Staff Scientist Medical University of South Carolina
Rationale: Temporal lobe epilepsy (TLE) has been shown to involve changes across brain regions located not only adjacent but also remotely from the area of ictal onset. Converging evidence from different methodological approaches, such as histopathology, functional, and structural connectivity, and volumetric analysis thus support the idea of epilepsy as a disorder of neural networks. The goal of this study was to implement a new tractography approach, “connectometry analysis,” to examine abnormalities in white matter tracts in patients with left and right TLE relative to healthy controls as well as in relationship to clinical phenotypes (namely, post-operative seizure outcomes). Methods: One hundred seventy-four participants (64 left TLE, 55 right TLE, 48 controls, seven patients with no information about side) were included from three different sites (MUSC, Emory University, Bonn University). Group connectometry analyses (Yeh et al. 2016) were performed using DSI-Studio (https://dsi-studio.labsolver.org). A T-score threshold of two was assigned to select local connectomes using a deterministic fiber tracking algorithm. The quantitative anisotropy values were normalized to account for variability in scanning protocols across sites. All tracks generated from bootstrap resampling were included. A false discovery rate (FDR) threshold of 0.02 was used to select fractional anisotropy (FA) and mean diffusivity (MD) tracks showing significant differences between groups. Results: Connectometry analysis revealed both lower and higher FA and MD values in a wide-spread bilateral temporal- and extra-temporal network in patients with TLE independent of seizure onset side compared to controls. Using non-parametric Spearman correlations, we investigated whether post-operative seizure freedom in TLE could be predicted based on FA and MD values of the defined tracks. Patients with left TLE and non-seizure freedom had lower MD values in ipsilateral Cortico Spinal and Corticopontine tracts. Patients with right TLE and no seizure-freedom had higher FA values in the bilateral Cingulum and Corpus Callosum. Conclusions: The present study illustrates the analytical advantage of using connectometry analysis to identify white matter fiber tracts in identifying network abnormalities in patients with TLE relative to controls. Moreover, connectometry is capable of identifying tracts related to distinct clinical phenotypes, namely, post-operative seizure freedom. Specifically, our results reveal FA and MD changes in a wide-spread temporal and extra-temporal network in patients with TLE, supporting the growing body of evidence for epilepsy being a network disease. This study demonstrates that connectometry analysis can complement conventional region of interest or tractography approaches. Understanding white matter fiber tracts affected in TLE may also contribute to models of preoperative assessment and planning. Funding: Please list any funding that was received in support of this abstract.: R01NS110347 (NIH/NINDS) - Predicting Epilepsy Surgery Outcomes Using Neural Network Architecture. R21 NS107739 (NIH/NINDS) - Prediction of seizure lateralization and postoperative outcome through the use of deep learning applied to multi-site MRI/DTI data: An ENIGMA-Epilepsy study.
