456 - Urinary Metabolomics of Bronchopulmonary Dysplasia in Extremely Preterm Infants

Poster Number: 456
Publication Number: 456.431

Dara G. Torgerson, University of California, San Francisco, School of Medicine, San Francisco, CA, United States; Miguel A. Guardado, University of California, San Francisco, San Francisco, CA, United States; Cheryl Chapin, UCSF Benioff Children's Hospital San Francisco, San Francisco, CA, United States; Martina Steurer, UCSF, San Francisco, CA, United States; Philip L. Ballard, University of California, San Francisco, School of Medicine, Bodega Bay, CA, United States

Background: Extremely preterm infants are at high risk of developing bronchopulmonary dysplasia (BPD), which is a serious developmental lung disease requiring ventilatory support and/or supplemental oxygen at 36 weeks post menstrual age (PMA). Characterizing early changes in the urinary metabolome may increase our understanding of the development of BPD.

Objective: Identify early changes in the urinary metabolome of extremely preterm infants diagnosed with BPD at 36 weeks PMA.

Design/Methods: Our study included 314 preterm infants < 29 weeks gestational age from the Trial of Late Surfactant Study (TOLSURF) and the Prematurity and Respiratory Outcomes Program (PROP), with a combined total of 178 BPD cases and 136 BPD controls. We performed untargeted metabolomics (UHPCL/MS) on urines collected at two timepoints for each infant between days 6-14 and 23-30 postnatal age, with normalization to osmolality and imputation of missing values to the minimum detectable level. Fold-change in the level of individual metabolites over time and at each timepoint separately were tested for an association with BPD using logistic regression, adjusting for sex, birthweight, TPN status, corticosteroid usage, and maternal race/ethnicity. Analyses were performed within each study and combined using a weighted meta-analysis. Pathway enrichment was evaluated using a Fisher’s exact test, and changes in the global metabolome using a principal component analysis.

Results: We identified 961 unique urinary biochemicals belonging to 83 metabolic pathways. A total of 38 biochemicals had urinary levels that varied by BPD status at timepoint 1 at p< 0.05, which were enriched for the sub-pathway leucine, isoleucine and valine metabolism (p=4.6x10-4, FDR=3.8%). Over all comparisons, 13 biochemicals met the criteria of having p< 0.05 in the meta-anlaysis, and p< 0.15 within each study and effects in the same direction (Table). Among these, levels of 10 biochemicals were lower in infants with BPD, including three involved in adrenal cortex steroid biosynthesis and two carnitine containing compounds at timepoint 2. Principal component analysis over all metabolites showed no clear distinction between BPD cases and controls.Conclusion(s): Among high risk, extremely preterm infants we identified candidate pathways and urinary biochemical changes associated with BPD that warrant further study. High variability of the urinary metabolome of preterm infants may reflect the degree of clinical heterogeneity in BPD and its contributing factors.
Table: Candidate urinary biochemicals associated with BPD at p < 0.05 in a meta-analysis across studies and having p < 0.15 within each study individually with effects in the same direction. A negative effect size/unit change indicates metabolites whose abundance or fold-change (Time 2/Time 1) was lower in infants with BPD.