Multiomics Reveals Biological Mechanisms Linking Macroscale Structural Covariance Network Dysfunction With Neuropsychiatric Symptoms Across the Alzheimer’s Disease Continuum

Abstract

BackgroundThe highly heterogeneity of neuropsychiatric symptoms (NPSs) hinder further exploration of their role in neurobiological mechanisms and Alzheimer’s disease (AD). We aimed to delineate NPS patterns based on brain macroscale connectomics to understand the biological mechanisms of NPSs on the AD continuum. MethodsWe constructed Regional Radiomics Similarity Networks (R2SN) for 550 participants (AD with NPSs [AD-NPS, n=376], AD without NPSs [AD-nNPS, n=111], and normal controls [n=63]) from CIBL study. We identified R2SN connections associated with NPSs, and then cluster distinct subtypes of AD-NPS. An independent dataset (n=189) and internal validation were performed to assess the robustness of the NPS subtypes. Subsequent multiomics analysis were performed to assess the distinct clinical phenotype and biological mechanisms in each NPS subtype. ResultsAD-NPS patients were clustered into severe (n=187), moderate (n=87), and mild NPS (n=102) subtypes, each exhibiting distinct brain network dysfunction patterns. A high level of consistency in clustering NPS was internally and externally validated. Severe and moderate NPSs showed significant cognitive impairment, increased plasma p-Tau181 levels, extensive decreased brain volume and cortical thickness, and accelerated cognitive decline. Gene set enrichment analysis (GSEA) revealed enrichment of differentially expressed genes in ion transport and synaptic transmission with variations for each NPS subtype. Genome-wide association studies (GWAS) analysis defined the specific gene loci for each subtype of AD-NPS (i.e, logical memory), aligning with clinical manifestations and progression patterns. ConclusionsThis study identified and validated three distinct NPS subtypes, underscoring the role of NPSs in neurobiological mechanisms and progression of the AD continuum.