Predictors of Cognitive Resilience in the Multi-Ethnic Study of Atherosclerosis

TL;DR

This study identifies education and income as key factors that help people maintain better cognitive function despite brain changes linked to aging.

Contribution

The study reveals that education and income are major modifiable predictors of cognitive resilience in a diverse U.S. population.

Findings

Higher educational attainment and family income were independently associated with greater cognitive resilience.

More depressive symptoms and higher waist girth were linked to lower cognitive resilience.

Education and income contributed significantly to explaining cognitive resilience, accounting for 39% and 22% of variance, respectively.

Abstract

Modifiable predictors of cognitive resilience (CR)—maintaining better-than-expected cognition despite neuropathology—may inform dementia prevention strategies. We quantified CR and identified main predictors in a U.S.-based multi-ethnic cohort. We used data from 1,335 individuals enrolled in 2000-02 (Exam-1; age=45-84) in the Multi-Ethnic Study of Atherosclerosis (MESA) with repeated global cognition scores (z-scores of Cognitive Abilities Screening Instrument, Digit Symbol Coding, Digit Span) administered at Exams 5 (2010-12), 6 (2016-18), and 7 (2022-24). Neuroimaging biomarkers assessed at Exam 6, included white matter fractional anisotropy, white matter hyperintensities, and total brain volume, adjusted for intracranial volume. First, using linear mixed-effects models adjusted for age, sex, and race/ethnicity, we estimated CR as random slopes of global cognition, capturing individual deviations from expected cognitive decline given a specific neuroimaging profile. CR estimates were z-scored; positive values indicated better CR. Second, using general linear models, we assessed whether modifiable social and health-related risk factors at Exam 1 predicted CR a decade later. Each predictor’s contribution to CR was assessed by percentage change in R-square by comparing fully-adjusted vs. reduced-models. At first neurocognitive assessment, study participants averaged 66 years old (SD = 8) and 46% were male. In fully-adjusted models, educational attainment (β = 0.19;95% CI=0.11,0.27) and family income (β = 0.12;95%CI=0.07,0.18) were independently associated with higher CR. More depressive symptoms and higher waist girth were independently associated with lower CR (β=-0.01;95% CI=-0.02,-0.01; β=-0.01;95% CI=-0.02,-0.01, respectively). Education and income were main contributors to CR (%R-squared Change=39% and 22%, respectively). Early favorable social (education, income) and health factors (depressive symptoms, waist girth) may promote CR decades later.