Synaptic dysfunction-related gene expression spatially correlates with widespread cortical network atrophy in temporal lobe epilepsy.

Temporal lobe epilepsy (TLE) exhibits marked lateralized gray matter alterations, yet whole-brain network vulnerability patterns, particularly those independent of seizure laterality, remain incompletely understood. Furthermore, the spatial correspondence between macroscopic network disruptions and underlying molecular architectures lacks systematic characterization, limiting mechanistic insights into epileptic network pathology. We utilized voxel-based morphometry (VBM) and surface-based morphometry (SBM) to characterize whole-brain gray matter alterations in 69 TLE patients (34 left-TLE, 35 right-TLE for lateralization analysis) and 47 controls. Twelve multivariate machine learning algorithms were employed to validate the robustness and importance of structural features. Additionally, imaging-transcriptomics analyses using the Allen Human Brain Atlas were conducted to examine spatial associations between gene expression profiles and brain morphological alterations. Beyond lateralization-specific atrophy, we identified bilateral somatomotor and default mode networks, along with the right ventral attention network, as primary vulnerable networks independent of seizure origin (all PFDR < 0.05). Machine learning confirmed right somatomotor network atrophy as a structurally robust signature of TLE. Imaging-transcriptomics analysis revealed a striking correspondence between cortical network atrophy and synaptic genes related to signaling and structural organization. Specifically, we uncovered distinct spatial covariation profiles among four synaptic hub genes: while CTNNB1, NLGN1, and GRM5 exhibited positive spatial correlations with atrophy patterns, SNAP25 displayed an inverse relationship. This study delineates lateralization-independent vulnerability networks in TLE and characterizes their spatial-molecular architecture. By linking macroscopic atrophy to microscopic synaptic signatures, our findings present a structural-molecular framework that offers insights for future mechanistic studies and highlights potential targets for synapse-based precision therapeutic strategies.