Post-ictal Dynamic Connectivity Across Seizure Types in Temporal Lobe Epilepsy

Introduction: Approximately 30% of epilepsy cases are drug-resistant [1]. For these patients, surgical intervention remains the most effective treatment, but its success relies on accurate localization of the seizure onset zone (SOZ), which can be costly, time-consuming, and uncertain. Stereotactic electroencephalography (SEEG), the current gold standard, fails to localize the SOZ in 10-15% of patients [2], prompting ongoing development of neuromodulation therapies. Our previous work demonstrates that network connectivity to the SOZ collapses during the late ictal period [3], though how this network reorganization extends into the post-ictal period remains unclear. Here, we delineate post-ictal connectivity profiles across three seizure types: focal aware seizures (FAS), focal impaired awareness seizures (FIAS), and focal-to-bilateral tonic-clonic (FBTC) seizures. We hypothesized that post-ictal connectivity patterns differ by seizure type, with FBTC seizures showing the most prolonged network disruption.

Methods: We obtained SEEG data from 78 patients with drug-resistant epilepsy undergoing presurgical evaluation at the Vanderbilt EMU. Channels were assigned as SOZs, non-involved zones (NIZs), or propagation zones (PZs) based on epileptologist designations. Directed dynamic connectivity was calculated using partial directed coherence (PDC) and isolated for the post-ictal period, which we defined as the ten minutes following seizure offset. Inward, outward, and net PDC were then computed from all channels to the PZ, SOZ, and NIZ relative to the interictal period. Relative net PDC was then stratified across FAS, FIAS, and FBTC. Group-level differences across subjects were computed using one-way ANOVAs, and significance was assessed using Bonferroni-Holm correction for multiple comparisons. Protocols were approved by the Vanderbilt IRB.

Results: We observed that FBTC seizures had greater net PDC to the PZ than FAS or FIAS within the first 200 seconds in the post-ictal period (one-way ANOVA, p < 0.01). In addition, FBTC seizures exhibited greater net PDC to the NIZ than FAS (one-way ANOVA, p < 0.05) but not FIAS within the first 200 seconds. No significant differences between seizure types were observed for net PDC to the SOZ.

Conclusion: Higher post-ictal net PDC in FBTC seizures may reflect continued disruption of connectivity to the SOZ, causing network interactions to be redirected through the PZ. This is consistent with the broader network disruption observed in FBTC seizures. In contrast, FIAS and FAS may have a faster recovery of the interictal network profile. These findings suggest that modulating post-ictal connectivity by targeting the PZ might reduce impairment or recurrence, informing new neuromodulation strategies for treating the post-ictal state.