Neurology. 2026 Apr 14;106(7):e214710. doi: 10.1212/WNL.0000000000214710. Epub 2026 Mar 2.
ABSTRACT
BACKGROUND AND OBJECTIVES: Vagus nerve stimulation (VNS) is the most common neuromodulation technique used to treat drug-resistant epilepsy (DRE) in children. Despite this, approximately half of those implanted do not realize a benefit and there are currently no means to preoperatively identify responders. Recent neuroimaging work has suggested that intrinsic differences in brain connectivity may explain some heterogeneity in VNS responsiveness. In the current work, we sought to study whether preimplantation functional network perturbations in relation to interictal epileptiform discharges (IEDs) are associated with VNS response in children with focal DRE.
METHODS: We retrospectively studied resting-state magnetoencephalography in children with focal DRE (n = 65), recorded before VNS implantation. Beamforming was used to reconstruct source-level estimations of neural activity within a parcellation of 52 cortical brain regions. Static functional connectivity was estimated using amplitude envelope correlation, followed by general linear modeling and network-based statistics to identify spatial networks associated with VNS response (>50% seizure reduction at 6 months). Perturbations in brain connectivity were estimated by inferring dynamic cortical microstates from amplitude envelopes and extracting event-related microstate probability time courses surrounding IEDs. Differences in microstate dynamics after IEDs were assessed using t tests at each time point, comparing responders and nonresponders, followed by temporal cluster-based correction for multiple comparisons.
RESULTS: A total of 44 children were included in the final analysis (mean age 15 years, 57% male, 52% responders). No clinical variables, including IED topographies, were associated with VNS response. Significant static networks were identified in alpha-band connectivity relating to both VNS response (anterior-dominant, t = 4.52) and nonresponse (posterior-dominant, t = -4.98). From the dynamic microstate analysis, one microstate related to a frontotemporal network showed significantly greater perturbation in nonresponders compared with responders (temporal cluster p < 0.05), in the 500 milliseconds after IEDs.
DISCUSSION: Our results provide evidence that connectivity of an intrinsic, anterior-dominant network is associated with response to VNS. Responders to VNS are characterized by stronger baseline connectivity of this network and greater resilience of this network to IED-related disruption.
PMID:41771009 | DOI:10.1212/WNL.0000000000214710