9/11 World Trade Center site are developing cognitive impairment (CI) at mid-life, decades before CI is usually detected. To date, one of the most consistent risk factors for CI in this population is symptoms of post-traumatic stress disorder (PTSD). However, little is known about the mechanistic cascade that drives stress-related neurological changes to accelerate cognitive decline in the human brain. We used machine learning to identify distinct brain signatures from functional magnetic resonance imaging between trauma-exposed healthy controls (TEHC; N = 30; 21 men), PTSD without CI (PTSD-CI; N = 19; 16 men), and PTSD with CI (PTSD + CI; N = 22; 18 men). We compared the spatial gradient of each functional signature to the distribution of mRNA expression in the brain. We applied structural equation modeling (SEM) to infer mechanistic cascades specific to each group. While modest accuracy was achieved for the PTSD–CI versus TEHC signature (0.67), clear differentiation was observed for PTSD + CI versus TEHC (0.73) and PTSD + CI versus PTSD–CI (0.85). Consistent significant correlations were found between PTSD + CI signatures and ZNF48 , TOMM40 , and GRIN1 expression distributions. The cortisol-induced neurotoxicity pathway was consistently found with the PTSD + CI signature, while the p53 signaling pathway was observed across all PTSD signatures. Our results reinforce peripheral biomarkers from a previous transcriptomic study and suggest functional biomarkers in PTSD and PTSD-related CI. Furthermore, our SEM results suggest that PTSD and PTSD-related CI may diverge at the mechanistic level, with neurotoxicity being specific to CI.
Kuang et al. (Thu,) studied this question.