Authors :
Presenting Author: Jeston Chin, BS – UTHealth Houston
Chaitanya Ganne, MD, PhD – UTHealth Houston
Vladimir Vashin, BS – UTHealth Houston
Jay Gavvala, MD – UTHealth Houston
Sandipan Pati, MD – University of Minnesota
Rationale:
Seizure timing in epilepsy is not random but governed by circadian and multidien cycles. While prior studies in focal epilepsy have described seizure periodicity, the dynamics of seizure cycles in thalamocortical network epilepsies, such as idiopathic generalized epilepsy (IGE), multifocal epilepsy (MF), and epileptic encephalopathy (EE), remain poorly understood. Here, we present the first comparative analysis of seizure and interictal epileptiform activity (IEA) cycles recorded from the centromedian (CM) thalamus using long-term ambulatory data from responsive neurostimulation (RNS).
Methods:
In this retrospective cross-sectional study we included 9 patients (IGE: MF: EE = 5:2:2) with drug-resistant epilepsy treated with centromedian (CM) Responsive Neurostimulation (RNS) over a minimum 1-year post-implantation period. We analyzed (1) self-reported seizures from the patient dairy, (2) timestamps of eletrographic-seizures/long episodes (ES - defined as >30 to >50 seconds) captured by the RNS (3) Interictal epileptiform activity (IEA), defined as hourly counts of detection of short-lived epileptiform discharges (< 30seconds), (4) periodic changes in stimulation/detection parameters as predictors of change in the cyclicity of IEA. Continuous Morlet-wavelet transform was used to estimate epileptiform activity cycles. The statistical significance of these cycles was tested against ten thousand surrogate distributions of IEA. Event phases were estimated using a Hilbert transform. Circular statistics of event phases were estimated and the resultant vector of circular distribution was evaluated using inter-trial phase clustering (ITPC) with statistical significance evaluated using Rayleigh’s test with false discovery rate correction. Linear mixed effects (LME) models were used to determine if frequency, burst duration, detector class, change in detection parameters, charge density and epilepsy type affected the IEA cyclicity.