Microstructural Integrity of Hippocampal-Posterior Cortical White Matter Is Associated With Phenomenological Properties of Trauma-Related Intrusive Memories.

BACKGROUND

Trauma-related intrusive memories (TR-IMs) are spontaneous and emotionally intense sensory recollections experienced as occurring in the here and now and a core feature of posttraumatic stress disorder (PTSD). Although central to symptom burden, the neurobiological substrates of specific phenomenological properties of TR-IMs remain poorly understood. Prior work has linked dynamic hippocampus (HPC) coactivation with default mode and visual networks to TR-IM properties, but whether structural connectivity tracks with these properties is unknown.

METHODS

In 114 symptomatic trauma-exposed adults (87 female), 2 weeks of ecological momentary assessment surveys captured TR-IM properties (intrusiveness, reliving, visual detail, vividness, emotional intensity). Diffusion-weighted imaging was used to quantify fractional anisotropy (FA) in 2 HPC-posterior cortical pathways: the parahippocampal-parietal cingulum, connecting the HPC to posterior default mode regions, and the inferior longitudinal fasciculus (ILF), linking the medial temporal regions to the visual cortex.

RESULTS

Bayesian models incorporating functional coactivation priors revealed that lower FA in the parahippocampal-parietal cingulum was associated with greater TR-IM intrusiveness, while lower FA in the ILF was linked to heightened reliving. Associations were strongest when weighted by prior dynamic connectivity findings in this sample.

CONCLUSIONS

These findings suggest that separable HPC-posterior cortical white matter pathways support different TR-IM properties, with the cingulum related to their intrusiveness and the ILF with reliving. This study identifies white matter correlates of TR-IM phenomenology and demonstrates the value of integrating real-world memory sampling with anatomically informed Bayesian modeling, thereby advancing the mechanistic understanding of PTSD reexperiencing.