Fig. 3
PVS enlargement is associated with AD biomarker profiles. PVS volumes in all plots were Box-Cox and z-transformed and adjusted for linear and quadratic age effects, sex, years of education, total intracranial volume and regional WMH of presumed vascular origin. A Histogram ridgeline plots of distribution of CSO-PVS rates of change across diagnostic groups. B Increase of CSO-PVS is dependent on the AD biomarker profile. Colour of individual trajectories corresponds to diagnosis at time of entry to DELCODE study. Subjects with A + T- or A + T + status show higher rates of change as compared to A-T-. C Histogram ridgeline plots of distribution of BG-PVS rates of change across diagnostic groups. D Increase of BG-PVS is dependent on the AD biomarker profile. Colour of individual trajectories corresponds to diagnosis at time of entry to DELCODE study. Subjects with A + T + status show higher rates of change as compared to A-T-. E Contrast image highlighting regions where PVS enlargement was more evident in A + T + vs. A-T- (puncorr < 0.05). We registered all PVS segmentation maps to a DELCODE-specific Multi-Brain (MB) toolbox template [58] and adjusted for local volume changes introduced by normalisation in PVS segmentation maps by modulation with Jacobian determinants [59, 60]. PVS maps were smoothed with Gaussian kernels (6 mm full width at half maximum). Model was aligned with regional marginal models [54] (PVS ~ Time*AT profile + Age + Age.2 + Sex + Years of Education + Total Brain Volume)