Localization of the Epileptogenic Zone Using Neural Spatial Volatility in Epilepsy Surgery

Abstract

Background and objectives: Diseased tissues in the human brain frequently exhibit marked alterations in structural and functional properties compared with adjacent, unaffected regions, whereas healthy tissues display intrinsic, milder topographical variability in these properties. We hypothesized that the epileptogenic zone responsible for drug-resistant focal seizures would be characterized by pronounced alterations in electrographic properties relative to adjacent regions.

Methods: This international, multicenter observational study aimed to test this hypothesis by studying patients with drug-resistant focal epilepsy who underwent curative epilepsy surgery following intracranial EEG (iEEG) recording, with at least 1 year of postoperative follow-up. In this study, "neural spatial volatility" was defined as differences in interictal high-frequency electrographic measures compared with adjacent areas. We tested whether greater neural spatial volatility in resected compared to preserved sites predicts postoperative seizure freedom, independently of clinical, neuroimaging, and ictal iEEG data. In addition, we examined whether incorporating neural spatial volatility into an outcome prediction model, based on the standard-care presurgical evaluation, would improve its performance.

Results: In a derivation cohort of 140 patients with 14,933 samples of iEEG electrodes (mean age 13.1 years; female 68 [48.6%]; postoperative seizure freedom 98 [70.0%]), greater neural spatial volatility in resected compared with preserved sites was an independent predictor of postoperative seizure freedom. The outcome prediction model considering the neural spatial volatility improved the classification performance, showing the area under the curve up to 0.825 (95% CI 0.747-0.902; p < 1.0 × 10^-15). This result was validated in an independent test dataset of 26 patients with 1,626 samples of iEEG electrodes (mean age 25.7 years; female 14 [53.8%]; postoperative seizure freedom 16 [61.5%]) at a different institution, showing the improvement of prediction performance up to 0.819 (95% CI 0.653-0.985; p < 0.0005).

Discussion: Despite the limitations of the retrospective study and the relatively small validation cohort, our findings suggest that increased heterogeneity in the mesoscale distribution of interictal high-frequency activity is a distinguishing feature of the epileptogenic zone in drug-resistant focal epilepsy, potentially offering a novel marker for its localization.