Genetic Architecture of Pure Alzheimer's Disease and Mixed Pathology

ABSTRACT It is widely acknowledged that a minority of people – around 20% - with neuropathologically confirmed Alzheimer’s dementia have pure Alzheimer’s pathology characterized by amyloid plaques and tau tangles. The remaining 80% of individuals with neuropathological Alzheimer’s pathology have one or more additional pathologies, including Lewy bodies, vascular brain injury, and TDP-43. A complementary and necessary strategy to clinical case-control Alzheimer’s disease is to perform genome-wide association studies (GWAS) using neuropathologically confirmed cases where “pure AD” and mixed pathology cases are distinct; this will provide more definitive risk loci to elucidate AD and related dementias (ADRD). The overarching goal of this proposal is to address this knowledge gap and further scientific understanding of the genetic architecture of Alzheimer’s disease and mixed pathology by integrating detailed neuropathology-related phenotypic information together with state-of-the-art statistical approaches and variety of omics data. Aim 1: Isolate the genetic architecture of individuals with pure AD vs individuals with mixed pathology in reference to controls and investigate if those associations are driven by a particular pattern of neuropathology. 1a. We will perform genome-wide association studies (GWAS) testing AD controls vs. pure AD vs. mixed pathology using a multinomial regression framework. We will also test if associations are driven by a particular neuropathology which will further our understanding of these neuropathological outcomes. 1b. We will refine standard genetic association results by integrating biological knowledge will identify additional candidate genes associated with mixed pathology by performing gene-network analysis by integrating data on human protein- protein interactions (PPIs) to further prioritize genes obtained from Aim 1. Aim 2: Identify risk factors for pure AD vs. mixed pathology. We hypothesize that among others, rapid decline in memory is associated with mixed pathology. 2a. We will use machine learning approaches and harmonized risk factors to isolate those that are associated with either pure or mixed pathology. 2b. We will perform polygenic score-based biobank-wide association scan and of our primary outcome using 1,738 heritable traits measured from UK Biobank and other consortia. This will help us gain novel insights to any shared genetic basis of biological processes between our phenotype and those traits. We will follow-up significant PRSs using Mendelian Randomization to probe causal risk factors of pure AD and mixed pathology. Aim 3. Test for disease risk that is impacted by variation in genes that affect the functions and interactions of diverse cells types. We will perform differential gene expression and abundance analyses across multiple brain regions for 3a) outcome of interest and 3b) loci from recent GWASs of AD and Aim 1 that will help us understand differential vulnerability of specific cell types, and push towards a mechanistic understanding of pure AD vs. mixed pathology progression by identifying dysregulated gene networks within affected cell types.