von Ellenrieder Nicolás, Dubeau François, Gotman Jean, Frauscher Birgit
Montreal Neurological Institute and Hospital, McGill University, 3801 University Street, Montreal H3A 2B4, Québec, Canada.
Montreal Neurological Institute and Hospital, McGill University, 3801 University Street, Montreal H3A 2B4, Québec, Canada; Department of Medicine and Center for Neuroscience Studies, Queen's University, 18 Stuart Street, Kingston K7L3N6, Ontario, Canada.
Neuroimage Clin. 2017 Feb 24;14:566-573. doi: 10.1016/j.nicl.2017.02.018. eCollection 2017.
The stage of sleep is a known modulator of high-frequency oscillations (HFOs). For instance, high amplitude slow waves during NREM sleep and the subtypes of REM sleep were shown to contribute to a better separation between physiological and pathological HFOs. This study investigated rates and spatial spread of the different HFO types (physiological and pathological ripples in the 80-250 Hz frequency band, and fast ripples above 250 Hz) depending on time spent in sleep across the different sleep cycles.
Fifteen patients with focal pharmaco-resistant epilepsy underwent one night of video-polysomnography during chronic intracranial EEG recording for presurgical epilepsy evaluation. The HFO rate and spread across the different sleep cycles were determined with an automatic HFO detector. We built models to explain the observed rate and spread based on time in sleep and other variables i.e. sleep stage, delta band and sigma band activity, and slow wave amplitude. Statistical significance of the different variables was determined by a model comparison using the Akaike information criterion.
The rate of HFOs depends significantly on the accumulated time of sleep. As the night advanced, the rate of pathological ripples and fast ripples decreased during NREM sleep (up to 15% per hour spent in the respective sleep stages), while the rate of physiological ripples increased during REM sleep (8% per hour spent in REM sleep). Interestingly, the stage of sleep but not the sleep cycle determined the extent of spread of HFOs, showing a larger field during NREM sleep and a more restricted field during REM sleep.
The different dependence with sleep time for physiological and pathological ripples is in keeping with their distinct underlying generating mechanisms. From a practical point of view, the first sleep cycle seems to be best suitable for studying HFOs in epilepsy, given that the contrast between physiological and pathological ripple rates is largest during this time.
睡眠阶段是高频振荡（HFOs）的已知调节因素。例如，非快速眼动睡眠期间的高振幅慢波以及快速眼动睡眠的亚型被证明有助于更好地区分生理性和病理性HFOs。本研究调查了不同HFO类型（80 - 250Hz频段的生理性和病理性涟漪，以及250Hz以上的快速涟漪）的发生率和空间传播情况，具体取决于不同睡眠周期中所花费的睡眠时间。
15例局灶性药物难治性癫痫患者在慢性颅内脑电图记录期间接受了一晚的视频多导睡眠图检查，用于术前癫痫评估。使用自动HFO检测器确定不同睡眠周期中HFO的发生率和传播情况。我们建立模型以根据睡眠时间和其他变量（即睡眠阶段、δ频段和σ频段活动以及慢波振幅）来解释观察到的发生率和传播情况。通过使用赤池信息准则进行模型比较来确定不同变量的统计学意义。
HFO的发生率显著取决于累积睡眠时间。随着夜晚的推进，非快速眼动睡眠期间病理性涟漪和快速涟漪的发生率下降（在各自睡眠阶段每小时下降高达15%），而快速眼动睡眠期间生理性涟漪的发生率增加（在快速眼动睡眠中每小时增加8%）。有趣的是，睡眠阶段而非睡眠周期决定了HFO的传播范围，在非快速眼动睡眠期间显示出更大的范围，而在快速眼动睡眠期间范围更受限。
生理性和病理性涟漪对睡眠时间的不同依赖性与其不同的潜在产生机制相符。从实际角度来看，第一个睡眠周期似乎最适合研究癫痫中的HFOs，因为在此期间生理性和病理性涟漪发生率之间的差异最大。
