Neuroimage-genomic fingerprints for subtyping and prediction of Alzheimer's Disease and related dementia

Wake Forest University Health Sciences, Winston-Salem NC

Investigators

Abstract

Project Summary/Abstract Alzheimer’s disease (AD) is a multifactorial neurodegenerative disease with heterogeneous pathologies that affect the cognitive function of the brain, eventually leading to dementia. The heterogeneity of AD is manifested both in terms of diverse neurodegeneration patterns and clinical presentations, which poses challenges in developing effective interventions. Specifically, individual genetic variations might be associated with such disease manifestation variability. It is crucial to unravel the intrinsic heterogeneity within the AD spectrum for achieving personalized diagnosis, and eventually developing effective targeted intervention and disease- modifying treatment. However, several obstacles and knowledge gaps remain: currently, the genotype- phenotype association that defines distinctive AD subtypes is poorly understood, and current AD prediction methods preclude the development of strategies for accurate selection of at-risk clinical trial participants and accurate disease onset time estimation. The objective of the proposed study is to establish the connection between AD-related genomic markers and neuroimaging phenotypes to determine the heterogeneity of AD subtypes and their joint association with the clinical onset of dementia. We hypothesize that a) distinctive genomic factors are associated with diverse AD-related neuropathological and clinical progression patterns; and b) the genotype-phenotype interaction is dynamic along the AD progression trajectory, which in turn regulates the clinical progression of dementia. We plan to develop data-driven computational models using multi-modal imaging-genomics information, to test these hypotheses with the following two Specific Aims: (1) to characterize distinctive subtypes of AD neuropathological patterns by constructing clinically relevant computational neuroimaging-genomic fingerprints, and (2) to achieve accurate prediction of AD progression through subtype-aware artificial intelligence models with novel genomic-neuroimaging integration. Our multi- disciplinary team of experts in AD mechanism and clinical manifestation, neuroimaging, genomics, gerontology, artificial intelligence, and biostatistics will construct and validate the generalizable artificial intelligence models by utilizing the available data from the Alzheimer's Disease Sequencing Project (ADSP) and the ADSP Phenotype Harmonization Consortium (ADSP-PHC), which is a multi-institutional effort for harmonizing genomic and phenotypical data to create the largest harmonized dataset available in Alzheimer's Disease research collected from over 39 AD-related cohorts to produce a large-scale, racially diverse, standardized set of clearly defined data. This study will be one of the first to establish a comprehensive picture of AD pathogenesis that (i) connects genomic-level risk factors with multi-modal neuroimaging-based neuropathological profiles, and (ii) determines their combined effects on the clinical syndrome that leads to dementia, thereby generating crucial knowledge to inform the development of personalized interventions.