Abstracts

Rationale: While epileptic spikes are considered interictal markers of epileptogenic cortex, less attention has been paid to the relationship between spectral features of the EEG and epileptogenic areas. We sought to evaluate the relationship between multi-band intracranial EEG spectra and regions of seizure onsets in a group of patients with multifocal epileptic spikes.Methods: We reviewed intracranial EEG from patients who underwent 2-stage epilepsy surgery at our center over the past 2 years and selected the first 10 awake records containing multifocal spikes for our analysis. Four minutes of artifact-free interictal EEG data from each record was subjected to power spectral analysis. EEG spectral power in the delta (1-4Hz), theta (4-8Hz), alpha (8-13Hz), Beta1 (13-21Hz), Beta2 (21-35Hz), low-Gamma (38-44Hz), and high-Gamma (80-110Hz) bands were computed and log-transformed before further analysis. Thirty minutes of EEG recording from each patient was subjected to automated spike detection to quantify spike frequency and amplitudes at each electrode. Maps of spike and spectral variables were normalized and a threshold of 80% of the maxima was applied to all maps. They were then compared to maps of ictal onset created by assigning a binary value of 1 or 0 to electrodes depending on involvement at seizure onsets. Spearman rank correlation was used to compare the ictal and interictal maps.Results: MRI abnormalities lateralized to seizure onset zones was only present in 3 patients (hippocampal sclerosis in two, uncal dysgenesis in one). Seizures onsets were in the anterior mesiobasal temporal region in 4 patients, posterior temporal neocortex in 1, temporo-parietal in 3, and of inferior frontal and parietal onset in 1 patient each. In 8 of these patients, correlation between electrode maps of seizure onset and interictal spectral powers were highest for low-frequency activity in either the 1-4Hz delta band (5 patients) or 4-8Hz theta band (3 patients). The Spearman rank correlations ranged from 0.34 to 0.61 (p<0.001). Only in one instance did interictal spike frequency show a comparable level of significant correlation (0.43, p=0.0011), while spike amplitudes showed significant correlations in 3 instances (r=0.4, p=0.00027; r=0.38, p=0.0006; r=0.44, p=0.00001). In two patients low-frequency spectral powers showed no significant correlation to ictal maps. In one of these patients spike amplitudes, but no other variables,showed a significant correlation of 0.38 (p=0.0006), while in the other Beta2 band (r=0.51, p=0.00009), low-Gamma band (r=0.5, p=0.000014), and spike frequency (r=0.43, p=0.00113) all showed significant correlations. Conclusions: Our results suggest EEG spectral powers in the low-frequency bands are highly correlated to areas of seizure onset. Low-frequency rhythms appear to represent an aspect epileptic dysfunction in the cortex that is distinct from spikes. Their quantification may provide a way to better define epileptogenic zones not only in intracranial studies, but also using non-invasive tools such as magnetic and electrical source imaging.