Heart failure (HF) remains a main cause of mortality worldwide. Risk stratification of patients with systolic chronic HF is critical to identify those who may benefit from advanced HF therapies. The aim of this study is to identify plasmatic proteins that could predict the early death (within 3 years) of HF patients with reduced ejection fraction hospitalized in CHRU de Lille. The subproteome targeted by an aptamer-based technology, the Slow Off-rate Modified Aptamer (SOMA) scan assay of 1310 proteins, was profiled in blood samples from 168 HF patients, and 203 proteins were significantly modulated between patients who died of cardiovascular death and patients who were alive after 3 years of HF evaluation (Wilcoxon test, FDR 5%). A molecular network was built using these 203 proteins, and the resulting network contained 2281 molecules assigned to 34 clusters annotated to biological pathways by Gene Ontology. This network model highlighted extracellular matrix organization as the main mechanism involved in early death in HF patients. In parallel, an adaptive Least Absolute Shrinkage and Selection Operator (LASSO) was performed on these 203 proteins, and six proteins were selected as candidates to predict early death in HF patients: complement C3, cathepsin S and F107B were decreased and MAPK5, MMP1 and MMP7 increased in patients who died of cardiovascular causes compared with patients living 3 years after HF evaluation. This proteomic signature of 6 circulating plasma proteins allows the identification of systolic HF patients with a risk of early death.