Sleep spindle activity correlates with implicit statistical learning consolidation in untreated obstructive sleep apnea patients

The aim of this study was to examine the relationship between overnight consolidation of implicit statistical learning with spindle frequency EEG activity and slow frequency delta power during non-rapid eye movement (NREM) sleep in obstructive sleep apnea (OSA). Forty-seven OSA participants completed the experiment. Prior to sleep, participants performed a reaction time cover task containing hidden patterns of pictures, about which participants were not informed. After the familiarisation phase, participants underwent overnight polysomnography. 24 h after the familiarisation phase, participants performed a test phase to assess their learning of the hidden patterns, expressed as a percentage of the number of correctly identified patterns. Spindle frequency activity (SFA) and delta power (0.5–4.5 Hz), were quantified from NREM electroencephalography. Associations between statistical learning and sleep EEG, and OSA severity measures were examined. SFA in NREM sleep in frontal and central brain regions was positively correlated with statistical learning scores (r = 0.41 to 0.31, p = 0.006 to 0.044). In multiple regression, greater SFA and longer sleep onset latency were significant predictors of better statistical learning performance. Delta power and OSA severity were not significantly correlated with statistical learning. These findings suggest spindle activity may serve as a marker of statistical learning capability in OSA. This work provides novel insight into how altered sleep physiology relates to consolidation of implicitly learnt information in patients with moderate to severe OSA. This study shows that sleep spindle activity derived from all-night polysomnography was positively correlated to implicit learning performance, whereas SWA and indices of OSA disease severity were not. This work provides novel insight into how altered sleep physiology relates to consolidation of implicitly learnt information in patients with moderate to severe OSA. These findings support the use of quantitative EEG markers as stronger correlates of OSA-related cognitive deficits than traditional measures such as the AHI.•We examined implicit statistical learning and NREM sleep EEG in untreated OSA.•Spindle activity in fronto-central regions were related to statistical learning.•Traditional disease severity metrics were not related to implicit learning.•This highlights the utility of sleep EEG markers of cognitive function in OSA.•It provides novel insights into altered sleep