In total, 249 Japanese individuals aged 89–94 years were evaluated using 3-Tesla magnetic resonance imaging to identify the structural correlates of cognition. White matter hyperintensities (WMH) were segmented on fluid-attenuated inversion recovery scans with TrUE-Net and classified into quartiles. Gray and white matter volumes were each divided into quartile categories via T1-weighted imaging using FreeSurfer. Diffusion tensor imaging was conducted to assess microstructural damage based on peak width of skeletonized mean diffusivity (PSMD), with PSMD and its interaction with education standardized (z-scores). Cognition was assessed using the Mini-Mental State Examination (MMSE), which is a 30-item test. Robust linear regression was used to examine the association of MMSE scores with age, educational level (z), sex, apolipoprotein E ε4 allele status, WMH quartile, quartile categories of volumes, PSMD (z), and the PSMD×education interaction (z). Voxel-wise lesion–symptom mapping identified the effects of WMH. Based on the tract-based spatial statistics, diffusion tensor imaging metrics were associated with cognition. Older age and WMH burden were associated with a lower MMSE, and education was positively associated with MMSE. PSMD did not have main effects. However, its interaction with education was significant, indicating reduced educational benefit with greater damage. The bilateral inferior fronto-occipital fasciculi and left inferior longitudinal fasciculus could have played a role. In particular, a higher axial diffusivity in the left inferior fronto-occipital fasciculus was associated with a lower cognition. In this cohort, cognition was correlated most strongly with small-vessel disease and lifetime education, particularly the integrity of white matter tracts.
UEDA et al. (Sun,) studied this question.