Structural morphometry distinguishes vagus nerve stimulation outcomes in individuals with epilepsy

Vagus nerve stimulation (VNS) is a commonly used treatment for drug-resistant epilepsy. However, response to VNS is variable, while understanding of this response variability is limited. The vagus afferent network (VagAN), the neural circuits thought to be involved in response to vagal afferents, has been implicated in response prediction in modalities such as diffusion-weighted imaging, functional imaging, and neurophysiology. Structural abnormality in the VagAN from T1w MRI has not yet been studied.

We assessed if VagAN abnormalities, derived from pre-implantation T1w MRI, were associated with VNS response in 92 individuals with epilepsy two years post-implantation (42 responders, 50 non-responders). We used the first principal component (PC1) from a robust principal component analysis for a variety of region sets trained using 100 healthy controls. We investigated two primary hypotheses. (i) If VagAN regions' PC1 differentiates response groups, (ii) if other region sets did not differentiate response groups.

Non-responders had significantly higher VagAN PC1 abnormality scores than both responders ( $r=0.30$ , $p=0.008$ ) and controls ( $r=0.37$ , $p<0.005$ ). By contrast, non-VagAN PC1 showed no discriminatory value ( $r=-0.03$ , $p>0.05$ ). VagAN outperformed all other region sets.

Structural abnormality of VagAN was associated with non-response to VNS. This suggests that volume abnormalities in key VNS regions correlate negatively with reduction of seizures. Abnormalities in other regions, such as the non-VagAN region set, were not associated with VNS response. This metric, when combined with other modalities, may be a useful measure to improve the selection of individuals with drug-resistant epilepsy for VNS therapy.

Vagus nerve stimulation (VNS) is a common treatment for drug-resistant epilepsy; however, it is not always effective. We tested whether volumes of brain regions related to VNS efficacy and found that abnormality within a set of regions known to be affected by VNS related to non-response. No other set of regions showed this effect implying that this set of regions is uniquely placed to determine if someone may respond to VNS. This work supports the idea that VNS efficacy is limited by an individual's brain abnormality and should be considered in further studies.