Identification of patient subtypes based on protein expression for prediction of heart failure after myocardial infarction

Summary: This study investigates the ability of high-throughput aptamer-based platform to identify circulating biomarkers able to predict occurrence of heart failure (HF), in blood samples collected during hospitalization of patients suffering from a first myocardial infarction (MI). REVE-1 (derivation) and REVE-2 (validation) cohorts included respectively 254 and 238 patients, followed up respectively 9 · 2 ± 4 · 8 and 7 · 6 ± 3 · 0 years. A blood sample collected during hospitalization was used for quantifying 4,668 proteins. Fifty proteins were significantly associated with long-term occurrence of HF with all-cause death as the competing event. k-means, an unsupervised clustering method, identified two groups of patients based on expression levels of the 50 proteins. Group 2 was significantly associated with a higher risk of HF in both cohorts. These results showed that a subset of 50 selected proteins quantified during hospitalization of MI patients is able to stratify and predict the long-term occurrence of HF.