Background

Acute coronary syndrome (ACS) is a heterogeneous syndrome that integrates pathoetiologically distinct phenotypes (non-ST-elevation myocardial infarction, NSTEMI; STEMI; unstable angina pectoris, UAP). A large-scale proteomic characterization of these phenotypes had not yet been performed in humans.

Methods

In the prospective, two-armed cohort study ProsPECTUS (N=2,546) of the University Medical Center Mainz, individuals presenting to the Chest Pain Unit on suspicion of ACS, and individuals referred for planned coronary angiography, were consecutively enrolled. Participants underwent a highly standardized medical examination, including biomaterial collection before percutaneous coronary angiography. In each individual, 181 plasma proteins were profiled in venous samples using immuno-qPCR multiplex assays. Elastic net-regularized logistic regression adjusted for age, sex, and estimated glomerular filtration rate (eGFR) was applied to identify phenotype-specific protein signatures, against an atherosclerosis-free reference group (Athero⁻). In the independent MyoVasc cohort (NCT04064450; N=3,289) representing chronic heart failure, in part of ischemic origin, the protein signatures were replicated and regressed on markers of subclinical organ damage and outcome. Immune cell deconvolution on the basis of RNA sequencing of whole blood samples in the MyoVasc cohort allowed inference regarding the cellular origin of the proteins.

Results

Proteomics data was available in 866 individuals with suspected ACS of the ProsPECTUS cohort (age 67.7±12.3 years; 24.5% women; NSTEMI: n=285; STEMI: n=137; UAP: n=124; Athero⁻: n=127) and 3,188 individuals in the MyoVasc study. STEMI was differentiated from Athero⁻ individuals on the basis of a 40-protein signature (leave-one-out cross-validated [CV] AUC: 0.91); NSTEMI vs Athero⁻ by 48 proteins (CV-AUC: 0.74), and UAP vs Athero⁻ by 18 proteins (CV-AUC: 0.69). STEMI (r=0.10, p=0.004) and NSTEMI protein signatures (r=0.12, p=0.001) correlated moderately with recency of symptom onset. STEMI [HR: 1.29, 95% CI: 1.06-1.57] and NSTEMI protein signatures [HR: 1.51, 95% CI: 1.22-1.87] predicted death following MI in ProsPECTUS, and incident MI, 3-point MACE, cardiac death and all-cause death in the independent MyoVasc cohort (all p<0.001). Both signatures also correlated positively with the presence of carotid plaques, intima-media thickness, arterial stiffness, left ventricular mass and left ventricular E/e’, while being inversely associated with ankle-brachial index, augmentation index, aortic pulse wave velocity, and left ventricular ejection fraction (all p<0.05). The STEMI signature was associated with a higher estimated monocyte (p<0.0001) fraction, while the NSTEMI signature was associated with a higher estimated fraction of CD8⁺ (p<0.0005) and regulatory T lymphocytes (p=0.015).

Conclusion

This study has defined ACS phenotype-specific protein signatures predictive of clinical outcome in two independent cohorts, and associated with markers of subclinical organ damage. The MI signatures were both partially reflective of acute-phase processes. The STEMI signature was particularly strongly enriched in acute atherothrombotic processes, including platelet degranulation, coagulation, and active fibrinolysis. Immune cell deconvolution provided additional evidence for a more acute phase nature of the STEMI-related proteins, while suggesting involvement of later-onset adaptive immunity in the NSTEMI signature.