Imaging Genetics and Biomarker Variations of Clinically Diagnosed Alzheimer's Disease

Abstract

Neuroimaging biomarkers play a crucial role in our understanding of Alzheimer’s disease. Beyond providing a fast and accurate in vivo picture of the neuronal structure and biochemistry, these biomarkers make up a research framework, defined in a 2018 as the A(amyloid)/T(tau)/N(neurodegeneration) framework after three of the hallmarks of Alzheimer’s disease. I first used imaging measures of amyloid, tau and neurodegeneration to study clinically diagnosed Alzheimer’s disease. After dividing subjects into early (onset younger than 65) and late-onset (onset of 65 and older) amyloid-positive (AD) and amyloid-negative (nonAD) groups, I saw radically differing topographical distribution of tau and neurodegeneration. AD subjects with an early disease onset had a much more severe amyloid, tau and neurodegeneration than lateonset AD. In the nonAD group, neurodegeneration was found only in early-onset FDG PET data and in a nonAlzheimer’s-like MRI and FDG pattern for late-onset. The late-onset nonAD resembled that of limbic-predominant age-related TDP-43 encephalopathy. I next utilized an imaging genetics approach to associate genome-wide significant Alzheimer’s risk variants to structural (MRI), metabolic (FDG PET) and tau (tau PET) imaging biomarkers. Linear regression was used to select variants for each of the models and included a pooled sample, cognitively normal, mild cognitive impairment and dementia groups in order to fully capture the cognitive spectrum from normal cognition to the most severely impaired. Model selected variants were replicated using voxelwise regression in an exploratory analysis of spatial associations for each modality. For each imaging type, I replicated some associations to the biomarkers previously seen, as well as identified several novel associations. Several variants identified with crucial Alzheimer’s biomarkers may be potential future targets for drug interventions.