Abstracts

Rationale: Intracranial electroencephalography (iEEG) is used in refractory epilepsy to localize seizure onset zones (SOZ) prior to surgical intervention. In ideal cases, patients will experience clinical seizures (CS) during iEEG, but some patients may only experience subclinical seizures (SCS), in which seizures have no behavioral signs and are only visualized on iEEG. The presence of SCS during iEEG has previously been associated with a longer disease course, younger onset of CS, and increased frequency of CS (Jin et al., 2017, Int. J. Neurosci.). However, electrophysiological characteristics of SCS compared to CS are incompletely explored. We investigated differences in how SCS and CS differ in onset relative to the SOZ using different frequency-based seizure detection approaches.

Methods: Forty-five patients at the University of Utah who underwent iEEG for refractory epilepsy between 2018 and 2021 were included in this study. Seizure reduction ratings for each patient were recorded post-intervention (median days, 753) by their epileptologist. During the patient’s iEEG monitoring, the types of seizures, whether CS or SCS, and SOZ were identified by the treating epileptologist. Seizure onset times were algorithmically determined by thresholding the spectral power during an ictal discharge. Frequencies analyzed were beta (12-30 Hz), high gamma (70-150 Hz), phase-locked high gamma, and dominant (frequency of maximum power). The first 4, first 8, and all detected channels using from CS and SCS EEG were compared to epileptologist-determined SOZ channels._x000D_
Results: Thirteen patients were identified to have both SCS and CS, and 32 patients had only CS during their IEEG. Patients with both SCS and CS vs. CS only had no significant difference in seizure reduction rates (Wilcoxon signed-rank test: p=0.45) regardless of therapy. Patients with resection and ablation therapies had significantly improved outcomes compared with patients with neuromodulation therapies regardless of clinical or subclinical presentation (Wilcoxon signed-rank test: p=0.003). In a subset of 6 patients with both SCS and CS, we compared clinically determined SOZ channels with SOZ channels detected algorithmically across frequency bands. Across different detection schemes, localized SOZ from SCS and CS were not significantly different using pairwise comparisons (Pairwise t-test: p >0.05)._x000D_
Conclusions: Although unfavorable outcomes have been previously reported in patients with both SCS and CS, we found that patients with both seizure types in this cohort had improved outcomes (non-significant) than patients with only CS. As expected, there were significantly improved outcomes for resection over neuromodulation regardless of seizure manifestation. Applying automated SOZ detection algorithms to CS and SCS iEEG performed comparably to identifying clinician-determined SOZ channels. This implies that SCS could be useful in the SOZ localization even if CS did not occur during iEEG. Further investigation of electrophysical characteristics of SCS compared to CS may help to understand the role of SCS in refractory epilepsy._x000D_ _x000D_ Funding: NIH NINDS F32, NS 114322 (Anderson), NIH NINDS K23, NS 114178 (Rolston), NIH NINDS R21 NS113031 (Rolston)