576 - "Unveiling Health through Urine: Metabolomic Insights into Predicting Bronchopulmonary Dysplasia"

Poster Number: 576
Publication Number: 576.1334

Background: Despite advancements in neonatal care, bronchopulmonary dysplasia (BPD) remains a significant cause of morbidity and mortality in preterm infants, with little improvement in its incidence. Early identification of high-risk infants holds promise for informing clinical decisions and facilitating the selection of candidates for future trials aimed at improving outcomes. The integration of metabolomics into the early prognosis of BPD represents an emerging and potentially transformative field.

Objective: This study aimed to identify urinary metabolites that could serve as early predictors of BPD.

Design/Methods: A prospective cohort study was conducted between October 2019 and December 2021, enrolling very low birthweight (VLBW) neonates with gestational ages of < 30 weeks and birthweights ≤1500 grams in a level four neonatal intensive care unit. Urine samples were collected on the fifth day of life and subjected to untargeted metabolomic analysis using liquid chromatography-mass spectrometry. The primary outcome was BPD, defined according to NICHD criteria as the need for supplemental oxygen at 28 days. Metabolite data were log-transformed, and significance was determined with a Wilcoxon rank-sum p-value < 0.05. The dataset was divided into training (66.7%) and testing (33.3%) sets, and neural network machine learning models were developed, with the area under the receiver operating characteristic curve (AUC) used as the performance metric. All statistical analyses were conducted in R.

Results: The study enrolled 47 neonates, with 20 (42.6%) diagnosed with BPD. Neonates who developed BPD were younger (27.1 weeks vs. 29.3 weeks, p< 0.01) and tended to have lower birthweights (920 grams vs. 1,120 grams, p=0.05). From an initial pool of 295 positive metabolites, 76 metabolites reached significance. Using a final set of ten positively charged metabolites, the AUC for predicting BPD was 90.7% (95% CI, 72.2%-100%), with a sensitivity of 100% and specificity of 88.9%. These markers were associated with key biological processes such as lipid and amino acid metabolism, arginine metabolism, and oxidation/reduction reactions. Among 161 negative metabolites, 46 had a p-value < 0.05, and the top 10 negative metabolites achieved an AUC of 77.8% (95% CI, 54.9%-100%). Pathways associated with BPD included mitochondrial dysfunction and gluconeogenesis.

Conclusion(s): Urinary metabolites represent a readily accessible resource for predicting neonates at risk of developing BPD. However, our preliminary findings require validation in an external cohort to further establish their clinical utility and potential impact.