BNP levels predict outcome in pediatric heart failure patients: post hoc analysis of the Pediatric Carvedilol Trial.

BACKGROUND: The ability of serum B-type natriuretic peptide levels (BNP) to predict outcomes in children with heart failure (HF) has not been well demonstrated. This study was designed to determine whether BNP levels predict outcomes in patients with moderate symptomatic HF.

METHODS AND RESULTS: We investigated whether enrollment BNP levels for the Pediatric Carvedilol Trial were associated with baseline characteristics. Freedom from a composite end point of HF hospitalization, death, or transplantation at 9 months was compared using a threshold BNP level identified using receiver operating curve analysis. Median BNP level was 110 pg/mL (interquartile range, 22.4 to 342.0 pg/mL) in 138 subjects. Median age was 3.4 years (interquartile range, 1.1 to 11.0 years). Diagnoses were cardiomyopathy (60%) and congenital heart disease (40%); 73% had a systemic left ventricle. BNP levels correlated moderately with left ventricular ejection fraction (R=0.39, P<0.001) but did not differ by HF class, age, diagnosis, sex, ventricular morphology, or left ventricular end-diastolic dimension Z-score (R=0.19). Outcome events included 25 HF hospitalizations, 4 deaths, and 2 transplants. Sensitivity was 71% and specificity 63%, for a BNP cutoff value of 140 pg/mL. BNP ≥140 pg/mL (hazard ratio, 3.7; 95% confidence interval, 1.62 to 8.4; P=0.002) and age >2 years (hazard ratio, 4.45; 95% confidence interval, 1.68 to 12.04; P=0.003) were independently associated with worse outcomes.

CONCLUSIONS: In children with moderately symptomatic HF, BNP ≥140 pg/mL and age >2 years identified subjects at higher risk for worse outcome. Further validation is needed to determine the BNP levels necessary to stratify risk in other pediatric cohorts.