Spike height improves prediction of future seizure risk

Abstract

ObjectiveWe evaluated whether interictal epileptiform discharge (IED) rate and morphological characteristics predict seizure risk. MethodsWe evaluated 10 features from automatically detectable IEDs in a stereotyped population with self-limited epilepsy with centrotemporal spikes (SeLECTS). We tested whether the average value or the most extreme values from each feature predicted future seizure risk in cross-sectional and longitudinal models. Results10,748 individual centrotemporal IEDs were analyzed from 59 subjects at 81 timepoints. In cross-sectional models, increases in average spike height, spike duration, slow wave rising slope, slow wave falling slope, and the most extreme values of slow wave rising slope each improved prediction of an increased risk of a future seizure compared to a model with age alone (p < 0.05, each). In longitudinal model, spike rising height improved prediction of future seizure risk compared to a model with age alone (p = 0.04) ConclusionsSpike height improves prediction of future seizure risk in SeLECTS. Several other morphological features may also improve prediction and should be explored in larger studies. SignificanceDiscovery of a relationship between novel IED features and seizure risk may improve clinical prognostication, visual and automated IED detection strategies, and provide insights into the underlying neuronal mechanisms that contribute to IED pathology.