Age-Related Brain Atrophy Mediates a Composite Outcome of One-Year Ischemic Stroke Recurrence and All-Cause Mortality Through YKL-40-Related Inflammatory Pathways: A Structural Equation Model.

Abstract

Chronological age is a strong predictor of poor outcomes after ischemic stroke but may not fully capture underlying biological vulnerability. This study investigated whether age-related brain atrophy and plasma YKL-40, a marker of astroglial inflammation, mediate the association between age and the one-year risk of ischemic stroke recurrence or all-cause mortality. Data were obtained from 4,305 participants enrolled in the Third China National Stroke Registry. Baseline brain atrophy was quantified from structural T1-weighted MRI using an automated deep learning-based pipeline (FastSurfer), yielding hemispheric cortical and white matter volumes that were modeled as indicators of a latent atrophy construct. Structural equation modeling was applied to estimate direct and indirect pathways linking age, brain atrophy, YKL-40, and one-year composite outcomes, adjusting for sex, atrial fibrillation or flutter, hypertension, and diabetes, with indirect effects evaluated using 5,000 bootstrap resamples. The total effect of age on one-year outcomes was not significant (β = -0.011; 95% CI, - 0.060 to 0.038), whereas the total indirect effect was significant (β = 0.028; 95% CI, 0.005-0.050). Mediation was primarily attributable to YKL-40 (β = 0.026; 95% CI, 0.010-0.041), while the independent contribution of brain atrophy was minimal. The indirect effects accounted for approximately 255% of the total effect, indicating a suppressor-type mediation pattern in which direct and indirect effects operated in opposite directions. These findings suggest that advancing age confers increased long-term risk of stroke recurrence or death predominantly through astroglial inflammation rather than through a direct age effect, with brain atrophy modestly amplifying this risk, supporting the integration of neuroimaging and inflammatory biomarkers into translational stroke risk stratification.