Peripheral sympathetic activations underlying PWA drops induce significant changes in EEG activity in a sleep-stage-specific manner

Objectives/Introduction: EEG‐microarousals during sleep are associated with surges in sympathetic activity and represent the hallmark of pathological sleep fragmentation common to many sleep‐related disorders. However, little is known about their time‐frequency composition and its modulation by the magnitude of concomitant sympathetic activity. Here we used drops in pulse wave amplitude (PWA) measured by photoplethysmography (PPG) as a marker of peripheral sympathetic activation [1][4] to investigate associated EEG changes. Methods: PWA‐drop detection [5] was performed on the overnight PPG of 16 subjects of the HypnoLaus Cohort [6]. EEG‐activity was analyzed in six channels: F3/4‐C3/4‐O1/2. For each sleep stage, a regression analysis was performed on the time‐varying root‐meansquare of band‐pass‐filtered EEG‐signals (delta: 0.54 Hz, theta: 48 Hz, alpha: 812 Hz, sigma: 1216 Hz, beta: 1625 Hz, gamma: 2545 Hz), using as regressor the PWA‐drop timing (40% amplitudethreshold). A permutation test was performed by randomly shuffling events within the regressor. Then, we explored drop‐induced EEG changes and their correlation with the PWA area under the curve (AUC). For each PWA‐drop (10% amplitude‐threshold), a time‐frequency representation of EEG changes (1s time‐windows; 75% overlap; 1 Hz frequency‐bin; 035 Hz) was computed in a 14s‐window centered on the PWA‐drop beginning. Power values were z‐scored using EEG‐activity ‘background’ values estimated from a 60 s timewindow centered on each drop (drop‐associated periods were discarded). Results: EEG‐activity was significantly modulated by PWA‐drops in all sleep stages, with the most consistent effect within delta and beta/gamma bands (p < 0.05 in >80% of subjects). In all sleep stages, PWA‐drops were preceded by sharp significant increases in delta power and followed by significant high‐frequency increases (alpha/beta/gamma). In light NREM sleep, relative high‐frequency increases were evident even for small PWA‐drops, while this was not the case for slow‐wave sleep. Conclusions: Our results demonstrate that sympathetic activations are associated with quantitative and qualitative EEG changes that depend both on the magnitude of the autonomic activation and the sleep stage. This time‐frequency characterization of EEG changes may be employed to define new automatic algorithms to detect sleep‐stage‐specific micro‐arousals. This will be helpful in order to understand arousal‐related pathological mechanisms involved in many sleep disorders.