Abstracts

Rationale: Cognitive impairment is a common comorbidity of long-standing, intractable temporal lobe epilepsy (TLE), the most common form of epilepsy. Tracking and understanding how brain network alterations relate to cognitive deficits may allow for timely surgical intervention and inform resection planning. Here, we employed a novel network-modeling algorithm to identify brain network alterations correlating with cognitive performance. The seizure network was mapped for each patient and spread of that network was correlated with standardized neuropsychology evaluations of memory dysfunction. Methods: Relevant demographic information and resting state functional MRI (rsfMRI) and EEG data was attained for nineteen prospectively enrolled patients diagnosed with TLE. rsfMRI and EEG data were pre-processed; patient-specific epileptic networks were mapped; and geographic spread was quantified using our novel algorithm. A volumetric functional atlas was employed to assess global functional connectivity. Pre-operative neuropsychology testing was conducted by a blinded, board-certified neuropsychologist. A Pearson's product-moment correlation was used to investigate the correlation of key demographic data and features of epileptic networks with neuropsychological performance. Results: Seizure network localization in the temporal lobes was significantly correlated with worse performance on the Wada memory performance contralateral to seizure onset (R2 = -.639, p = .006). Increased localization of the seizure network to the contralateral temporal lobe was significantly correlated with worse scores on the contralateral Wada memory test (R2 = -.669, p = .003). Finally, increased pre-operative global connectivity was correlated with worse performance on the Wada memory test for the contralateral lobe but not the ipsilateral lobe (R2 = -.532, p = .028; R2 = -.164, p = .528). Conclusions: Seizure networks distributed in the temporal lobe contralateral from the seizure onset correlated with dysfunction in the memory circuits in that lobe. Diffuse network alterations may complicate attempts at surgical resection, and this network modeling method has potential for mapping extratemporal involvement of seizure networks and resulting cognitive impairment. Further studies correlating pre-operative network distribution with post-operative cognitive outcome are warranted. Funding: No funding