Abstracts

Rationale: Previous studies have suggested that epileptiform activity and seizures have a circadian distribution, but were limited by a) short-term recording in an artificial EMU environment with altered AEDs and sleep, or b) subjective reports of seizures by diary. We studied circadian patterns using long-term ambulatory intracranial recordings and epileptiform detection count data recorded by the NeuroPace® RNS® System.Methods: ECoG recordings and device detection counters were retrospectively reviewed for all subjects participating in the RNS® System pivotal trial who met the inclusion criterion for this study of stable detection parameters over a continuous 84-day period. Data analyzed included “detections” (D = hourly counts of device-detected epileptiform events based on investigator-specified patterns) and “long episodes” (LE = hourly counts of device-detected events that exceed an investigator-specified duration). In a subset of subjects, LE represented electrographic seizures. To identify these subjects, the most recent 100 ECoG records with long episodes were reviewed by 2 board-certified EEG readers. Spectrum resampling was used to determine the dominant frequency periodicity of the dataset. Cosinor analysis was used to identify subjects with significant circadian peaks in their detected activity, and the peak time of day. Subjects were grouped by region of electrographic seizure onset; chi-squared analysis was used to assess for non-uniform distribution of peak activity.Results: Across all 134 subjects , the dominant periodicity for D and LE was 24.0 hours (95% CI: 23.7 – 24.3 for both metrics), and in 40% (53 /134) of subjects, LEs indicated electrographic seizures. Within the 24-hour period, a significant circadian peak in D and LE was detected for the majority of subjects (94% for D, 66% for LE). In mesial temporal lobe (MTL) subjects with significant circadian patterns (p < 0.05), epileptiform activity (as measured by D) peaked consistently at night (0000-0400 Hrs), but the peak times of sustained patterns (as measured by LE) were variable across subjects (fig 1). Those with extratemporal (ETL) onsets showed the reverse pattern with peak LE occurring between 0200 – 0300 Hrs, but no consistent peak time for D (fig 2). The subset of ETL subjects with frontal lobe onset (FLE) had consistent circadian distributions of D and LE, including LE that represented electrographic seizures.Conclusions: We conclude that: 1) long-term intracranial recording of epileptiform activity in ambulatory subjects is feasible; 2) epileptiform activity has a strong 24 hour periodicity with peak nocturnal occurrence; and 3) there is a dissociation between peaks of brief and sustained epileptiform activity in MTL and ETL epilepsy. These findings support the assertion that circadian patterns of epileptiform activity vary by site of seizure onset zone, which may have implications for treatment and safety, including injury and SUDEP.