Neuroimaging PheWAS and molecular phenotyping implicate PSMC3 in Alzheimer's disease.

Neuroimaging genetics has advanced our understanding of Alzheimer's disease (AD); however, frameworks using functional genomics are needed to elucidate mechanisms connecting loci to neurological outcomes. To address this need, we explored relationships between AD-associated variants and disease via their impact on gene expression and neuroanatomical phenotypes. We mapped established AD genes to neuroimaging traits using the NeuroimaGene Atlas and predicted transcript-driven neurological features of AD by comparing gene-derived neuroimaging features with clinical neuroimaging data. Genetic covariance analyses were performed to characterize shared genetic architecture between AD endophenotypes and neuroimaging features, and to identify neuroimaging features associated with a family history of dementia. Our analyses implicate PSMC3 as a contributor to AD pathophysiology and identify AD endophenotypes, including dementia family history, linked to frontal cortex thickness and volume, as well as changes in cerebrospinal fluid volume. Our findings prioritize AD genes whose regulation is associated with vulnerable brain regions, offering a potential mechanistic framework for downstream functional validation.