Abstracts

Rationale: In refractory focal epilepsy, cycles of epileptic brain activity organize seizures over multi-day (multidien) timescales, but the effect of treatment on these cycles and their link to seizure rates is unknown. We hypothesized that cycles of epileptic brain activity may vanish with successful treatments.

Methods: Epileptiform activity (EA) was recorded over years in 163 participants in the RNS System clinical trials, who were implanted with an intracranial brain stimulator for detecting and treating seizures. Participants kept a seizure diary, and changes in medications were logged. Using a wavelet transform, we extracted underlying multidien cycles from recordings of EA. We identified timepoints where a new anti-seizure medication (ASM) was started and compared seizure rates among epochs with present or absent multidien cycles of EA after beginning ASM. This measure was evaluated for predictive power by the area under the curve (AUC) of the receiver operating characteristic.

Results: We identified 334 new ASM trials, of which 91 (27%) led to a >=50% decrease in seizure rate (responders). Relative seizure rate was significantly lower (p< 0.05 Wilcoxon test) when multidien rhythms of EA vanished after introduction of a new ASM. When measuring the sensitivity-specificity trade-off of using decreases in multidien rhythms of EA as predictor for reduction in seizures at a 3-month horizon, we found an AUC of 0.77-0.79 and 0.80 when predicting 50-90% reduction of seizure rate and seizure freedom, respectively. The same method yielded an AUC of 0.82 when predicting seizure freedom at 6 and 12 months.