Ex Vivo Imaging of the Aging Brain to Discover Morphology/Pathology Associations

点击翻译按钮获取中文摘要

Alzheimer's disease (AD) is associated with surprisingly high degree of pathologic heterogeneity. In most individuals diagnosed with AD at autopsy, the brain not only harbors the β-amyloid and tau pathologies that are the hallmarks of AD, but also one or more co-pathologies, including TDP-43, α-synuclein, non-AD tauopathy, and cerebral small vessel disease (SVD). The primary AD pathologies and co-pathologies all contribute to neurodegeneration in AD, but their relative contribution in different brain regions and the degree in which co- pathologies modulate the progression of primary pathologies is not well understood. It is widely recognized that it is important for clinical trials in AD to account for these additional drivers of neurodegeneration, but there is a lack of in vivo biomarkers that can reliably detect and quantify co-pathology. Pathologic heterogeneity may help explain why AD treatments targeting a single pathological mechanism have been largely ineffective. This project seeks to address this limitation by using ex vivo human brain MRI to characterize the contributions of primary AD pathologies and co-pathologies to neuronal loss and cortical thinning in AD. The project leverages a prospective dataset from 100-120 autopsies conducted at the University of Pennsylvania AD Research Center that will include high-resolution 7 Tesla MRI of intact brain hemispheres with co-registered histology at selected gray matter locations and around white matter lesions. Moreover, the temporal lobe, part of the brain where earliest and most severe AD-related neurodegeneration occurs, will be scanned at 9.4 Tesla, and undergo serial histological imaging, allowing three-dimensional mapping of tau pathology (tangles, threads, etc.) and neuronal density across the entire temporal lobe. This unique ex vivo imaging dataset will represent a convergence of structural and pathological imaging data in the same 3D space, allowing a broad range of studies analyzing trajectories of pathology deposition and pathology-neurodegeneration relationships. The specific aims of the proposal are as follows. Aim 1 is to develop deep learning-based image analysis techniques for 7 Tesla whole- hemisphere MRI, which are currently lacking, including segmentation of cortical gray matter, white matter lesions, normal-appearing white matter, and subcortical structures; groupwise registration to both ex vivo and in vivo MRI templates; and extraction of both MRI-based and histological features to characterize white matter lesions associated with SVD. Aim 2 is to analyze the complete 100-120 specimen dataset to characterize the distribution of tau pathology, neuronal loss, and cortical thinning both in the temporal lobe and in the whole brain and to describe the impact of co-pathologies on these distributions and on the relationships between them. Aim 3 is to leverage pathology-specific “signatures” extracted from analyzing this ex vivo dataset to improve the sensitivity of in vivo biomarkers for inferring the presence of co-pathology and tracking disease progression.