Clinical Prediction, Genetic Architecture, and Environmental Modification in Alzheimer’s Disease, LATE, and Mixed Pathology - Project Summary Recent studies highlight the fact that the large public health impact of Limbic-predominant Age-related TDP-43 Encephalopathy (LATE), which affects approximate one-third of adults over age 85. LATE is frequently under- recognized or misclassified as Alzheimer’s Disease (AD). Currently, autopsy-confirmed AD and LATE neuropathologic changes (ADNC and LATE-NC) remain the gold standard for diagnosis. ADNC and LATE-NC often co-occur (Mixed-NC) in advanced age. To enable early prediction, it is urgent to: (1) accurately predict underlying neuropathologic (NP) phenotypes using ante-mortem prognostic approaches; (2) identify disease- specific genetic contributors; and (3) characterize time-varying environmental factors modifying genetic risk. Previous predictive models focused on ADNC, overlooking LATE-NC and Mixed-NC. Most existing genetic analyses relied on categorical NP outcomes and utilized biallelic variants excluding multiallelic variation. Prior gene–environment (G×E) interaction analyses typically examined one-by-one and single-time-point models. To address these gaps, we will integrate multimodalities to develop predictive models for ADNC, LATE- NC, and Mixed-NC by constructing latent scores within a unified framework. We will employ two complementary approaches: supervised structural equation modeling (SEM) to yield “Latent Brain Function Scores (LBFSs)” capturing the age-dependent and multidimensional brain functions, and pseudotime analysis to derive “Latent Disease Progression Scores (LDPSs)” tracing age-anchored trajectories of disease progression. We will include multiallelic variants in genetic analyses and construct midlife and later-life environmental risk scores (ERSs) to test comprehensive G×E interactions. By leveraging LBFSs and LDPSs as continuous NP proxies in participants with and without NP data, we will increase sample sizes and statistical power and enhance the discovery of novel genetic risks and insightful environmental modifiers. Specific Aim 1: Develop and evaluate predictive models for ADNC, LATE-NC, and Mixed-NC. Using data from the National Alzheimer’s Coordinating Center, the Alzheimer’s Disease Neuroimaging Initiative, the Religious Orders Study, the Memory and Aging Project, and the Adult Changes in Thought Study, we will perform SEM and pseudotime analysis to construct LBFSs and LDPSs and develop predictive models. Specific Aim 2: Uncover novel genetic contributors to ADNC, LATE-NC, and Mixed-NC. The Alzheimer’s Disease Sequencing Project whole-genome sequencing data will be used to identify susceptibility loci including biallelic and multiallelic variants. Derived LBFSs and LDPSs will serve as continuous NP proxies. Specific Aim 3: Examine environmental factors modifying ADNC, LATE-NC, and Mixed-NC genetic risk. NACC, ROS, and MAP data, including social, physical, behaviroal, and lifestyle factors, will be used to create midlife and later-life ERSs, by applying the item response theory-based multidimensional generalized partial credit model, and assess how these ERSs modify genetic risk on LBFSs and LDPSs.
