No abstract

No abstract

BackgroundClinicians lack objective tests to help determine the severity of bronchiolitis or to distinguish a viral from bacterial causes of respiratory distress. We hypothesized that children with respiratory syncytial virus (RSV) infection would have a different metabolomic profile compared to those with bacterial infection or healthy controls, and this might also vary with bronchiolitis severity.MethodsClinical information and urine-based metabolomic data were collected from healthy age-matched children (n = 37) and those admitted to hospital with a proven infection (RSV n = 55; Non-RSV viral n = 16; bacterial n = 24). Nuclear magnetic resonance (NMR) measured 86 metabolites per urine sample. Partial least squares discriminant analysis (PLS-DA) was performed to create models of separation.ResultsUsing a combination of metabolites, a strong PLS-DA model (R2 = 0.86, Q2 = 0.76) was created differentiating healthy children from those with RSV infection. This model had over 90 % accuracy in classifying blinded infants with similar illness severity. Two other models differentiated length of hospitalization and viral versus bacterial infection.ConclusionWhile the sample sizes remain small, this is the first report suggesting that metabolomic analysis of urine samples has the potential to become a diagnostic aid. Future studies with larger sample sizes are required to validate the utility of metabolomics in pediatric patients with respiratory distress.

Establishing the severity of hypoxic insult during the delivery of a neonate is key step in the determining the type of therapy administered. While successful therapy is present, current methods for assessing hypoxic injuries in the neonate are limited. Urine Nuclear Magnetic Resonance (NMR) metabolomics allows for the rapid non-invasive assessment of a multitude breakdown products of physiological processes. In a newborn piglet model of hypoxia, we used NMR spectroscopy to determine the levels of metabolites in urine samples, which were correlated with physiological measurements. Using PLS-DA analysis, we identified 13 urinary metabolites that differentiated hypoxic versus nonhypoxic animals (1-methylnicotinamide, 2-oxoglutarate, alanine, asparagine, betaine, citrate, creatine, fumarate, hippurate, lactate, N-acetylglycine, N-carbamoyl-β-alanine, and valine). Using this metabolomic profile, we then were able to blindly identify hypoxic animals correctly 84% of the time compared to nonhypoxic controls. This was better than using physiologic measures alone. Metabolomic profiling of urine has potential for identifying neonates that have undergone episodes of hypoxia.

No abstract