Rationale:
According to the former meta-analysis (Englot et al., 2016), responder rates tend to increase between the second year and the fifth year after vagus nerve stimulation (VNS). In this study we tried to demonstrate long term effect of invasive VNS to drug-resistant epilepsy with its change over time and reveal predictor of good outcome.
Methods:
Patients who had undergone VNS surgery from July 1997 to May 31, 2022 were enrolled. Gender, age, time of epilepsy onset, time of VNS surgery, baseline seizure frequency and frequency per month at three months, six months, and annually after VNS surgery were investigated. Seizure reduction rate at each follow-up was calculated. For patients who were judged to be ineffective and discontinued VNS by the doctor, the last observation was carried forward even after the point of discontinuation. Paired comparison of the seizure frequency before VNS and after VNS was performed. After that, logistic regression was performed to see if there were factors affecting responder (reduction rate >= 50%) status at the longest time point that showed a significant difference compared with baseline. Independent variables included seizure frequency before VNS, age at VNS, duration of epilepsy before VNS, lateralization (whether seizure onset is lateralized or not), the number of antiseizure medication (ASM) and VNS intensity. Repeated measures ANOVA was conducted based on the variables identified as significant in logistic regression to see whether the trend of the VNS effect differed depending on the variables.
Results:
Retrospectively, 208 patients were enrolled. Up to four years after VNS surgery, there was a significant decrease in seizure frequency compared to preoperative seizure frequency, no significant change in seizure frequency were shown thereafter (Table 1). When logistic regression was performed at four years after surgery, the higher the preoperative seizure frequency, the higher the likelihood of becoming a responder (Odds ratio[OR]=1.024,
p=0.002) and the higher the likelihood of becoming a responder (OR=0.217,
p=0.008) in the absence of mental retardation. There was a significant difference in the change trend according to the difference in preoperative seizure frequency (< 20 or >=20) (F=12.020,
p=0.000, ηp2=0.092). Although the frequency showed a decreasing trend, patients whose frequency was less than 20 before surgery did not show a clear decrease. Even in patients with a baseline frequency of 20 or more, after a significant decrease in the first three months, the effect was maintained at a similar level without additional reduction (Figure 1). On the other hand, the mental retardation showed no interaction over time (F=0.371,
p=0.898, ηp2=0.003).
Conclusions: Up to four years after VNS surgery, there was a significant decrease in seizure frequency. The higher the preoperative seizure frequency, the greater the VNS effect, and the less effective in the presence of mental retardation. The effect was achieved mostly at the first three months and there seemed to be no additional effect thereafter.
Funding: This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.