White Matter Integrity and Default Mode Network Disruption in Cognitive Impairment: A Multimodal Imaging Study From the Alzheimer's Disease Neuroimaging Initiative

Cognitive impairment involves large‐scale brain systems, but the relation between white matter microstructure and default mode network (DMN) organization remains unclear in clinically heterogeneous populations. In this secondary analysis of Alzheimer's Disease Neuroimaging Initiative data, we examined structural abnormalities, DMN organization, and cognition. The structural analysis included 1431 classified participants with diffusion tensor imaging (DTI) data. After deduplication, the multimodal source merge yielded 1452 unique participant‐visits from 905 participants. Primary analyses used participant‐level random‐intercept models adjusted for scan age, sex, education, and diagnosis, alongside first‐available‐visit HC3‐robust sensitivity analyses. Global fractional anisotropy (FA), DMN within‐network connectivity, DMN segregation, and Mini‐Mental State Examination (MMSE) scores were lower in Alzheimer's disease than in cognitively normal participants. Global FA was positively associated with DMN within‐network connectivity in the primary model (standardized beta = 0.080, 95% confidence interval [] 0.022 to 0.137, p = 0.006; q = 0.019). The prespecified tract analysis identified FDR‐supported associations for splenial, callosal, and cingulate‐cingulum FA. The global association attenuated after scanner‐platform adjustment (beta = 0.052, 95% CI −0.007 to 0.112, p = 0.082), quantifying an important multicenter limitation. DMN within‐network connectivity was not independently associated with MMSE in the primary mixed model (beta = 0.035, 95% CI −0.003 to 0.073, p = 0.071). In a same‐visit amyloid/tau subset (216 observations, 180 participants), diagnosis‐adjusted analyses showed no non‐monotonic imaging pattern. In this audited multimodal cohort, we identified diagnosis‐related group differences and a positive participant‐level structure–function association between white matter integrity and DMN organization, underscored by anatomically heterogeneous tract‐level effects. Acknowledging cross‐sectional and scanner‐variance constraints, these findings provide a hypothesis‐generating basis for future investigations into coupled structural‐functional network disruptions across the Alzheimer's continuum.