Abstracts

Rationale: In partial epilepsy, seizures arise from brain regions that are thought to be fundamentally altered in their structure and function, leading to cognitive deficits observed during interictal (seizure-free) epochs. While the role of interictal spikes leading to transitory cognitive impairment is well recognized, the underlying mechanism of long-standing cognitive impairment seen in interictal periods is poorly understood. We hypothesized that recurrent neural interactions over longtime scales interictally will localize brain region/s that are functionally disconnected as evident from disruption of specific cognitive and metabolic networks. We have recently developed a focus localization algorithm (FLA) based on directed connectivity between brain sites using the measure of generalized partial directed coherence (GPDC) over a broad frequency band. By exploring the concept of effective inflow, we subsequently correlated the identified focus with neuropsychological, and FDG-PET (Positron Emission Tomography) scans. Methods: Five adults with intractable non-lesional partial epilepsy have undergone stereo depth EEG implantation targeting mesial and lateral temporal regions, cortical -subcortical structures in frontoparietal regions. Multivariate GPDC analysis (model order 7, 0.1-50 Hz and 70-110 Hz) was performed on the EEG sampled at 2KHz from all recording channels over 6 hrs awake and 6 hrs sleep EEG. Qualitative analysis of pre-operative FDG-PET scan and standard neuropsychological tests (WAIS-IV, BNT-60, CVLT-2, WRAT-3 Reading, Word Fluency, Digit span) performed interictally were analyzed and correlated with the identified maximum effective inflow brain region. Results: Clinically identified seizure onset zones (SOZ) were non-dominant mesial temporal region (N=2), dominant mesial temporal region (N=1), anterior cingulate (N=1) and dorsolateral orbitofrontal (N=1).  Regions identified by GPDC overlapped with clinically identified SOZ in 4/5 patients (80% overlap), and the results were more consistent with existence of higher gamma activity (70-110 Hz) during sleep. Hypometabolic regions identified by PET scan overlapped with all except one GPDC identified regions. There was also a correlation between cognitive deficits specific to regions identified by GPDC. Conclusions: Inter-regional connectivity analysis over long time scales during seizure-free epochs successfully identified regions that are metabolically hypoactive and part of functional networks necessary for cognitive processing. Exploration of the relation between the ictal and the identified interictal network topology should add more value to the clinical utility of this novel quantitative analysis for epilepsy. Funding: Acknowledgement: This study was supported by NSF grant OIA 1632891