Elucidating the Role of Endothelial Dysfunction in Alzheimer Disease: Towards A New Data-Driven Disease Model

Project Abstract Structural and functional alterations in brain endothelium are observed in ≥90% of Alzheimer Disease (AD) brains in the absence of cerebrovascular disease or amyloid angiopathy. Despite its prevalence, the role of endothelial dysfunction in AD pathogenesis has not been adequately investigated due to the absence of reliable endothelial biomarkers. We have identified a novel panel of endothelial biomarkers whose brain ribonucleic acid (RNA) and cerebrospinal fluid (CSF) protein levels are significantly altered in AD compared to controls. Exciting preliminary data suggest that these proteins offer promise as novel diagnostic and prognostic biomarkers of brain endothelial dysfunction and strongly correlate with cognitive and radiological outcomes in AD. Importantly, our preliminary AD models suggest that endothelial dysfunction is an early pathological substrate which may precede amyloid and tau. We here propose to conduct the first in-depth comprehensive translational study of endothelial dysfunction in AD. In this study, we will utilize proteomics and transcriptomics to measure CSF protein and brain RNA levels of these endothelial markers in 3 large well-characterized longitudinal cohorts of late-onset sporadic AD, healthy controls, and non-AD dementias (combined n>3,400) who have been followed for 20-25 years and for whom we have detailed longitudinal cognitive, CSF, magnetic resonance imaging (MRI including ASL, FLAIR, and DTI), amyloid-positron emission tomography (amyloid-PET), and tau-PET imaging data. In Aim 1, we will examine the diagnostic and prognostic utility of CSF endothelial markers in AD and their ability to predict cognitive decline and brain atrophy over time, independently of amyloid and tau. We will also characterize baseline and longitudinal associations of endothelial dysfunction with CSF and/or imaging markers of amyloid, tau, inflammation, and neuronal/synaptic injury over 20-25 years of follow-up, including 10-15 years prior to symptom onset. In Aim 2, we will leverage event-based modeling (EBM; Aim 2A) in our cross-sectional cohorts and multifactorial data driven analyses (MFDDA; Aim 2B) in our longitudinal cohorts to propose new data-driven predictive models for AD which will elucidate the temporal ordering of fluid and imaging AD markers and the spatiotemporal progression of AD imaging markers (cerebral blood flow, amyloid and tau PET, and brain atrophy) across 37 brain regions and over 10-15 years prior to symptom onset. Functional pathway analyses of proteomic and RNA-seq data from our cohort and others will identify novel mechanisms by which endothelial dysfunction contributes to neurodegeneration, and influences synaptic plasticity, axonal repair, and abnormal protein aggregation in AD (Aim 2C). Through in-depth multi-omics analyses of endothelial dysfunction and the integration of multimodal data from large well-characterized longitudinal multiracial AD cohorts, we aim to propose a new disease model which will be the first to integrate endothelial dysfunction into the AD paradigm and will capture the molecular and pathological heterogeneity of AD beyond amyloid and tau. Findings from this study will provide novel insights into disease mechanisms and new molecular targets for AD drug discovery.