Relative entropy as a biomarker for vagus nerve stimulation response in epilepsy: Insights from intracerebral...

This study aims to analyze differences in pre-implantation invasive electroencephalographic (IEEG) data between vagus nerve stimulation (VNS) responders (>50% seizure reduction) and VNS nonresponders (<50% seizure reduction).

We retrospectively identified 19 patients with drug-resistant focal epilepsy who underwent IEEG followed by VNS implantation instead of resective brain surgery. Resting-state IEEG recordings (30 min) were selected for all patients. Three biomarkers were analyzed in the IEEG data: (1) the number of interictal epileptiform discharges (IEDs), (2) the number of high-frequency oscillations (HFOs; ripples: 80-250 Hz, fast ripples: 250-600 Hz), and (3) relative entropy (all frequency bands, 80-250 Hz, and 250-600 Hz). Analyses were performed in three regions of interest: (1) the all contact zones (ACZ), (2) the seizure-onset zones (SOZ), and (3) the non-seizure onset zones (NON SOZ). Differences between VNS responders and nonresponders were then evaluated. Finally, a statistical classifier was constructed to predict VNS efficacy based on all three biomarkers in the ACZ, SOZ, and non-SOZ.

The only biomarker that revealed significant differences between VNS responders and nonresponders was Relative Entropy (SOZ: ripples [U = 1, p = 0.18] and fast ripples [U = 8.0, p = 0.009]), but analysis of effect size using Cliff's delta demonstrated large effects across all three zones in the raw signal, ripple, and fast ripple frequency bands. The best accuracy of 0.84 was obtained for the classifier based on the ACZ; the accuracy of the classifier for the SOZ and NON SOZ was 0.72 and 0.79, respectively.

Invasive electroencephalographic recordings can be used to predict pre-implantation VNS efficacy. In this context, Relative Entropy appears to be a crucial biomarker for distinguishing between VNS responders and nonresponders.

This study shows that brain recordings taken with internal electrodes can help doctors predict whether vagus nerve stimulation will successfully reduce a patient's seizures. We found that a specific measure of brain activity patterns, called Relative Entropy, is a key indicator of whether the treatment will be effective. These findings suggest that analyzing these brain signals can help identify the patients most likely to benefit from vagus nerve stimulation before the device is implanted.