Network Changes in Eating Epilepsy: An MEG Study

Objective: To examine the network differences in eating epilepsy at resting state and during eating period using MEG recording.

Methodology Resting state MEG data were analyzed in 17 patients with eating epilepsy (M:F = 12:5, age at onset = 9.66 ± 8.63 years, age at evaluation = 20.6 ± 8.7 years). All underwent presurgical workup including MRI, EEG, video-EEG, MEG, and neuropsychological assessment. MEG recording while eating (rice) was obtained in five patients (M:F = 2:3). MRI (brain) showed normal-9; left perisylvian gliosis-4, bilateral peri-sylvian gliosis-2, left MTS-1, and left temporal astrocytoma-1. Data were down sampled to 500 Hz, artifacts removed and bandpass filtered between 0.5 and 150 Hz. Data were transformed to source space by beam forming and parcellated. Amplitude envelope correlation was applied to six frequency bands (delta: 1–4 Hz, theta: 5–7 Hz, σ: 8–12 Hz, β: 13–30 Hz, gamma1: 30–60 Hz, gamma 2: 60–90 Hz) to derive adjacency matrices for both conditions. Network measures (clustering coefficient, modularity, local efficiency, betweenness centrality, degree, and path length) were derived using graph theory. Group comparisons were FDR corrected.

Results: Resting state networks showed increased clustering and pathological local information processing in the superior, inferior, and medial parietal regions and prefrontal regions predominantly in the β and gamma frequency bands (p < 0.036). While eating, network changes involved the precuneus, inferior parietal, right prefrontal, bilateral calcarine, and entorhinal regions in the β and gamma bands (p < 0.027).

Conclusion: Previous electrophysiological and functional imaging studies have implicated involvement of perisylvian region. Pathological information processing was noted across the perisylvian regions dominantly on the right hemisphere in this study. Eating resulted in recruitment of bilateral hubs between the parietal, temporal, and occipital structures.