Abstracts

Rationale: Temporal lobe epilepsy (TLE) is a recognized surgically-amenable disorder, however, up to 50% of operated patients continue to have seizures. Here, we developed a novel resting-state fMRI framework that parameterized functional connectivity distance (i.e., the overall anatomical distance of an area’s functional connections), thereby consolidating functional and geometric properties of brain networks. Specifically, we examined the balance between short- and long-range functional connections in drug-resistant TLE and assessed its utility in predicting post-surgical outcome. As such, our work allowed in vivo testing of previously described experimental and histopathological pathoconnectomic models, in which epileptogenic regions show decreased long-range connectivity yet aberrant local connections. Methods: We studied 30 drug-resistant TLE patients (15 males, age=26.9±8.7 years) with histologically-verified hippocampal sclerosis and 57 age- and sex-matched healthy controls using multimodal 3T MRI. We mapped the resting-state fMRI time-series to each participant’s cortical surface and computed pairwise correlations between all pairs of regions to generate individualized functional connectomes. For each region within the z-transformed connectomes, we retained the top 10% of weighted connections and calculated the average geodesic distance to all other regions. The resulting connectivity distance maps recapitulated a region’s geodesic distance to its functionally connected areas within the brain and, as such, characterized the relationship between physical distance and functional connectivity. Results: Marked connectivity distance reductions in dorsomedial prefrontal and temporo-insular regions were observed in TLE (pFWE<0.005; FIG. 1A), with the latter having stronger effects ipsilateral to the focus in left and right TLE patients (pFWE<0.001; FIG. 1B). Patterns of distance reductions in temporo-insular cortices were driven by concurrent increases in short-range and decreases in long-range connections, indicating topological segregation of functional networks (FIG. 1C). Supervised pattern learning with 5-fold cross-validation further revealed that unfavorable surgical outcome imparts a distinct signature on connectivity distance profiles, characterized by contralateral and extratemporal distance reductions prior to anterior temporal lobectomy (accuracy=74.8±4.3%; FIG. 2). Conclusions: Our novel resting-state functional connectivity analysis consolidates physical and architectural aspects of macroscale connectomes and, as such, can provide useful markers to study focal epilepsy syndromes that are increasingly recognized to perturb whole-brain network organization. While a shift in the distribution of short- and long-range connections is generally associated to the side of the seizure focus in TLE patients, those characterized by more bilateral imbalances were found to have less favorable surgical outcome, therefore supporting the potential utility of our approach in surgical prognostics. Funding: Fonds de la Recherche du Québec – Santé (FRQ-S) Canadian Institutes of Health Research (CIHR)National Science Foundation of China (NSFC)