Associations between sleep quality, plasma neurofilament light, and cognition in older adults without dementia.

Abstract

The relationship between sleep quality, neurofilament light chain (NFL), and cognitive impairment, including the potential effect of plasma NFL in this association, remains unclear. Using the Alzheimer's Disease Neuroimaging Initiative (ADNI) cohort, we excluded individuals with dementia or a history of sleep-related medication use at baseline, including 640 participants with complete sleep assessments and covariates. Sleep quality was assessed using the Neuropsychiatric Inventory sleep subscale, which includes ratings of frequency, severity, and their product, with higher scores indicating poorer sleep quality. Baseline and follow-up demographics, sleep indices, plasma NFL levels, and cognition scores (including Mini-Mental State Examination [MMSE], Montreal Cognitive Assessment [MoCA], Alzheimer's Disease Assessment Scale-Cognitive Subscale [ADAS13], Clinical Dementia Rating Scale-Sum of Boxes [CDRSB], Executive Function [EF], Language [LAN], and Memory [MEM]) were also collected. Multivariable linear regression examined the associations between baseline sleep quality, plasma NFL, and cognition, as well as the relationship between sleep quality and longitudinal cognitive decline, calculated using linear mixed-effects models. Mediation analysis evaluated the role of plasma NFL in the sleep-cognition association. Multiple testing significance was corrected using false discovery rate, with results presented as Q-values. Poor sleep quality scores were associated with elevated plasma NFL levels (β: 0.055 to 2.645, P < 0.05), poorer cognition (ADAS13, CDRSB, EF, LAN, MEM; β: -0.188 to 1.279, Q < 0.05), and accelerated longitudinal cognitive decline (MoCA; β: -0.005, Q < 0.05) in both models, with sensitivity analyses supporting these findings. Furthermore, plasma NFL levels partially mediated the relationship between sleep quality and both baseline cognition (ADAS13, CDRSB, LAN, MEM; P < 0.05) and longitudinal cognitive decline (MoCA; P < 0.05), with mediation proportions ranging from 9.2% to 26.7%. Poorer sleep quality was associated with cognitive impairment and accelerated cognitive decline, suggesting its potential role in Alzheimer's disease. These associations may be partially mediated by neuroaxonal injury.