Fig. 3
WMH volume is associated with worse baseline cognitive performance and accelerated decline over time. Analysis: We used multiple linear regression with WMH segmentation maps as the dependent variable and cognitive performance as the independent variable. We accounted for the effects of hypertension, Aβ positivity, age, sex, years of education, mean background intensity, and total intracranial volume. We used an explicit mask to constrain the analysis to voxels in which data for at least five subjects were available. We thresholded contrast maps at 5% and adjusted p-values for FDR. Illustration: Regression results with memory (top left), executive function (bottom left), PACC5 (top right), and ΔPACC5 (bottom right) as independent variables. Findings: We found WMH to be significantly associated with worse cognitive performance at baseline and sharper decline within a 3-year period. Such relationships were consistently evident around portions of the anterior thalamic radiation neighbouring the thalamus (memory: xyzMNI = [− 8, − 1, 3], T = 7.00, DoF = [1.0, 366.0], pFDR = 1.44 × 10−5; executive: xyzMNI = [− 9, 0, 5], T = 6.74, DoF = [1.0, 366.0], pFDR = 2.85 × 10−5; PACC5: xyzMNI = [− 8, 1, 4], T = 7.20, DoF = [1.0, 366.0], pFDR = 8.43 × 10−6; ΔPACC5: xyzMNI = [− 7, 2, 2], T = 4.53, DoF = [1.0, 217.0], pFDR = 5.12 × 10−3). Frontal and occipital WMH also coincided with a faster cognitive decline (frontal peak at the level of the genu of the corpus callosum: xyzMNI = [− 1, 23, 4], T = 5.19, DoF = [1.0, 217.0], pFDR = 1.37 × 10−3; occipital peak at the level of the forceps major and inferior fronto-occipital longitudinal fasciculus: xyzMNI = [17, − 81, 2], T = 4.69, DoF = [1.0, 217.0], pFDR = 1.23 × 10.−2)