Abstracts

Rationale: Genetic generalized epilepsy (GGE) is a collection of non-lesional epileptic disorders (Scheffer et al. 2018) characterised by widespread abnormal bilateral brain networks (Zhang et al. 2011). Around 30% of patients with GGE are refractory to anti-epileptic drug (AED) treatment (Gelisse et al. 2001). The mechanisms underlying persistent seizures after AED treatment are poorly understood. Sophisticated MRI techniques can be used to model brain networks with the opportunity of developing biomarkers for treatment outcome. Here, we characterize structural brain network alterations in patients with GGE and determine whether brain network alterations differ between patients with well-controlled and poorly-controlled seizures. Methods: Thirty-four patients with GGE (11 seizure free and 23 persistent seizures) and 39 age- and sex-matched healthy controls were studied. Structural connectivity networks were built from T1-weighted and diffusion weighted MRI data. For network nodes, 82 cortical and subcortical regions were segmented from T1-weighted images using FreeSurfer software. Network connection strengthes between regions were reconstructed using the diffusion MRI data for mean diffusivity, radial diffusivity, fractional anisotropy and count (number of streamlines). Diffusion MRI data was corrected for artefacts using the ENIGMA pipeline and DSI studio was used for tractography. Differences between GGE seizure free, GGE persistent seizures and control groups were computed using Network Based Statistics (NBS). Results: Statistically significant bilateral network differences were observed between patients and controls. Network changes were dependant on the particular diffusion-based parameters analyzed. All patients had decreased fractional anisotropy and count measures of connectivity across both cerebral hemispheres compared to controls. Only patients with persistent seizures had additional bi-hemispheric mean diffusivity and radial diffusivity based network alterations relative to healthy controls (figure 1); patients who were rendered seizure free did not show evidence of such network changes. No differences were found in direct comparisons between patients rendered seizure free and those with persistent seizures. Conclusions: Although a non-lesional disorder, network analysis revealed bi-hemispheric structural network alterations in patients with GGE, which may provide some insight into the mechanisms of the disorder. Given that particular network alterations of mean diffusivity and radial diffusivity were unique to patients with persistent seizures, network analysis may be useful for identifying imaging biomarkers of refractory GGE. Funding: Funding: SSK: UK Medical Research Council (Grant Numbers MR/S00355X/1 and MR/K023152/1)