Alzheimer's Disease (AD) is the most common form of dementia in the United States, nearly 5 million people have the disease. Perivascular space (PVS) changes were previously considered an inconsequential finding of AD, but there is recent interest regarding its role in AD progression 1,2 . Here, we investigate the association between several PVS morphological features (from in vivo MRI) and AD pathology (from postmortem histology). Data were derived from Alzheimer's Disease Neuroimaging Initiative (ADNI)3 . Demographic information (sex, race, age) and ADNI pathologic scores were collected. Scores were for Amyloid-B, Tau, and neurofibrillary tangle status from histological images. We also assessed scores for neuronal loss and cerebral amyloid angiopathy (CAA). Using the last premortem MRI, manual segmentation of basal ganglia PVS of each individual was conducted. PVS of the white matter was also automatically mapped using our recently developed technique 1,4 . We then used linear regression to assess if premortem PVS features are predictors of postmortem AD pathology. Basal ganglia PVS count was a predictor of glial Tau pathology in the superior and middle temporal gyri (t=4.1, p=0.00003) and the parahippocampal gyrus (t=3.5, p=0.0005). Basal ganglia PVS volume was also a predictor of glial Tau pathology in the Entorhinal Cortex (t=3.14, p=0.002). Interestingly, PVS volume fraction in white matter centrum semiovale was a predictor of vascular pathology, including CAA in the Amygdala (t=2.6, p=0.008) and Entorhinal cortex (t=2.9, p=0.003). We found that PVS morphology in the basal ganglia was a predictor of glial Tau pathology but not of neurofibrillary tangle pathology. Furthermore, PVS morphology in the basal ganglia was not a predictor of Amyloid-Bdiffuse or cored plaques but interestingly was a predictor for vascular damage.
Read full abstract