Transcriptomic analysis reveals immune and metabolic signatures correlated with prognosis in DCM and NDLVC

Introduction

DCM and NDLVC are common non-ischemic HFrEF subtypes driven by genetic and environmental factors, resulting in substantial patient heterogeneity (1). The relationships between etiology, outcomes, and molecular mechanisms remain unclear. Though etiology strongly influences cardiac transcriptomics (2), prior studies were small, limiting the identification of robust molecular subtypes and their links to clinical outcomes.

Purpose

This study's goal was to identify transcriptomic subtypes in cardiac biopsies of a large cohort of DCM and NDLVC patients, connecting discovered subtypes to etiology, severity and outcome.

Methods and results

Diagnostic cardiac biopsies from 153 DCM and 30 NDLVC patients were processed for RNA-sequencing. We then performed consensus clustering and found three stable patient clusters. Linking clusters to clinical data showed that cluster C3 comprised patients with markedly lower LVEF, whereas C2 and C3 showed only moderate reduction. The C1 and C2 appear clinically similar except for higher inflammation in C1. Notably, survival analysis identified C2 with improved outcome compared to C1 and C3, independent of LVEF (Figure 1).

Differential expression (DE) analysis highlighted known markers for the severe subtype such as NPPB and NPPA (Figure 2A). Gene set enrichment analysis (GSEA) revealed that inflammatory and proliferative gene sets were suppressed in the mild subtype with positive prognosis C2, and upregulated in the severe subtype C3. Metabolic genes, on the other hand, were downregulated in the severe subtype C3 (Figure 2B). Further, gene set variability analysis (GSVA) revealed that individual scores for gene sets enriched in the C1, such as INF-α response and fatty acid metabolism, were significantly correlated with higher LVEF (Figure 2C).

Adjusting for systolic dysfunction severity, using LVEF as covariate, resulted in suppressing most DE between clusters. However, GSEA retained strong cluster distinction, with inflammatory gene sets being reduced and oxidative phosphorylation being upregulated in the good prognosis subtype C2 (Figure 2D).

Conclusions

We identified distinct transcriptomic profiles in non-end stage DCM and NDLVC, with distinctive inflammatory, metabolic, and proliferative signatures. These subtypes carry prognostic information beyond LVEF, suggesting molecular phenotyping could improve risk stratification and guide treatment strategies.

Figure descriptions

Figure 1: Principal component plot colored by clusters (A); Time from HF diagnosis, Echo LVEF, and biopsy CD3+ cells across clusters (B); Keplan-meier curve for major adverse cardiac events per clusters over 8 years and hazard ratios from a survival model correcting for LVEF (C).

Figure 2: Differential expression analysis comparing clusters: volcano plots (A) and gene set enrichments (B). GSVA results for TGF-B signaling and fatty acid metabolism, against LVEF (C). Gene set enrichment between clusters after adjusting for LVEF (D)Figure 1For image description, please refer to the figure legend and surrounding text.Figure 2For image description, please refer to the figure legend and surrounding text.