Reduced muscle expression heterogeneity by categorising diabetic subjects resulted in improved statistical significance for trascriptome comparison

Abstract

The information gained through studies investigating muscle gene expression profiles (GEPs) for diabetes biomarkers identification, has resulted in the general belief that muscle transcriptome is relatively unaltered while the subject gradually progresses from a healthy to a diabetic condition. In this study, we focused on identifying the diabetes subgroup based on the subject's muscle gene expression profile to reduce the expression heterogeneity and regain statistical significance in transcriptome comparison. Given the muscle GEPs of type-II diabetic (T2D) and normal glucose tolerant (NGT) subjects, the R-package WGCNA was used to identify a set of correlated genes termed a module. The Random Forest classifier trained over expression profiles of genes belonging to respective modules was used to classify subjects into subgroups. The study was able to categorize early-stage diabetic (ESD) and advance stage diabetic (ASD) subjects based on muscle gene expression profiles. The ASD subjects differed significantly from ESD subjects with respect to diabetes-related clinical traits. Comparison between ASD and NGT subjects showed significant up-regulation of muscle atrophy marker atrogin-1 along with the FOXO signaling pathway. The expression of PDK4, an important inhibitor of glycolytic flux was up-regulated whereas, glucose oxidation biomarkers, HK2 and LDHB were down-regulated in ASD subjects. The result obtained from this study highlights muscle-associated gene expression changes that could be used to categorize diabetic subjects into biologically relevant subgroups.