Plasma multi-omic and cardiac imaging network signatures predict poor long-term outcomes after acute myocardial infarction

BackgroundPrognostic biomarkers for patients admitted for a myocardial infarction (MI) episode are of great interest for risk stratification and follow-up care after discharge. Multi-omics analysis is a standard approach for the discovery of diagnostic and prognostic biomarkers, but few studies have evaluated the prognostic potential of molecular markers in combination with echocardiographic imaging variables. MethodsWe measured the plasma proteome and lipidome in patients discharged from an acute MI and followed for secondary outcomes in New Zealand for a median time of 4.85 years (CDCS, N=741 for network inference, N=464 for predictive analysis) and in Singapore for a median time of 2.0 years (IMMACULATE, N=190 for validation). Using a network-based integrative analysis framework iOmicsPASS+, we mapped proteins, lipids, echocardiographic imaging variables and clinical biomarkers to a unified network and identified predictive subnetwork signatures of major adverse cardiac events (MACE) and heart failure hospitalization (HFH) in CDCS, with validation in IMMACULATE. ResultsSpecific plasma proteins and lipids showed direct connections to cardiac imaging variables in the network. The gold standard biomarker, NT-proBNP, remained one of the best prognostic marker of MACE and HFH, but a number of plasma proteins involved in extracellular matrix organization, chemotaxis, inflammation, and apoptosis were also strong predictors of both outcomes. Hub proteins of subnetwork signatures were enriched in the heart, arteries, kidneys, liver and lungs. BMP10, CAPG, EFEMP1, FSTL3, RSPO4, and RELT were those directly connected to the echocardiographic variables and natriuretic peptides. In particular, EFEMP1 and FSTL3 in combination with diastolic function (E/e) were strongly predictive of HFH in both CDCS (AUC 0.78, 95%CI 0.72-0.83) and IMMACULATE (AUC 0.72, 0.61-0.84). ConclusionsOur integrative analysis revealed competing signatures beyond established biomarkers of post-MI HFH, comprised of plasma proteins correlated with impaired diastolic function after the primary MI episode.