Dynamic risk stratification using serial measurements of plasma concentrations of natriuretic peptides in patients with heart failure

Background Prognostic models for patients with chronic heart failure are generally based on a single assessment but treatment is often given with the intention of changing risk; re- evaluation of risk is an important aspect of care. The prognostic value of serial measurements of natriuretic peptides for the assessment of changes in risk is uncertain. Aims To evaluate the prognostic value of serial measurements of plasma amino-terminal pro-brain natriuretic peptide (NT-proBNP) during follow-up of out-patients with chronic heart failure (CHF). Methods Patients diagnosed with CHF between 2001 and 2014 at a single out-patient clinic serving a local community were included in this analysis. NT-proBNP was measured at the initial visit and serially during follow-up. Only patients who had one or more measurements of NT-proBNP after baseline, at 4, 12 and/or 24 months were included. Results At baseline, amongst 1998 patients enrolled, the median age was 73 (IQR: 64–79) years, 70% were men, 31% were in NYHA class III/IV, and 77% had NT-proBNP >400 pg/mL. Median follow-up was 4.8 (IQR: 2.5–8.6) years. Serial measurements of NT-proBNP improved prediction of all-cause mortality at 3 years (c- statistic = 0.71) compared with using baseline data only (c-statistic = 0.67; p < 0.001) but a model using only the most recent NT-proBNP had an even higher c-statistic (0.72; p < 0.001). Similar results were obtained based on long-term prediction of mortality using all available follow-up data. Conclusions Serial measurement of NT-proBNP in patients with CHF improves prediction of all-cause mortality. However, using the most recent value of NT-proBNP has similar predictive power as using serial measurements