Cross-cohort blood transcriptome analysis identifies an eight-gene diagnostic index and inflammatory programs in acute myocardial infarction
- 1Emergency Medicine Department, Cangnan Hospital of Wenzhou Medical University
Abstract
Background: Globally, acute myocardial infarction (AMI) is the foremost cause of cardiovascular fatality, with dysregulated immune-inflammatory cascades serving as master regulators of both acute cardiomyocyte necrosis and subsequent maladaptive remodeling. Despite the prognostic potential of individual immune biomarkers, no multi-omics machine learning (ML)-derived immune gene signature has been established for predicting AMI outcome.
Methods: Peripheral blood transcriptomes from four Gene Expression Omnibus datasets (GSE59867 as the discovery cohort; GSE48060, GSE60993, and GSE66360 as independent validation cohorts; total 348 participants) were subjected to study-wise differential expression analysis and marker-based immune scoring without cross-platform batch harmonization. Weighted gene co-expression network analysis (WGCNA) and nested machine learning (L2-regularized logistic regression, linear support vector machine (SVM), random forest, and gradient boosting) were used to derive an eight-gene diagnostic index. Signature performance was evaluated in three independent validation cohorts, contextualized at the single-cell resolution (GSE121893), and experimentally supported by in vitro reverse transcription quantitative polymerase chain reaction (RT-qPCR) using American Type Culture Collection (ATCC)-authenticated cell lines (THP-1 TIB-202 and EA.hy926 CRL-2922).
Results: AMI specimens showed marked enrichment of pro-inflammatory M1-like macrophages, neutrophils, and exhausted CD8+ T cells compared to healthy donors. The WGCNA turquoise module exhibited the strongest phenotypic correlation (r = 0.78, p < 0.001). Convergent machine learning analysis yielded an eight-gene diagnostic index (BAX, EDA2R, MCEMP1, SOCS3, ASGR2, RNASE1, ADM, and KPNA5) with a repeated out-of-fold area under the receiver operating characteristic curve (AUC) of 0.984 in the discovery cohort and independent validation AUCs of 0.863, 0.655, and 0.876. Two exploratory binary-outcome analyses (6-month heart failure progression and 18-month recurrent events) yielded AUCs of 0.819 and 0.569, respectively; these small-sample results are descriptive only and do not support Kaplan–Meier or hazard ratio claims. In vitro RT-qPCR in ATCC-authenticated THP-1-derived macrophages (lipopolysaccharide/ATP stimulation) and EA.hy926 cells (hypoxia/reoxygenation) confirmed elevated levels of CXCL10 (3.82 ± 0.31-fold), IL6 (4.15 ± 0.44-fold), NLRP3 (2.94 ± 0.28-fold), and CASP1 (3.47 ± 0.38-fold) under AMI-relevant stimulation conditions ( n = 3 independent biological replicates).
Conclusions: An eight-gene diagnostic index (BAX, EDA2R, MCEMP1, SOCS3, ASGR2, RNASE1, ADM, and KPNA5), derived using nested machine learning and validated in three independent cohorts, achieved robust discriminative accuracy for AMI diagnosis. Elevated inflammatory transcripts (CXCL10, IL6, NLRP3, and CASP1) confirmed by in vitro RT-qPCR in ATCC cell lines reflect concurrent inflammasome activation and immune dysregulation, offering a mechanistic rationale for precision immune-targeted intervention strategies.
Related
- Editorial: Exploring the impact of inflammatory processes on the course of acute myocardial infarction: pathways and therapies
- Inflammatory Signatures from the Complete Blood Count and In-Hospital Mortality in Acute Myocardial Infarction
- Phenotypic screening identifies kenpaullone as a prolymphangiogenic compound to improve heart repair following myocardial infarction
- Interpretable machine learning analysis of immune-inflammatory biomarkers for predicting no-reflow in patients with acute myocardial infarction after percutaneous coronary intervention.
