Abstracts

Rationale: Epileptiform activity (EA) and seizure occurrence are known to fluctuate in a circadian pattern (i.e. 24 hour cycles). Prior studies of circadianicity of EA have been limited to small populations and short recording periods. This is the first study to evaluate long-term circadian patterns of EA in ambulatory patients. Our goals were: (1) to determine if EA exhibits a circadian rhythm in patients implanted with the NeuroPace RNS ^TM System, (2) to determine which times of the day are associated with peaks of EA, and (3) to determine if circadianicity is present for subgroups, segmented by location of seizure focus and by electrocorticographic (ECoG) epileptiform pattern. Methods: We plotted chi-square (X²) periodograms of EA for each of 65 patients implanted with the NeuroPace RNS ^TM System to examine 6 to 42-hour cycles of EA, to measure circadianicity (24-hour periodicity), and to quantify changes in the amount of circadianicity using the methods of Sokolove & Bushnell. Significance cut-offs at alpha < 0.001 and 0.05 were used. To estimate the time of the day with the peak EA rate, we used a cosinor analysis to locate the temporal phase of the circadian peaks. We generated histograms by hour of day to visualize what clock times were associated with the highest rates of EA. We subjected the above data to subgroup analysis to determine if X² was related to: (1) laterality of stimulation, (2) lobe of stimulation, (3) neocortical vs, hippocampal stimulation, and (4) detection type (segmented into the following classes of ECoG detection: beta, alpha, theta, delta, generic/burst, gamma, spike, voltage attenuation). Results: The majority of patients had EA that clustered in a circadian rhythm for most the recording period (months to years), however the amount of circadianicity varied over time, and could have been influenced by seizures or medications. Circadianicity appeared to be a general characteristic of EA irrespective of electrode location or electrographic pattern. A significant circadian pattern was detected in both left and right hemispheres, all lobes (temporal, occipital, parietal, and frontal) and in the hippocampus. Significant circadian patterns were also observed for all types of electrographic activity (theta, delta, beta, bursts, gamma, spiking, and voltage attenuation). Phase analyses revealed peaks of EA across and within all subgroups between approximately 10 PM and 4 AM with another minor peak in the late afternoon. Conclusions: EA as detected by the NeuroPace RNS ^TM System occurs in a circadian rhythm with the most robust peaks of EA between approximately 10 PM and 4 AM. This appears true regardless of site or type of ECoG detection. Circadianicity of EA should be taken into account in the research, design, development, use, and study of intracranial stimulation for epilepsy. To maximize the clinical benefit of any device, device parameters may need to be titrated to match endogenous patterns of EA; this is an area of active and ongoing research.