Abstracts

Rationale: Patient-specific predictors of positive response to vagus nerve stimulation (VNS) remain poorly defined. Given the recent expansion in treatment options for patients with refractory epilepsy, there is increased value in identifying patients who are likely to benefit from VNS therapy. We propose a prediction system for identification of patients who are likely to be VNS responders.

Methods: We present a retrospective analysis of 162 patients with medically refractory epilepsy who underwent VNS at our three institutions. Patient clinical characteristics and seizure frequency before and after VNS implantation were recorded from the electronic record. Patients were categorized as VNS responders (≥ 50% seizure frequency at last follow-up) or non-responders. Time elapsed from VNS implantation to sustained treatment effect was determined for VNS responders. Univariate Cox regression (UVR) was used to identify predictors of VNS response. Variables with p ≤ 0.20 were included in a backward stepwise multivariate Cox regression (MVR). Recursive partitioning analysis (RPA) was used to stratify likely VNS responders compared to non-responders, and the resulting prognostic groups were compared with Kaplan-Meier analysis. Comparisons were considered statistically significant for p ≤ 0.05.

Results: Ninety-one (56.2%) patients achieved ≥ 50% seizure frequency reduction. (Table 1) Left-hand dominance (HR = 1.741, p = 0.028), age at epilepsy onset ≥ 15 years (HR = 2.076, p = 0.003), duration of epilepsy ≥ 8 years (HR = 2.014, p = 0.005) and age at implantation ≥ 35 years (HR = 1.859, p = 0.013) were significant univariate predictors of VNS response, with a trend towards significance for baseline seizure frequency < 5/month (HR = 1.518, p = 0.060). Following MVR, left-hand dominance (HR = 1.920, p = 0.011), age at epilepsy onset ≥ 15 years (HR = 2.514, p < 0.001), and duration of epilepsy ≥ 8 years (HR = 2.348, p = 0.001) remained significant predictors of VNS response.

RPA was performed using variables with p ≤ 0.20 on UVR. Duration and age at implantation were excluded from RPA, because it is unlikely that a patient’s outcome would improve by postponing intervention until the prognostic group nominally changes. (Figure 1A) Patients with either age at epilepsy onset ≥ 15 years, left-hand dominance, or baseline seizure frequency < 5/month were stratified into Group A; the remaining patients were placed in Group B. In Group A, 74.3% of patients were VNS responders at last follow-up with a median Kaplan-Meier time-to-response (TTR) of 24.0 months, compared to a response rate of 42.4% and TTR of 70.5 months in Group B (Figure 1B, p = 0.001).

Conclusions: Patients with age at epilepsy onset ≥ 15 years, left-hand dominance, or baseline seizure frequency < 5/month are ideal candidates for VNS. Genetics and differences in environmental factors in patients with epilepsy onset ≥ 15 years and differences in functional connectivity in left-handed individuals may potentiate the effects of VNS.

Funding: Please list any funding that was received in support of this abstract.: none.