Association of Self-Reported Sleep Characteristics With Neuroimaging Markers of Brain Aging Years Later in Middle-Aged Adults

Abstract

Objectives: To determine the association between early midlife sleep and advanced brain aging patterns in late midlife.

Methods: Using the CARDIA study, we analyzed sleep data at baseline and 5 years later, focusing on short sleep duration, bad sleep quality (SQ), difficulty initiating and maintaining sleep (DIS and DMS), early morning awakening (EMA), and daytime sleepiness. These were categorized into 0-1, 2-3, and >3 poor sleep characteristics (PSC). Brain MRIs obtained 15 years later were used to determine brain age through a machine learning approach based on age-related atrophy.

Results: This cohort study included 589 participants (mean age 40.4 ± 3.4 years, 53% women). At baseline, around 70% reported 0-1 PSC, 22% reported 2%-3%, and 8% reported >3 PSC. In multivariable linear regression analyses, participants with 2-3 or >3 PSC had 1.6-year (β = 1.61, 95% CI 0.28-2.93) and 2.6-year (β = 2.64, 95% CI 0.59-4.69) older brain age, respectively, compared with those with 0-1 PSC. Of the individual characteristics, bad SQ, DIS, DMS, and EMA were associated with greater brain age, especially when persistent over the 5-year follow-up.

Discussion: Poor sleep was associated with advanced brain age in midlife, highlighting the importance of investigating early sleep interventions for preserving brain health.