Vagus nerve stimulation-induced laryngeal motor evoked potentials for response prediction and intensity titration in drug-resistant epilepsy. (March 2023)

• Record Type: Journal Article
• Authors: Berger, Alexandre; Carapancea, Evelina; Vespa, Simone; Danthine, Venethia; Doguet, Pascal; Delbeke, Jean; Nonclercq, Antoine; El Tahry, Riëm
• Is Part Of: Clinical neurophysiology. Volume 147(2023)
• Page Start: 99
• Page End: 107
• Publication Date: 2023-03
• Subjects: Vagus nerve stimulation -- Epilepsy -- Larynx -- Biomarker -- Motor-evoked potentials -- SVM; Neurophysiology -- Periodicals; Electroencephalography -- Periodicals; Electromyography -- Periodicals; Neurology -- Periodicals; 612.8
• DOI: 10.1016/j.clinph.2023.01.009
• Languages: English
• ISSNs: 1388-2457

Abstract

Highlights: A support vector machine model based on laryngeal motor evoked potential features discriminated responders and non-responders with an accuracy of 80%. Responders received vagus nerve stimulation at an intensity between 1-fold and 2-fold the recruitment saturation of laryngeal motor evoked potentials, which may correspond to an optimal target. Laryngeal motor evoked potentials could be proposed as a guidance to scale vagus nerve stimulation intensity based on a physiological indicator of fiber engagement. Abstract: Objective: The objective of the study was to record Laryngeal Motor Evoked Potentials (LMEPs) in Vagus Nerve Stimulation (VNS)-implanted patients suffering from Drug-Resistant Epilepsy (DRE). Based on these recordings, LMEPs characteristics were evaluated and compared between responders (R) and non-responders (NR). Finally, possible under- or over-stimulation was assessed based on a physiological indicator of fiber engagement. Methods: Mean dose–response curves were compared between R and NR. A Support Vector Machine (SVM) model was built based on both LMEP and dose–response curves features, to discriminate R from NR. For the exploration of possible under- or over-stimulation, a ratio between the clinically applied stimulation intensity and the intensity yielding to LMEP saturation was computed for each patient. Results: A trend towards a greater excitability of the nerve was observed in R compared to NR. The SVM classifier discriminated R and NR with an accuracy of 80%. An ineffective attempt to overstimulate at current levels above what is usually necessary to obtain clinical benefits was suggested in NR. Conclusions: The SVM model built emphasizes a possible link between vagus nerve recruitment characteristics and treatment effectiveness. Most of the clinically responding patients receive VNS at a stimulation intensity 1-fold and 2-fold the intensity inducing LMEP saturation. Significance: LMEP saturation could be a practical help in guiding the titration of the stimulation parameters using a physiological indicator of fiber engagement.