Key result
Machine learning-identified 5-gene signature demonstrates AUC >0.83 for diagnosing ischemic cardiomyopathy.
Why the study?
A systematic computational dissection of the molecular mechanisms of ischemic cardiomyopathy and a coherent pipeline from discovery to potential therapeutics was lacking.
Effect estimate: AUC > 0.83
A machine learning and single-cell transcriptomic analysis identified a 5-gene signature associated with fibroblasts in ischemic cardiomyopathy and proposed LDN-193189 as a potential therapeutic candidate.
Hypothesis-generating for fibroblast biomarkers in ischemic cardiomyopathy; leaves open LDN-193189 as a candidate pending validation.
Ischemic cardiomyopathy (ICM), a leading cause of heart failure, is characterised by complex cellular heterogeneity and a dysregulated microenvironment. A systematic computational dissection of its molecular mechanisms and a coherent pipeline from discovery to potential therapeutics is currently lacking. We integrated single‐cell RNA sequencing (scRNA‐seq) data from ICM patients with four independent bulk transcriptomic cohorts. A cardiac cellular atlas was constructed, and candidate genes were filtered through differential expression analysis. Subsequently, a benchmark of 127 machine learning algorithm‐feature selection combinations was performed to identify robust diagnostic hub genes. Their functions were validated at single‐cell resolution via UCell scoring, pseudotime trajectory analysis, and virtual knockout perturbations using scTenifoldKnk. The immune infiltration landscape was assessed using CIBERSORT and MCP‐counter. Finally, computational drug repositioning and molecular docking were employed to screen for potential compounds targeting the hub genes. Machine learning identified a core 5‐gene signature ( NPPA, HTRA1, LUM, ASPN , and OGN ) demonstrating excellent diagnostic performance across independent datasets (AUC > 0.83). Single‐cell analysis revealed that these genes were most abundantly expressed in fibroblasts and were consistently upregulated in ICM. Pseudotemporal trajectory analysis illustrated their dynamic expression patterns. Virtual knockout and functional enrichment indicated that four of these genes ( ASPN, HTRA1, LUM, OGN ) significantly perturbed pathways related to the regulation of inflammatory response. Immune profiling revealed increased infiltration of fibroblasts and plasma cells in ICM. Molecular docking identified the compound LDN‐193189 as a potential lead molecule with high predicted binding affinity (binding energy < −9 kcal·mol −1 ) for ASPN, LUM , and OGN . Through multi‐omics integration and computational biology, this study systematically delineates a fibroblast‐centric molecular network involving key hub genes and an altered immune microenvironment in ICM and computationally proposes a potential therapeutic candidate. These findings provide a crucial computational foundation and experimental direction for understanding ICM pathology and developing novel therapeutic strategies.
No takes yet. Share an insight, caveat, or question.
Yu et al. (2026) studied Ischemic cardiomyopathy. 5-gene signature (NPPA, HTRA1, LUM, ASPN, OGN) was evaluated on Diagnostic performance (AUC > 0.83). A machine learning-identified 5-gene signature (NPPA, HTRA1, LUM, ASPN, OGN) demonstrated excellent diagnostic performance for ischemic cardiomyopathy across independent datasets (AUC > 0.83).
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: