Prior research shows a consistent link between Alzheimer’s disease (AD) and poor sleep quality; however, most studies have been limited by subjective self-report sleep measures and a lack of diagnostic biomarker support. A better understanding of the intersection between objective and subjective sleep, blood-based AD biomarkers, and cognitive aging may help clarify the role of sleep as a non-pharmacological modifiable risk factor for cognitive decline. We examined the independent and interactive effects of sleep and plasma pTau181 on cognitive functioning across two independent cohorts.

This study included two non-overlapping cohorts of community-dwelling older adults enrolled in research studies at the UC San Francisco Memory and Aging Center. The first cohort (Fitbit) included 39 adults (age = 70.9 ± 9.5yrs; 54% female; 83% cognitively unimpaired, 14% mild cognitive impairment, 3% dementia; 77% White; mean education = 17.8yrs) who completed 30 days of Fitbit^TM Inspire 2 monitoring, which tracked total time asleep each night (averaged within-persons across nights). The second cohort (PSQI) included 157 adults (age = 74.6 ± 7.1yrs; 51% male; 87% cognitively unimpaired, 89% white, mean education= 17.4yrs) who completed the Pittsburgh Sleep Quality Index (PSQI), a self-report measure of sleep quality (higher score = worse sleep quality). Both cohorts completed in-person neuropsychological assessments and blood draws with plasma analyzed for pTau181 concentration. Composite cognitive z-scores were calculated for memory and executive functioning. In each cohort, multiple linear regression models examined the interaction between sleep and plasma pTau181 on cognitive domain z-scores, controlling for age, sex, and education.

In the Fitbit cohort, wearable-based total sleep time did not have a significant bivariate association with plasma pTau181 concentration (r= -0.233, p=0.142). Plasma pTau181 was not significantly correlated with executive functioning (r= -0.140, p=0.396); however, this relationship was moderated by sleep (b=0.003, p=0.001) such that more total time asleep attenuated the negative relationship between pTau181 and executive functioning. The interaction between pTau181 and sleep on memory was not significant (b= -0.001, p=0.485). In the PSQI cohort, sleep quality also did not have a significant bivariate association with plasma pTau181 concentration (r= -0.058, p=0.468). However, self-reported sleep quality moderated the relationship between pTau181 and memory (b= -0.058, p=0.045) such that better sleep quality attenuated the negative relationship between pTau181 and memory z-scores. PSQI scores did not moderate the relationship between pTau181 and executive z-scores (b= -0.003, p=0.907).

Using both objective Fitbit^TM total sleep time and PSQI self-reported sleep quality, the relationship between pTau181 and cognitive functioning was moderated by sleep such that better sleep characteristics associated with better cognitive performance despite the adverse effects of elevated plasma pTau181 on cognition. Results highlight the clinical relevance of supporting better sleep both in terms of total hours and quality, monitoring participants’ sleep characteristics in patients at elevated risk for AD, and support additional research on sleep interventions in AD. Longitudinal studies of sleep as a potential modifiable risk factor for incident cognitive decline in neurodegenerative populations are warranted.