NT-proBNP and Zlog-transformed NT-proBNP values predict extubation failure in critically ill neonates with pulmonary hypertension and ventricular dysfunction

Objectives: Critically ill neonates with a history of pulmonary hypertension (PH) or ventricular dysfunction are at risk to experience an extubation failure (EF) after liberation from mechanical ventilation (MV). Due to insufficient data from neonatal cohorts, it remains unclear whether NT-proBNP is an appropriate biomarker to predict EF in this cohort. The Zlog-transformation of NT-proBNP (further named NT-proBNP_Zlog) is an additional tool to optimize the interpretation of NT-proBNP since absolute NT-proBNP values are varying with the age of these infants. Patients and Methods: This was a retrospective single-center analysis at the University Children's Hospital, Bonn, Germany, during the study period from January 2020 until December 2021. Forty-three neonates met the inclusion criteria and were screened for study participation. Inclusion criteria: prolonged (>24 h) MV with at least one extubation attempt, with a history of PH and/or ventricular dysfunction in the echocardiographic assessment at admission to the neonatal intensive care unit or during the period of MV, NT-proBNP measurements before (max. 24 h, baseline) and after (max. 24 h, follow-up) the first extubation attempt. The primary clinical endpoint was defined as EF with need for reintubation (0–72 h). Neonates with an EF were allocated to group A and neonates with successful liberation from MV to group B. Main Results: The primary clinical endpoint (EF) was reached in 21% (nine infants). Absolute mean NT-proBNP values (NT-proBNP_abs) at baseline did not differ significantly in infants of group A and B (6931 vs. 7136 pg/ml, p = 0.227). NT-proBNP_Zlog values at baseline (2.35 vs. 1.57, p = 0.073) tended to higher values in group A. NT-proBNP_abs values measured at follow-up were significantly higher in infants allocated to group A (11120 vs. 7570 pg/ml, p = 0.027). Likewise, NT-proBNP_Zlog values at follow-up were significantly higher in infants allocated to group A (3.05 vs. 1.93, p = 0.009). NT-proBNP_abs values at follow-up and NT-proBNP_Zlog values at baseline correlated significantly with the severity of PH. Regarding the receiver operating characteristic-analysis, a NT-proBNP_abs value at follow-up of ≥4622 pg/ml was calculated as optimal cut-off value for the prediction of EF (area under the curve [AUC] 0.742, p = 0.001). A NT-proBNP_Zlog value at baseline of ≥1.63 and at follow-up of ≥2.14 was calculated as optimal cut-off for the prediction of EF (AUC: 0.690/p = 0.027, and 0.781/p = 0.000, respectively). Conclusion: NT-proBNP_abs and NT-proBNP_Zlog might be valuable biomarkers for the prediction of EF in critically ill neonates. The Zlog-transformation of NT-proBNP allows an age-independent interpretation of NT-proBNP and should be considered for clinical routine.