The clinical utility of brain natriuretic peptide in paediatric left ventricular failure

Aims To assess the rote of brain natriuretic peptide (BNP) in both the acute and chronic settings in children with left ventricular (LV) failure. Methods and results We undertook a retrospective review of all BNP levels taken over a 2-year period in our institution. Minimum follow-up was 90 days. Ninety-two BNP samples from 48 patients were reviewed. Twenty patients (42%) reached the combined endpoint of death, transplantation, or listing for transplantation. Median age was 3 years and 3 months. Mean BNP levels in NYHA or Ross classes I-IV were 29, 239, 744, and 1593 pg/mL, respectively, with significant differences between mean logBNP in classes I-III (P < 0.001). LogBNP levels correlated with fractional shortening (P < 0.001), LVEDd z-score (P < 0.001), and tissue Doppler velocities (P < 0.02). From serial data there was a strong correlation between change in BNP and change in clinical status (F 9.5, P < 0.001). Receiver-operator curve (ROC) demonstrated that BNP > 290 pg/mL predicts poor outcome with sensitivity of 0.80, specificity of 0.87, and likelihood ratio of 6.4 in paediatric patients with chronic LV dysfunction. A separate ROC from acute presentations did not demonstrate superiority of BNP over other assessments. Conclusion BNP levels in paediatric heart failure (HF) patients show a strong correlation to both impaired heart function on echocardiogram and clinical status. Serial BNP levels follow the clinical course. In chronic HF, a BNP level of >290 pg/mL is predictive of an adverse outcome.