Abstract
OBJECTIVE: To determine the independent and joint associations of five markers of white matter integrity, including white matter hyperintensities (WMH), extracellular free water (FW), fractional anisotropy (FA), peak width of skeletonized mean diffusivity (PSMD), and Diffusion tensor image analysis along the perivascular space index (ALPS) on cognitive performance and its trajectory in cognitively diverse individuals. METHODS: 574 participants from the University of California, Davis Alzheimer's Disease Research Center (UCD ADRC) longitudinal cohort, aged 77 ± 7 years received yearly comprehensive clinical evaluations and a baseline MRI exam. Baseline MRI measures, including WMH, FW, FA, PSMD, and ALPS, were computed for each individual and used as independent variables to explain baseline and change in episodic memory (EM) and executive function (EF) using linear regression with stepwise adjustment. Bayesian Model Averaging (BMA) was then applied to derive robust estimates of each marker's contribution to cognition and its longitudinal trajectory, accounting for their joint inclusion in the same model. RESULTS: Analyses showed that higher baseline WMH, FW, and PSMD, as well as lower FA and ALPS, were significantly associated with poorer cognitive performance (p < 0.05). These associations remained robust after adjusting for relevant covariates-including age, sex, education, hypertension, diabetes, and hippocampal volume-except for FA and ALPS, which were no longer associated with EM (p > 0.05). Higher baseline WMH, FW, and PSMD, and lower FA, were also significantly associated with annual decline in EF and EM, whereas ALPS showed no association with cognitive change (p > 0.05). After covariate adjustment, these associations remained significant, except for PSMD and FA which were no longer significantly associated with EM trajectory. Joint modeling using BMA identified FW and WMH as the most likely contributors to both baseline performance and change in EF and EM, with posterior inclusion probabilities exceeding 50 %. CONCLUSIONS: This study identified both cross-sectional and longitudinal associations between five markers of white matter integrity and cognitive performance and decline. Using BMA-a method designed to disentangle the specific contribution of each marker while accounting for multicollinearity-we found that, among the five markers, FW and WMH emerged as the most probable candidates to explain the course of cognitive decline in EM and EF.