Rationale:
Surgical management of medically refractory epilepsy relies on accurate localization of seizure onset zone (SOZ) and spread. Interictal epileptiform discharges (IEDs), captured in artifact free resting iEEG data, are used in surgical planning and leveraged by clinicians to localize SOZ and spread. The underlying neurophysiology of spread and its relation to outcomes remains unclear. One hypothesis relates spread of the ictal and IED activity to the functional connectivity of the networks. IED channels can be modularized into communities of spiking networks based on cospike probabilities (CP). This study tested the match between SOZ modules identified by our algorithm (a-SOZ) and by clinician identified ictal channels (c-SOZ), functional connectivity between modules and their relationship to CP, and the correlation between out-module connectivity strength and likelihood to have a generalized seizure.
Methods:
Participants underwent surgical placement of intracranial electrodes for seizure monitoring.
Asymmetric cospike probability (CP) was calculated for each channel pair and assigned to a module. Figure 1 shows algorithm and SOZ-module identification. Matching between c-SOZ and a-SOZ was compared against a shuffled null. Functional connectivity was calculated using normalized granger causality (GCW). To assess connectivity relationship to spread, within module GCW was compared between a-SOZ modules and non-SOZ modules, and all module GCWs were correlated to module CP (Figure 2). To assess how seizure propagation can utilize functionally connected networks, out-module GCW of the a-SOZ was correlated to patient percentage of generalized seizure (GTCs).
Results:
Data from 11 patients underwent analysis. SOZ module match between a-SOZ and c-SOZ is 81% compared to 27% of with shuffled data (Figure 1), demonstrating that IEDs offer a proxy for epileptic networks that are clinically relevant in identifying the SOZ. We pose that IEDs, at least in part, utilize functional connectivity to spread both within and across modules (Figure 2), shown by significantly correlated module GCW to module CP (r = 0.45 p= 0.001), and by the increased GCW within a-SOZ compared to non a-SOZ modules (p= 0.04). Further, SOZ out-module GCWs significantly correlates with patient GTC percentage (r= 0.65 p= 0.03) and this correlation strengthens in the subset (n= 5) of only patients who have had a GTC (r= 0.92 p= 0.02). This offers that these functional connections between modules allow for spread during seizures.
Conclusions:
Our data suggests that these modules can proxy epileptic networks and that the underlying functional connections allows both IEDs and seizures to propagate out of, and within, those networks.
Funding: The National Institutes of Health (NIH) Medical Research Scholars Program