Abstract 14044: Leveraging Multi-Omics to Identify Biologically-Informed Clusters of Individuals With Differing Underlying Disease Risk Mechanisms Using an Innovative Network Modelling Approach
Academic Article
Overview
Research
Identity
Additional Document Info
Other
View All
Overview
abstract
Background: The rates of known cardiometabolic risk elevation are highly variable among patients suggesting that there are other risk mediators underlying diseases. We hypothesized that using cutting-edge and innovative multi-omics integrative network analysis in conjunction with the consensus clustering approach would identify clusters of individuals which may represent novel functional and/or pathophysiological risk mechanisms of cardiometabolic disorders. Methods: The genetically predicted levels of metabolites and proteins constructed from the GWAS data of Framingham/Malm and KORA-TWINS were imputed in 13,562 BioVU participants. These 2 layers of omics data were integrated using similarity network fusion in conjunction with consensus clustering approaches. The identified clusters of individuals were tested for associations with 14 cardiometabolic diseases using logistic regression adjusting for known risk factors to identify endotypic determinants of diseases. Pathway analysis comparing each cluster to the rest of participants was performed using Reactome and MetaboAnalyst. Results: We identified 5 clusters across which LDL-C was significantly (p=0.001) different. The significant associations were found for cluster 1 with coronary artery disease (p=0.04), cluster 2 with cardiomyopathy(p=0.004) and atrial fibrillation (p=0.03) and cluster 4 with congenital heart disease (p=0.03) [ Fig 1 ]. Thus, clusters 1, 2 and 4 are endotypic determinants of the above diseases. Each of these clusters was associated with distinct pathways indicating candidate mechanisms of diseases related to the clusters. Conclusions: This novel discovery-oriented multi-omics clustering method identified biologically-related subgroups of individuals and revealed novel risk mechanism and biomarkers of cardiometabolic diseases. The innovative network-based models advance precision medicine by enabling future prognostic risk prediction tools.
