Urine Metabolomics to Predict Bronchopulmonary Dysplasia: A Single-Center Prospective Study

Bronchopulmonary dysplasia (BPD) remains a major cause of morbidity in extremely preterm infants, yet early prediction remains limited using clinical factors alone. Urine is a non-invasive and routinely available specimen in the NICU, making it well suited for early biomarker discovery.

To determine whether urinary metabolites collected on day of life five can predict BPD development in preterm neonates and to identify associated metabolic pathways.

In this prospective single-center study, urine samples collected on day of life five from preterm infants were analyzed using liquid chromatography-mass spectrometry-based metabolomics. A multivariable modeling approach integrating urinary metabolites with clinical features was used to identify metabolic patterns associated with subsequent BPD development. Model performance was assessed using area under the receiver operating characteristic curve (AUC).

Early urinary metabolomic profiling identified a multivariate metabolic signature associated with later BPD development (40.9% of infants; n = 44). The integrated clinical-metabolic model demonstrated good predictive performance (AUC = 0.805) using a limited number of metabolites and clinical variables. Key metabolic pathways implicated included purine metabolism, oxidative stress, and amino acid degradation, consistent with known mechanisms of lung injury in prematurity.

A single urine sample collected early in life can provide clinically meaningful information about BPD risk. These findings support the feasibility of urine-based, non-invasive metabolic profiling as a complement to existing clinical risk assessment tools. Larger, multicenter studies are needed to validate these findings and define their role in clinical care.