Comprehensive Summary
Liu et al. utilized an integrated proteomic, transcriptomic, and machine-learning strategy to illustrate molecular changes in acute myocardial infarction (AMI). Using plasma samples from 48 AMI patients and 50 controls, the authors found over 400 differentially expressed proteins enriched in inflammatory, immune, oxidative-stress, and remodeling pathways. Clustering analyses showed two inflammatory AMI subtypes, with a blood transcriptomic meta-analysis producing more than 1,300 differentially expressed genes (DEGs) that overlapped with prior proteomic hits. Co-expression network analysis underscored a key module, yielding 17 candidates, and an optimized feature-selection pipeline further distilled these to nine core proteins (CAMP, CLTC, CTNNB1, FUBP3, IQGAP1, MANBA, ORM1, PSME1, SPP1) with robust diagnostic performance. Cross-validation in atherosclerosis plaques, single-cell data, and spatial myocardial transcriptomics confirmed cell- as well as region-specific expression, highlighting macrophage-driven inflammation and remodeling. The study provides a concise yet informative molecular signature of AMI and identifies promising biomarker candidates.
Outcomes and Implications
The findings demonstrate that integrating proteomics with transcriptomics and machine-learning feature engineering is capable of revealing clinically relevant molecular signatures of AMI beyond standard cardiac enzymes. By identifying reproducible core proteins implicated in processes such as tissue remodeling, this approach provides potential diagnostic markers and targets that current assays do not capture. While this study is limited by modest sample size and acute-phase sampling, the analytic pipeline demonstrates a scalable strategy for biomarker discovery and points toward future work evaluating whether these protein signatures improve risk stratification, therapeutic targeting, or mechanistic understanding of AMI in broader populations.