0192: A proteomic score improves risk stratification in stable chronic heart failure patients

Abstract

Risk stratification of patients with stable chronic heart failure (CHF) is critical to better identify those who may benefit the most from invasive strategies such as heart transplantation. To improve cardiovascular (CV) death prediction in CHF, we performed a proteomic analysis using high throughput surface enhanced laser desorption ionization – time of fight – mass spectrometry (SELDI-TOF-MS). Plasma samples were pre-treated to access the deep proteome. The proteomic analysis was first performed in a case (CV death within 3 years) /control (survivors at 3 years) study including 198 patients with a left ventricular ejection fraction (LVEF) <45%. A proteomic score was developed in this derivation population using the support vector machine (SVM) method. The score was then validated in an independent cohort of 309 consecutive patients (CV death at 3 years) with CHF. Altogether, 203 ion m/z peaks were detected. Among them, 42 peaks were significantly differentially expressed between cases and controls after Bonferroni correction (P value at 0.00025). Then, the SVM method was applied to develop a proteomic score. In the derivation population, the score level was higher in cases as compared to controls: 0.7 vs. 0.25 (P=5.10 -29 ). The ROC curve showed an AUC of 0.87 to predict CV mortality. In the validation population, the score level was still higher in patients who experienced a CV death as compared to survivors: 0.53 vs. 0.39 (P=0.0002). The ROC curve showed an AUC of 0.68. After adjustment on confounders (NYHA class, LVEF, BNP, creatinine, Peak VO 2 ), the score was still significantly associated with CV death (HR=15.1, P=0.007) and it allowed a significant improvement of CHF patient reclassification. The net reclassification index (NRI) and the integrated discrimination improvement (IDI) reach both significant p values. Proteomic analysis of low abundance plasma proteins is highly promising to improve CV death risk prediction in CHF.