Abstract
Abstract
Based on an edge-centric functional network model, this study investigated the longitudinal evolution of overlapping network structures after mild traumatic brain injury and explored their potential molecular basis through transcriptomic analysis. Sixty-four acute-phase patients with mild traumatic brain injury and thirty-two healthy controls were recruited, and thirty-one patients completed the chronic-phase follow-up. Resting-state functional magnetic resonance imaging data and neuropsychological assessments were acquired, and correlations between imaging measures and neuropsychological scores were analyzed. Community similarity (S-index) was quantified from an edge-centric connectome. Subsequently, partial least squares (PLS) regression, functional enrichment, and cell-type enrichment analyses were performed to examine the relationships between regional S-index changes and gene expression patterns. Compared with patients in the acute phase, patients in the follow-up phase showed significant improvement in cognition during the chronic phase. The S-index significantly decreased in a significantly connected component comprising seven edges and seven nodes (the right olfactory cortex, right parahippocampal gyrus, right calcarine cortex, bilateral lingual gyri, and bilateral postcentral gyri) from the acute phase to the chronic phase (Network-Based Statistic-corrected,
P
= 0.027). However, no significant correlations between neuropsychological scores and S-index were detected after false discovery rate correction. Separately, to examine the whole-brain transcriptional correlates of regional S-index changes, each region’s S-index was averaged across all its connections and entered into PLS regression against regional gene expression profiles. PLS1 explained 14.98% of the variance in regional S-index changes and was significantly correlated with the longitudinal regional S-index t map (
r
= 0.412,
P
spin
= 0.020). Specifically, the longitudinal change in regional S-index was spatially coupled with the PLS1
−
gene sets, which were functionally enriched in synaptic architecture and signaling pathways, with excitatory and inhibitory neurons also enriched. In contrast, the increase in S-index was linked to the PLS1+ gene sets; these genes were functionally characterized by metabolic adaptation, developmental regulation, and cytoskeletal remodeling, while astrocytes, microglia, and oligodendrocyte precursor cells were enriched. These findings identify a longitudinal imaging feature of overlapping network reorganization and provide preliminary transcriptomic and cellular hypotheses concerning its molecular basis.