Abstracts

Rationale:
In patients with drug-resistant epilepsy (DRE), high-frequency oscillations (HFOs), divided into ripples (80-250 Hz) and fast-ripples (250-500 Hz), are promising biomarkers of the epileptogenic zone (EZ) whose removal is needed for seizure-freedom. Fast-ripples are closely linked to the EZ but are seen in only a few patients. Ripples are observed in almost all patients with epilepsy but can be found in physiological areas that should not be resected. Here, we assess the temporal concordance of spikes, ripples, and fast-ripples. We studied whether HFOs, spikes, or their combinations can delineate the EZ in a cohort of children with DRE by correlating their removal with surgical outcome. We hypothesize that spikes overlapping on HFOs are better biomarkers of the EZ compared to spikes that occur in isolation, as well as ripples occurring in isolation or fast ripples.

Methods:
We retrospectively analyzed interictal intracranial EEG (iEEG) data (19 ± 23 minutes) from 34 children (11 ± 6 years, 14 females) with DRE. We automatically detected spikes, ripples, and fast ripples and classified them into categories based on their temporal occurrence (Fig. 1).  The rates of these biomarkers were compared (Wilcoxon rank-sum) between SOZ and non-SOZ as well as resected and non-resected electrodes for both good (Engel I) and poor outcome (Engel ≥ II) patients, and the extent to which these biomarkers were removed was correlated with outcome (Fisher).

Results: Spikes and ripples were found in all patients, while fast ripples were found in only 15 patients. The average rate of spikes, ripples, and fast ripples was 3.4 ± 3.7 spikes/min, 1.4 ± 1.3 ripples/min, and 0.41 ± 0.6 fast ripples/min, respectively, for good outcome patients, and 2.8 ± 2.3 spikes/min, 2.3 ± 3.2 ripples/min, and 0.58 ± 1.2 fast ripples/min, respectively, for poor outcome patients; no differences were observed between good and poor outcome. For good outcome patients, we observed higher rates in resected compared to non-resected electrodes only for all spikes, all ripples, all fast ripples, spikes alone, and spike + ripple (p < 0.05). For poor outcome patients, we did not observe increased rates in resected compared to non-resected electrodes for any of the biomarkers evaluated in this study except for all spikes and spikes alone (p < 0.01). For all fast ripples, spike + ripple, and spike + ripple and fast ripple, good outcome patients had higher resection ratio (p < 0.05; Fig. 2) than poor outcome patients. Complete resection of all fast ripples, spike + ripple, and spike + ripple + fast ripple predicted good outcome (p < 0.05; Fig 2) with an AUC of 0.75, 0.68, and 0.81; a positive predictive value of 86%, 90%, and 80%; and a negative predictive value of 88%, 50%, and 86% respectively, while complete resection of all spikes did not.