205 | Structural connectivity alterations and Contributions to prediction biomarkers in drug-resistance Epilepsy

Disorders of the Nervous System

Author: Mariana Nahir Vallejo-Azar | Email: mvallejoazar@unaj.edu.ar

Mariana Nahir Vallejo-Azar , Juan Pablo Princich , Paula Gonzalez

1° Unidad de Estudios en Neurociencias y Sistemas complejos (ENyS), CONICET- Hospital SAMIC El Cruce- Universidad Nacional Arturo Jauretche

Epilepsy is a chronic and frequent disease. One-third of the patients have drug-resistance, most commonly in the form of Temporal Lobe Epilepsy with Hippocampal Sclerosis (TLE-HS) or Focal Cortical Dysplasia (FCD). Previous studies reported network alterations, but few have compared both types. This study aimed to characterise the global network topology and structural connectivity in TLE-HS, FCD and healthy controls (HC) using diffusion-weighted images (DTI). 115 adults of both sexes (FCD= 20, TLE-HS= 48 and HC= 47) were scanned on a 3T MRI scanner. Weighted connectivity matrices were obtained in DSIstudio. The epilepsy effect on network topology was assessed by general linear models. Sex, age, lateralisation and duration of disease were included as covariates. Results revealed clustering and small-wordness coefficient, transitivity and global efficiency decrease in FCD. The predictive models demonstrated loss of fractional anisotropy in patients, using linear discriminant analysis in NBS predict. Connectivity decrease in right TLE-HS was associated with the duration of epilepsy, particularly in subcortical edges. The model of left TLE-HS revealed significant performance as a classifier, using edges linking temporal and orbitofrontal regions in the ipsilateral hemisphere. FCD model suggested a tendency for connectivity loss. Overall a distinct connectivity reorganization related to the type of epilepsy. The predictive models could be clinically relevant as classifiers.