Abstracts

Rationale: Most temporal lobe epilepsy (TLE) patients show marked hippocampal sclerosis (HS) upon pathological examination. Yet, a considerable subgroup of around 40% present with no significant cell loss but gliotic changes only. To evaluate effects of hippocampal pathology on brain structure and functional networks, we aimed at dissociating multimodal MRI characteristics in patients with HS (TLE-HS) and those with gliosis only (TLE-G). Methods: We retrospectively selected a cohort of 39 drug-resistant TLE patients (16 males; 339 years) who met the following inclusion criteria: i) evaluated with a research-dedicated 3T MRI in addition to clinical imaging, ii) underwent selective amygdalo-hippocampectomy, iii) no mass lesion (malformations of cortical development, tumor, vascular malformations), iv) no history of severe traumatic brain injury or encephalitis; v) resected specimen suitable for histological analysis based on the current ILAE HS classification scheme. Twenty patients showed hippocampal cell loss and gliosis (TLE-HS), including 10 with marked cell loss in CA1 and CA3 (HS-1), 6 with CA1 predominant (HS-2), and 4 with CA4 predominant cell loss (HS-3). Nineteen patients presented with gliosis only (TLE-G). The control group consisted of 25 age- and sex-matched healthy individuals (12 males; 328 years). At 3T, we acquired submillimetric T1-weighted images, T2-weighted images, together with diffusion and resting-state fMRI. We automatically segmented the hippocampal formation into three consistently identifiable subfields (Cornu Ammonis 1-3 (CA1-3), CA4-DG, subiculum) and extracted a medial sheet for each subfield label (i.e., a surface running along its central path), allowing for dense feature sampling with only minimal partial volume contamination. All modalities were co-registered into the same space, allowing for multi-parameter feature aggregation. Data were analyzed across hippocampal subfield surfaces, and findings were corrected for multiple comparisons using random field theory. Results: Compared to controls, TLE-HS presented with marked ipsilateral atrophy, T2-hyperintensity, and mean diffusivity increases across all subfields, while TLE-G presented with dentate gyrus hypertrophy and focal increases in T2-intensity and mean diffusivity (Fig 1). Multivariate assessment confirmed a more marked ipsilateral load of anomalies across all subfields in TLE-HS, while anomalies in TLE-G were restricted to the subiculum. Post-hoc analyses in clusters of atrophy in TLE-HS indicated high effects in HS-1 (Cohen's d=-0.99) and HS-2 (d=-0.93), while they were slightly lower in HS-3 (d=-0.76). In TLE-G, effects were marginal and rather in opposite (i.e., hypertrophic) direction (d=0.12). A between-cohort dissociation was independently suggested by resting-state functional connectivity analysis, revealing marked hippocampal decoupling from anterior and posterior default mode hubs in TLE-HS, while TLE-G did not differ from controls (Fig. 2). Back-projection connectivity analysis from cortical targets revealed consistently decreased network embedding across all subfields in TLE-HS, while changes in TLE-G were limited to the subiculum. Hippocampal disconnectivity strongly correlated to T2-hyperintensity and marginally to atrophy. Conclusions: Multimodal MRI reveals diverging structural and functional connectivity profiles across the TLE spectrum. Pathology-specific modulations of large-scale functional brain networks lends novel evidence for a close interplay of structural and functional disruptions in focal epilepsy. Funding: This research was funded by the Canadian Institutes of Health Research (CIHR MOP-57840 and CIHR MOP-123520). BCB received a CIHR postdoctoral fellowship.