To assess the risk of clinical worsening in heart failure (HF) is an important goal because HF is a progressive disease with a severe prognosis. The risk is usually estimated by using clinical, biological and clinical variables and by the dosage of natriuretic peptides. Other biomarkers have been tested in this setting and among candidates, the soluble ST2 seems a relevant one. The objective of our study was firstly, to analyze the predictive value of blood levels of sST2 in a cohort of patients with chronic heart failure and secondly, to study the variables associated with sST2 levels and to compare its predictive value to that of BNP levels.

Measurement of both BNP and sST2 was performed in patients who were followed in our department. Baseline clinical and biological characteristics were also obtained. The primary endpoint was a composite of unplanned HF-related hospitalizations and all-cause death or cardiac transplantation. Predictive values of sST2 and BNP levels were estimated by using survival curves and Cox models.

Our cohort included 298 patients: age 61y (IQR 52-69), 27% female, 46% ischemic heart disease, 22% diabetes, 25% atrial fibrillation, LVEF 0.38 (IQR 0.28-0.48), eGFR 75 (IQR 55-92)ml/min/1.73m². Median sST2 levels were 31ng/L (24-46) and BNP levels were 192pg/mL (IQR 87-420). Increase in sST2 levels was significantly associated with age, BNP and renal function. During a median follow-up of 36 months, there were 76 events (25.5%) including 32 deaths. The increase in sST2 levels was strongly associated with clinical events. The Figure 1 shows survival curves by using the usual sST2 cut-off of 35ng/l; after adjustment on BNP levels, the Hazard Ratio was 2.85 [1.77-4.58].

In patients with chronic HF, sST2 levels are significantly predictive of outcome on top of BNP. The clinical usefulness of systematic measurement sST2 in addition to natriuretic peptides deserves dedicated studies.