An integrative, hypothesis‐free, multi‐omics approach uncovers biological pathway alterations in Alzheimer’s disease

Christopher Clark,Loic Dayon,Mojgan Masoodi,Julius Popp

DOI: https://doi.org/10.1002/alz.038563

2020-12-01

Abstract:In recent years, different ‘omics approaches have identified multiple pathway alterations in Alzheimer’s disease (AD). However, the pathogenic mechanisms are still not fully understood. Here, we test an integrative approach combining multiple ‘omics results to identify and explore in depth central nervous system (CNS) pathway alterations in AD. Multi‐omics data were obtained from a well‐characterised cohort of community‐dwelling elder subjects including healthy volunteers with normal cognition and memory clinic patients with either mild cognitive impairment or mild AD dementia. Inflammatory biomarkers were quantified by sandwich immunoassays in cerebrospinal fluid (CSF) samples; metabonomics (covering amino acids, carboxylic acids, and central energy metabolism), ionomics, proteomics, and lipidomics were performed using nuclear magnetic resonance or mass spectrometry coupled to liquid, supercritical fluid or gas chromatography. Alterations associated with AD were identified using Elastic‐Net regression at single ‘omics level and Multi‐Omics Factor Analysis, a hypothesis‐free Bayesian approach, for multi‐omics integration. Pathway enrichment analysis was performed in the Reactome database for selected analytes. We identified five major dimensions of heterogenicity (latent factors, LFs) within the cohort. We evaluated the performance of this model by separating participants in the cohort with low vs. high CSF expression levels of amyloid, tau and phosphorylated tau (P‐tau). Best performance was obtained for tau and P‐tau with 93 and 96% sensitivity, respectively. Using these LFs we could furthermore identify analytes associated with either amyloid pathology (21) or neurodegeneration (37). We also identified specific patterns of expression of these analytes across LFs, revealing interactions between multiple ‘omics modalities. Pathway enrichment analysis also revealed an over representation of the hemostasis, immune response and extracellular matrix signalling pathways. In this proof‐of‐principle work, we demonstrate that integrating expression from multiple ‘omics levels allows for a more comprehensive exploration of alterations related to AD pathology at CNS level. We identify novel metabolic alterations at molecular and pathway levels, and describe their relationships with amyloid pathology, neuronal injury, and tau hyperphosphorylation.