THE STABILITY OF SLOW WAVE SLEEP AND EEG MICROSTRUCTURE MEASURES ACROSS TWO CONSECUTIVE NIGHTS OF LABORATORY POLYSOMNOGRAPHY IN COGNITIVELY NORMAL OLDER ADULTS

Abstract Introduction Healthy and sleep disordered populations show high night-to-night variability of polysomnographic (PSG) macrostructure metrics, however there is evidence of stability in EEG microstructure. In-laboratory PSG is critical to gold standard measures of sleep physiology but multi-night investigations are resource heavy and burdensome to participants. Given the theoretical link between sleep and Alzheimer's disease (AD) pathology (tau and β-amyloid burden), we assessed the night-to-night reliability of sleep macrostructure and EEG microstructure in a group of cognitively normal elderly participating in aging and memory studies. Methods 107 participants (mean = 67±8 yrs., range [54-84 yrs.], 72% female) attended 2 consecutive nights PSG scored according to AASM guidelines for sleep staging, respiratory and leg movement events. Midline EEG (Fz, Cz and Pz referenced to average mastoid signals) were analyzed in 98 participants using an automatic algorithm (DETOKS) for detection of relative slow wave (0.5-4Hz) activity (SWA), NREM2 spindle and K-complexes (KC) densities. Differences between night 1 and 2 for total sleep time (TST), slow wave sleep (SWS), rapid eye movement (REM), stage 2 (%NREM 2), sleep efficiency (SE), apnea hypopnea (AHI) and periodic limb movement (PLM) indices, and EEG microstructure were assessed using t-tests and Wilcoxon rank sum tests where appropriate. Two-way intraclass correlations (ICC) for single unit and absolute agreement were used to determine variability between nights for all measures. Results Night 2 PSGs showed significantly greater TST (6.3 vs 6.8 hours, p<0.001), %REM (17.5 vs 19.7, p<0.001), SE (84.9 vs 87.4%, p<0.02) and SWA (Fz:76.8 vs 78.0%, p<0.01). There were no significant differences between nights for %SWS, %NREM2, AHI, PLMs, spindle and KC densities. ICC and 95% confidence interval estimates were low for TST(0.28), %REM(0.32) and SE(0.32), moderate for %SWS(0.67) and %NREM2(0.59), good for AHI(0.78), SWA(Fz:0.86) and KCs(Fz:0.85), and excellent for PLMs(0.91) and spindles (Pz:0.97). Conclusion SWS, SWA, spindles, KC’s, AHI and PLM indices show good to excellent intra-individual stability across two consecutive nights of PSG. Although there were differences in %REM, SE and SWA, these were numerically small and perhaps functionally or clinically less significant. One night of in-lab PSG is enough to provide reliable estimates of individuals’ SWS, SWA, spindles, KC’s and sleep disorders. Support (If Any) Funding NIH R21AG055002, K24HL109156, R01AG056682, R01AG056531