Purpose: This study investigates the potential of preoperative MEG functional connectivity networks to predict the efficacy of vagus nerve stimulation (VNS) in patients with drug-resistant epilepsy (DRE).
Methods: A total of 18 DRE patients and 18 healthy controls were enrolled. Resting-state MEG data were collected preoperatively, and brain network connectivity was assessed across seven frequency bands (δ, θ, α, β, γ, ripple, and fast ripple) using corrected amplitude envelope correlation (AEC-c). Network-based statistics (NBS) were employed to identify differences in connectivity patterns.
Results: Compared to healthy controls, DRE patients, particularly non-responders (NR-VNS), exhibited widespread abnormal functional connectivity, including significant increases in low-frequency bands and mixed alterations in mid-to-high frequency bands. Responders (R-VNS) showed marked normalization of brain connectivity, with reductions in differences from controls, especially within alpha and beta bands. These connectivity patterns were significantly associated with treatment outcomes, indicating their potential as predictive biomarkers.
Conclusions: Preoperative brain network patterns derived from multi-frequency MEG, particularly in alpha and beta bands, hold promise for predicting VNS treatment response in DRE patients. The "health status" of the brain's network prior to implantation appears to be a crucial factor influencing therapeutic efficacy.
Keywords: Drug-resistant epilepsy; Functional connectivity; Magnetoencephalography; Network-based statistical; Vagus nerve stimulation.
© 2026. The Author(s).