Abstracts

Rationale: The timing of seizures has a circadian pattern in many patients, but it is unknown whether other epileptic biomarkers also show circadian patterns. Our objective was to determine whether the rate high-frequency oscillations (HFOs) varies with a 24 hour cycle, either overall or per each sleep stage. Methods: HFOs were automatically detected in 18 subjects undergoing invasive EEG monitoring, and sleep scoring was manually performed using surface EEG electrodes. HFO rates were modeled as an both a homogeneous and inhomogeneous Poisson process, in the latter case using least squares for fitting the third order Fourier series. All p-values were computed using chi squared tests.  Results: HFO rates were not found to have statistically significant circadian patterns in any of the 18 patients at the 0.05 confidence level. However, HFO rates were found to differ between awake and REM (p < 0.002), awake and NREM (p < 10^-12), and between stage 2 and REM (p<10^-5), with the median rate of 1.2 HFOs/min (stage 2), 1.1 HFOs/min (NREM), 1.1 HFOs/min (REM), and 0.8 HFOs/min (awake). Conclusions: HFO rates do not appear to have circadian variations either overall or within specific sleep stages. However, HFO rates do vary between states of vigilance, though the magnitude of the change is small. Funding: NIH BD2K K01-ES026839 (SG), NIH NINDS K08-NS069783 (WS) and R01-NS094399 (WS), and the Doris Duke Foundation Clinical Scientist Development Award #2015096 (WS)