Urine metabolomics in neonates with late-onset sepsis in a case-control study

Abstract

Although late-onset sepsis (LOS) is a major cause of neonatal morbidity and mortality, biomarkers evaluated in LOS lack high diagnostic accuracy. In this prospective, case-control, pilot study, we aimed to determine the metabolic profile of neonates with LOS. Urine samples were collected at the day of initial LOS evaluation, the 3^rd and 10^th day, thereafter, from 16 septic neonates (9 confirmed and 7 possible LOS cases) and 16 non-septic ones (controls) at respective time points. Urine metabolic profiles were assessed using non-targeted nuclear magnetic resonance spectroscopy and targeted liquid chromatography-tandem mass spectrometry analysis. Multivariate statistical models with data from either analytical approach showed clear separation between the metabolic profiles of septic neonates (both possible and confirmed) and the controls. Metabolic changes appeared to be related to disease progression. Overall, neonates with confirmed or possible LOS exhibited comparable metabolic profiles indicating similar metabolic alternations upon the onset of clinical manifestations. This methodology therefore enabled the discrimination of neonates with LOS from non-septic individuals, providing potential for further research toward the discovery of LOS-related biomarkers.

Figure 1. PCA scores plots.

First and second component explaining the variation among data are plotted at three time-points D0, D3 and D10. Top Frame: ¹H-NMR data: Samples of DO show clear separation of the two groups (septic patients versus controls). On D3 separation is not as evident, while on D10 samples cluster all together Bottom frame: LC-MS/MS data: Samples of DO and D3 show clear separation of the two groups (septic patients versus controls). On D10 samples cluster all together. The arrow indicates a deceased septic neonate on day 3. Green circles correspond to controls and red circles to sepsis subjects (smaller circles denote possible and large circles confirmed sepsis). Pink correspond to day 3 and light pink to day 10.

Figure 2. Higher impact pathways involving significant metabolites of both techniques as resulted from the online web software MetaboAnalyst.

The significance of a model increases in the y-axis. The color bar indicates the p-value of the pathway; darker color corresponds to more significant p-value (smaller value).