(375) - Weighted Gene Co-expression Network Analysis Reveals Distinct Interaction Patterns in Failing and Recovering Human Myocardium

Abstract

PurposePrevious microarray studies in LVAD myocardial samples using standard statistical methods provided inconsistent results and failed to identify a common pathway for myocardial recovery process. We sought to investigate gene-gene interaction patterns in failing and post-LVAD samples to identify novel regulators of reverse remodeling.MethodsGene expression dataset for human failing and post-LVAD paired myocardial samples (n=38) were obtained from Gene Expression Omnibus Database (series GSE974). Expression data file was log-transformed and normalized in PARTEK GS, and transferred to R for weighted gene co-expression network analysis using package WGCNA. WGCNA uses network terminology to describe co-expression patterns between genes of a network and identifies clusters of highly correlated genes (modules), which can be related to sample traits.ResultsHierarchical clustering using TOM dissimilarity measure with dynamic tree cut algorithm yielded 9 distinct HF and 7 Recovery gene modules. Recovery modules were poorly preserved in HF network suggesting distinct signaling mechanisms for the two process (Figure 1A). Module-trait relationship analysis identified a unique LVAD module which correlated with duration of support. Cellular component analysis of the "myocardial plasticity" module showed genes that were significantly mapped to extracellular region including 9 different collagen genes. Hub genes of this module included lysyl oxidase (LOX) and CD44, both of which have been implicated in the myocardial remodeling process (Figure 1B).ConclusionOur findings suggest that alterations in extracellular matrix occur in a time-dependent fashion following LVAD support and may contribute to reverse remodeling process. Analysis of gene-gene interaction patterns rather than traditional differential gene expression analysis may provide valuable mechanistic insights into myocardial recovery. PurposePrevious microarray studies in LVAD myocardial samples using standard statistical methods provided inconsistent results and failed to identify a common pathway for myocardial recovery process. We sought to investigate gene-gene interaction patterns in failing and post-LVAD samples to identify novel regulators of reverse remodeling. Previous microarray studies in LVAD myocardial samples using standard statistical methods provided inconsistent results and failed to identify a common pathway for myocardial recovery process. We sought to investigate gene-gene interaction patterns in failing and post-LVAD samples to identify novel regulators of reverse remodeling. MethodsGene expression dataset for human failing and post-LVAD paired myocardial samples (n=38) were obtained from Gene Expression Omnibus Database (series GSE974). Expression data file was log-transformed and normalized in PARTEK GS, and transferred to R for weighted gene co-expression network analysis using package WGCNA. WGCNA uses network terminology to describe co-expression patterns between genes of a network and identifies clusters of highly correlated genes (modules), which can be related to sample traits. Gene expression dataset for human failing and post-LVAD paired myocardial samples (n=38) were obtained from Gene Expression Omnibus Database (series GSE974). Expression data file was log-transformed and normalized in PARTEK GS, and transferred to R for weighted gene co-expression network analysis using package WGCNA. WGCNA uses network terminology to describe co-expression patterns between genes of a network and identifies clusters of highly correlated genes (modules), which can be related to sample traits. ResultsHierarchical clustering using TOM dissimilarity measure with dynamic tree cut algorithm yielded 9 distinct HF and 7 Recovery gene modules. Recovery modules were poorly preserved in HF network suggesting distinct signaling mechanisms for the two process (Figure 1A). Module-trait relationship analysis identified a unique LVAD module which correlated with duration of support. Cellular component analysis of the "myocardial plasticity" module showed genes that were significantly mapped to extracellular region including 9 different collagen genes. Hub genes of this module included lysyl oxidase (LOX) and CD44, both of which have been implicated in the myocardial remodeling process (Figure 1B). Hierarchical clustering using TOM dissimilarity measure with dynamic tree cut algorithm yielded 9 distinct HF and 7 Recovery gene modules. Recovery modules were poorly preserved in HF network suggesting distinct signaling mechanisms for the two process (Figure 1A). Module-trait relationship analysis identified a unique LVAD module which correlated with duration of support. Cellular component analysis of the "myocardial plasticity" module showed genes that were significantly mapped to extracellular region including 9 different collagen genes. Hub genes of this module included lysyl oxidase (LOX) and CD44, both of which have been implicated in the myocardial remodeling process (Figure 1B). ConclusionOur findings suggest that alterations in extracellular matrix occur in a time-dependent fashion following LVAD support and may contribute to reverse remodeling process. Analysis of gene-gene interaction patterns rather than traditional differential gene expression analysis may provide valuable mechanistic insights into myocardial recovery. Our findings suggest that alterations in extracellular matrix occur in a time-dependent fashion following LVAD support and may contribute to reverse remodeling process. Analysis of gene-gene interaction patterns rather than traditional differential gene expression analysis may provide valuable mechanistic insights into myocardial recovery.