Abstracts

Rationale: Successful epilepsy surgery requires accurate identification of the epileptogenic zone (EZ). While dipole localization of interictal epileptiform discharges (“spikes”) can identify the irritative zone, this is not always within the EZ. Using functional connectivity to characterize the spike-associated network could improve EZ localization, subsequent surgical planning, and outcomes. Methods: We retrospectively identified eight individuals with resting-state MEG recordings with well-formed spikes who had subsequently undergone intracranial electrographic monitoring. Individual spikes were manually marked by a board-certified epileptologist trained in clinical magnetoencephalography. Whole-brain connectivity matrices using the Human Connectome Project Multimodal Parcellation anatomic atlas were generated in MNE-Python (v0.17.0) using the imaginary part of coherence during one-second epochs prior to and during each spike. The connectivity matrices imported into the Network-Based Statistic toolbox (NBS; v1.2), and permutation testing was used to identify significant spike-associated networks. Results: In individuals with at least 20 spikes, a significant spike-associated network (p<0.01 for all spike populations across a range of thresholds) was identified with a concentration in the same region predicted by spike source localization. All individuals subsequently underwent intracranial monitoring, and the extent of the IED-associated network overlapped with the brain regions involved in spike and seizure generation as determined by intracranial recording (Figure 1). Depending on the NBS threshold selected, the spike-associated networks demonstrated widespread cortical involvement far beyond the irritative, and likely epileptogenic, zones. Dynamic functional connectivity revealed connectivity changes briefly preceding, and persisting beyond, the spikes themselves (Figure 2). Conclusions: Spike-associated networks were concentrated around spike sources and included other electrographically-active regions seen during intracranial monitoring. Notably, the networks also extended beyond regions included in the intracranial recordings, which might explain the more widespread deficits commonly seen in individuals with focal epilepsy (the “functional deficit zone”). Spike-associated network analysis could provide insights into the underlying epileptic network beyond traditional dipole localization. Funding: NIH NINDS K12-NSADA (1K12NS089417-01)