Abstracts

Rationale: Clinical trials of devices for epilepsy present unique challenges. These include increases in efficacy over time, a need for longer treatment periods, and a decision between active and sham controls. In July 2013, FDA granted a conditional IDE to NeuroSigma to commence a pivotal trial of eTNS for DRE. In planning for this trial, methods from the phase II trial of eTNS and prior device trials were reviewed to help optimize study design. Methods: The methods and results of major neuromodulation trials (VNS, DBS, RNS, eTNS) were reviewed for inclusion criteria, duration of the treatment periods, response in the first 4-6 weeks, and use of active control settings. Based on this review, a re-analysis of the eTNS phase II trial was performed using modified inclusion criteria and incorporation of an induction phase (exclusion of the first treatment visit from the efficacy analysis). Statistical methods include Analysis of Variance, p < 0.05.Results: The DBS, RNS, and eTNS device trials demonstrated progressive increases in efficacy over the acute treatment period. Data for VNS is only available for the entire treatment period. In the DBS, RNS, and eTNS trials, efficacy for the treatment groups was not significantly different from controls at the first treatment visit at 4-6 weeks (p=0.5 DBS, p = 0.27 RNS, p = 0.89 eTNS). Rather, peak efficacy occurred at the final treatment visit. This supports the concept that an induction period is common in device studies. As for variability in seizure frequency at baseline, between-group baseline seizure frequency was best matched for DBS and RNS studies, but more variable for the VNS and eTNS studies. Taking these facts into consideration, we performed a re-analysis of our phase II eTNS trial to model how the pivotal trial might perform using revised inclusion criteria, and incorporating an induction period. First, we excluded subjects with < 4 and > 60 seizures/month. Second, we excluded the first treatment visit from the primary efficacy assessment. Using the revised criteria, the re-analysis indicates that the mean reduction in seizures for the treatment group is -34% versus +6.6% for controls (n = 26, p = 0.03, ANOVA; median reduction in seizures = -30.3% versus -1.1%, p = 0.03, ANOVA). Increasing the minimum number of seizures required for enrollment, capping the maximum allowable seizures, and eliminating the first treatment visit helps reduce baseline variation and increases the likelihood of statistical separation during the treatment period. Conclusions: We expect tightening the inclusion criteria, and introducing an induction phase early in the treatment period may increase the likelihood of executing a successful phase III trial. Decisions regarding control parameters and the advantages of treatment periods longer than 12-weeks will be discussed. We hope to incorporate these innovations into our pivotal trial of eTNS for drug resistant partial seizures, and believe they may help future device studies for epilepsy.