Abstracts

Rationale: Vagus nerve stimulation (VNS) is an established therapy for drug resistant epilepsy with around half of those receiving VNS showing meaningful seizure improvement. For improved patient care it is vital that response to VNS can be ascertained prior to insertion for improved outcomes and better individual treatment. A key feature often present in epilepsy is the existence of pathology visible on MRI. However, it is unclear if and how such pathology relates to VNS response. Here, we hypothesised greater structural abnormality in non-responders to VNS, compared to responders and healthy controls.

Methods: We assessed T1w MRI prior to implantation for 43 VNS responders (age 36.4±19.9, 21 Females(F)) and 54 non-responders (age 40.5±17.7, 35 F). This was done by computing grey matter volumes from 6 regions of interest in each hemisphere and the brainstem, split into 107 total sub-regions from areas thought to be most affected by VNS (namely the Vagus Afferent Network (VagAN) (Hachem et al., 2018), Figure 1A). We used 100 healthy controls (age 39±21, 62 F) to correct for age, sex, and scan type using ComBat. These controls were then used to z-score individuals for each region. Group differences were then analysed using a one-tailed Mann-Whitney U test across group pairs of non-responders, responders, and controls. We additionally calculated Cohen’s d for effect size between these groups.

Results: We found less structural abnormality in responders than non-responders in T1w MRI across the VagAN. Within the VagAN (Figure 1B) non-responders showed significantly greater structural abnormality than responders (p=0.02, d=-0.40) and the control group (p<0.01, d=-0.66). Comparatively, no significant difference was found between responders and controls (p=0.11, d=-0.27).

Conclusions: Overall, we showed significant structural differences between VNS responders and non-responders pre-implant. These structural differences may allow for delineation of responders and non-responders prior to VNS implantation and in the future could be used in predictive models.

References: Laureen D Hachem, Simeon M Wong, and George M Ibrahim. The vagus afferent network: emerging role in translational connectomics. Neurosurgical focus, 45(3):E2, 2018.


Funding: We thank members of the Computational Neurology, Neuroscience & Psychiatry Lab (www.cnnplab.com) for discussions on the analysis and manuscript. YW and PT were supported by UKRI Future Leaders Fellowships (MR/V026569/1, MR/T04294X/1).