Sequential analysis of myocardial gene expression with phenotypic change: Use of cross-platform concordance to strengthen biologic relevance

Objectives To investigate the biologic relevance of cross-platform concordant changes in gene expression in intact human failing/hypertrophied ventricular myocardium undergoing reverse remodeling. Background Information is lacking on genes and networks involved in remodeled human LVs, and in the associated investigative best practices. Methods We measured mRNA expression in ventricular septal endomyocardial biopsies from 47 idiopathic dilated cardiomyopathy patients, at baseline and after 3-12 months of beta-blocker treatment to effect left ventricular (LV) reverse remodeling as measured by ejection fraction (LVEF). Cross-platform gene expression change concordance was investigated in reverse remodeling Responders (R) and Nonresponders (NR) using 3 platforms (RT-qPCR, microarray, and RNA-Seq) and two cohorts (All 47 subjects (A-S) and a 12 patient "Super-Responder" (S-R) subset of A-S). Results For 50 prespecified candidate genes, in A-S mRNA expression 2 platform concordance (C-cpT), but not single platform change, was directly related to reverse remodeling, indicating C-cpT has biologic significance. Candidate genes yielded a C-cpT (PCR/microarray) of 62% for Responder vs. Nonresponder (R/NR) change from baseline analysis in A-S, and ranged from 38% to 100% in S-R for PCR/microarray/RNA-Seq 2 platform comparisons. Global gene C-cpT measured by microarray/RNA-Seq was less than for candidate genes, in S-R R/NR 17.5% vs. 38% (P = 0.036). For S-R global gene expression changes, both cross-cohort concordance (C-ccT) and C-cpT yielded markedly greater values for an R/NR vs. an R-only analysis (by 22 fold for C-ccT and 7 fold for C-cpT). Pathway analysis of concordant global changes for R/NR in S-R revealed signals for downregulation of multiple phosphoinositide canonical pathways, plus expected evidence of a beta(1)-adrenergic receptor gene network including enhanced Ca2+ signaling. Conclusions Two-platform concordant change in candidate gene expression is associated with LV biologic effects, and global expression concordant changes are best identified in an R/NR design that can yield novel information.