- Global Disparities in Diagnostic and Therapeutic Strategies for Myocardial Infarction With Non-obstructive Coronary Arteries: An International Survey
BibTeX
@article{pan2026crosscohort,
title = {{Cross-cohort blood transcriptome analysis identifies an eight-gene diagnostic index and inflammatory programs in acute myocardial infarction}},
author = {Chengfeng Pan and Zunjiang Chen and Jie Li and Shangli Xie and Xingken Fan},
journal = {Frontiers in Cardiovascular Medicine},
year = {2026},
month = oct,
doi = {10.3389/fcvm.2026.1904246},
url = {https://www.frontiersin.org/articles/10.3389/fcvm.2026.1904246/full},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Cross-cohort blood transcriptome analysis identifies an eight-gene diagnostic index and inflammatory programs in acute myocardial infarction
AU - Chengfeng Pan
AU - Zunjiang Chen
AU - Jie Li
AU - Shangli Xie
AU - Xingken Fan
PY - 2026
DA - 2026/10/19
JO - Frontiers in Cardiovascular Medicine
DO - 10.3389/fcvm.2026.1904246
UR - https://www.frontiersin.org/articles/10.3389/fcvm.2026.1904246/full
AB - Background: Globally, acute myocardial infarction (AMI) is the foremost cause of cardiovascular fatality, with dysregulated immune-inflammatory cascades serving as master regulators of both acute cardiomyocyte necrosis and subsequent maladaptive remodeling. Despite the prognostic potential of individual immune biomarkers, no multi-omics machine learning (ML)-derived immune gene signature has been established for predicting AMI outcome. Methods: Peripheral blood transcriptomes from four Gene Expression Omnibus datasets (GSE59867 as the discovery cohort; GSE48060, GSE60993, and GSE66360 as independent validation cohorts; total 348 participants) were subjected to study-wise differential expression analysis and marker-based immune scoring without cross-platform batch harmonization. Weighted gene co-expression network analysis (WGCNA) and nested machine learning (L2-regularized logistic regression, linear support vector machine (SVM), random forest, and gradient boosting) were used to derive an eight-gene diagnostic index. Signature performance was evaluated in three independent validation cohorts, contextualized at the single-cell resolution (GSE121893), and experimentally supported by in vitro reverse transcription quantitative polymerase chain reaction (RT-qPCR) using American Type Culture Collection (ATCC)-authenticated cell lines (THP-1 TIB-202 and EA.hy926 CRL-2922). Results: AMI specimens showed marked enrichment of pro-inflammatory M1-like macrophages, neutrophils, and exhausted CD8+ T cells compared to healthy donors. The WGCNA turquoise module exhibited the strongest phenotypic correlation (r = 0.78, p < 0.001). Convergent machine learning analysis yielded an eight-gene diagnostic index (BAX, EDA2R, MCEMP1, SOCS3, ASGR2, RNASE1, ADM, and KPNA5) with a repeated out-of-fold area under the receiver operating characteristic curve (AUC) of 0.984 in the discovery cohort and independent validation AUCs of 0.863, 0.655, and 0.876. Two exploratory binary-outcome analyses (6-month heart failure progression and 18-month recurrent events) yielded AUCs of 0.819 and 0.569, respectively; these small-sample results are descriptive only and do not support Kaplan–Meier or hazard ratio claims. In vitro RT-qPCR in ATCC-authenticated THP-1-derived macrophages (lipopolysaccharide/ATP stimulation) and EA.hy926 cells (hypoxia/reoxygenation) confirmed elevated levels of CXCL10 (3.82 ± 0.31-fold), IL6 (4.15 ± 0.44-fold), NLRP3 (2.94 ± 0.28-fold), and CASP1 (3.47 ± 0.38-fold) under AMI-relevant stimulation conditions ( n = 3 independent biological replicates). Conclusions: An eight-gene diagnostic index (BAX, EDA2R, MCEMP1, SOCS3, ASGR2, RNASE1, ADM, and KPNA5), derived using nested machine learning and validated in three independent cohorts, achieved robust discriminative accuracy for AMI diagnosis. Elevated inflammatory transcripts (CXCL10, IL6, NLRP3, and CASP1) confirmed by in vitro RT-qPCR in ATCC cell lines reflect concurrent inflammasome activation and immune dysregulation, offering a mechanistic rationale for precision immune-targeted intervention strategies.
SN - 2297-055X
ER -