1031 Association Between Objective Sleep Architecture and Biomarkers of Alzheimer's Disease

Abstract Introduction Growing evidence suggests a bi-directional association between sleep and Alzheimer’s disease (AD), with several studies showing an association between self-reported sleep disturbances and AD biomarkers. However, the association between polysomnography (PSG)-assessed sleep architecture and AD biomarkers remains unknown. We aim to investigate the associations between sleep architecture features and AD biomarkers in a clinical population with and without cognitive impairment. Methods We examined 128 participants mean age of 70.9 years (±9.7), 43.8% of men from an outpatient memory clinic, of whom 23 were healthy controls, 41 patients with mild cognitive impairment, and 64 patients with AD. Sleep architecture features were derived from PSG data and categorized by tertiles. Four AD biomarkers were measured, including tau-181, Brain-Derived Neurotrophic Factor (BDNF), and neurofilament light (NFL) obtained from blood samples, along with Aβ levels measured from amyloid positron emission tomography scans. Multivariable linear regressions models were performed to assess the associations between sleep architecture features and AD biomarkers. Results After adjustment for age, gender, APOE4 status, diabetes mellitus, smoking habits, and body mass index, participants in the highest REM latency tertile (i.e., with longer REM latency; 192.7 minutes) were associated with increased levels of tau-181 (β= 0.25, 95% confidence interval (CI)= 0.01;0.55) and Aβ (β= 0.15, 95% CI= 0.06;0.26), and with decreased level of BDNF (β= -0.51, 95% CI= -0.70;-0.20) compared to participants in the lowest tertile (i.e., with shorter REM latency; 98.2 minutes). Moreover, being in the middle SWS percentage tertile (i.e., SWS percentage between 2.5 and 10.2) was associated with a lower level of tau-181 (β= -0.18, 95% CI= -0.34;-0.02) compared to being in the lowest tertile (i.e., shorter SWS percentage; 2.5)). We did not find any association between AD biomarkers and sleep duration, efficiency, latency, or other sleep architecture features. Conclusion In this cross-sectional study of a clinical population, we highlight the importance of REM latency, and to a lesser extent, SWS percentage, in relation to AD biomarkers. Future research is needed to examine the longitudinal association between sleep architecture and AD biomarkers and clarify underlying mechanisms. Support (if any)