Abstracts

Rationale: Recent studies using chronic intracranial electroencephalography (cEEG) have yielded convergent evidence that epilepsy is a cyclical disorder. Interictal Epileptiform Activity (IEA) fluctuates in cycles at circadian and multidien (multi-day) time-scales, and electrographic seizures cluster preferentially during certain phases of these cycles. Here, using cEEG from the NeuroPace^® RNS^® System and parallel clinical data from seizure diaries, we show that clinical seizures also tend to occur at preferred phases of IEA cycles. Methods: Seizure and IEA data from 16 patients involved in the RNS System pivotal trial were evaluated. Diaries of patient-reported seizures were available for up to seven years. Wavelet decomposition was applied to IEA time-series data (i.e. the rate of detected epileptiform discharges per hour by the neurostimulator) to resolve component circadian and multidien rhythms. The instantaneous phase of multidien rhythms was calculated and, using circular statistics, the average phase at which clinical seizures occurred was determined. Results are reported as phase-locking values, an index of phase clustering, and Rayleigh test statistics. Results: All patients were found to have circadian and multidien IEA cycles, the latter with patient-specific periodicities ranging from 7 to 33 days. For 10 out of 16 patients, we found that seizures clustered at preferred phases of the IEA rhythms with phase-locking values ranging from 0.2 to 0.8 (mean=0.32, p<0.05 to <0.001, Rayleigh test).The Figure shows 20 months of IEA data from one representative patient. Cyclical variation of IEA (a, 25-day periodicity) is shown alongside weekly clinical seizure frequency (b).  A polar plot with phase histogram of seizure occurrence (c, red triangles) reveals phase preference of seizures. Vector (red arrow) direction indicates average phase of seizure occurrence and vector length corresponds to the phase-locking value (c). Conclusions: We extend previous findings on the phase-relationship between electrographic seizures and IEA cycles recorded with cEEG by showing that a similar relationship exists for clinical seizures. This small series corroborates an emerging view that seizures are not random events, and suggest that it may be possible to forecast clinical seizure risk on the scale of days. Funding: None