Regional associations of sleep architecture and Alzheimer's disease pathology

ObjectiveRecent evidence suggests that disturbances of sleep architecture are linked to Alzheimers disease (AD) pathology. Here, we assessed the association between sleep architecture and regional amyloid and tau pathology employing a portable sleep-monitoring device in addition to PET imaging. Methods18 cognitively normal adults (CN; M(Age) = 64.06 (8.63), Sex (M/F) =6/12) and 18 patients with MCI/early AD (M(Age) = 67.33 (8.25), Sex (M/F) =9/9) were included from the "Tau Propagation Over Time" (T-POT) study. All subjects underwent amyloid ([11C]-PiB) and tau ([18F]-AV1451) PET imaging. PET images were normalized to MNI-space and intensity standardized to the whole cerebellum ([11C]-PiB) or the inferior cerebellum ([18F]-AV1451). Sleep monitoring was performed at home using the portable "Dreem" EEG-headband (Beacon Biosignal), which is a reliable and comfortable wireless alternative to the gold-standard polysomnography (PSG). Sleep recordings were performed within six months of the PET acquisitions. At least, one sufficient night had to be acquired, which was used to assess the sleep macrostructure for each individual. Total duration of sleep phases per minutes (i.e. REM, N1, N2, N3) and total sleep time were extracted. In a first step, a linear mixed model (LMM) was used to compare the groups in terms of duration of the different sleep stages across the nightly recordings. Given the results of this comparison, mean N1 and N3 duration were subsequently correlated with regional amyloid and tau pathology SUVRs of 34 cortical regions using Spearmans correlation. The reported results are based on one-tailed tests. All analyses were corrected for age. ResultsPatients with MCI/AD showed reduced N3 duration (p = .007) compared to the CN group. A trend was observed indicating that patients with MCI/AD exhibited longer N1 durations (p = .094); however, this difference did not reach statistical significance. Shorter N3 duration was associated with higher regional amyloid load in the paracentral lobe and the posterior cingulate gyrus, whereas longer N1 duration was linked to higher amyloid pathology in several regions, including the medial temporal lobe, cingulate cortex and the occipital lobe. Moreover, associations were observed between longer N1 duration and greater tau burden in regions comprising the temporal lobe, cingulate cortex, and medial-frontal areas of the brain. ConclusionDifferences in sleep architecture between healthy controls and MCI/AD may arise from regionally-specific accumulation patterns of AD pathologies. Although it remains unknown whether disruptions in sleep architecture are a cause or a consequence, a complex relationship between AD-aggregation pathology in specific brain regions and the different phases of sleep appears to emerge.