Biochemical Profiling of Urine Metabolome in Premature Infants Based on LC-MS Considering Maternal Influence

Abstract

In this study, Liquid Chromatography-Mass Spectrometry (LC-MS)-based metabolomics profiling was conducted to elucidate the urinary profiles of premature infants during early and late postnatal stages. As a result, we discovered significant excretion of maternal drugs in early-stage infants and identified crucial metabolites like hormones and amino acids. These findings shed light on the maternal impact on neonatal metabolism and underscore the beneficial effects of breastfeeding on the metabolism of essential amino acids in infants. This research not only enhances our understanding of maternal-infant nutritional interactions and their long-term implications for preterm infants but also offers critical insights into the biochemical characteristics and physiological mechanisms of preterm infants, laying a groundwork for future clinical studies focused on neonatal development and health.

Keywords: LC−MS; amino acid; human milk; maternal nutrition; metabolomics; neurotransmitter; premature infant.

Figure 1

(a) PCA plot; (b) PLS−DA plot for Early group (red; n = 22) and Late group (green; n = 12); (c) Cross−validation of PLS−DA; (d) The top 30 VIP scores. The marker of (*) in (c) represents the highest value in the performance measure.

Figure 2

(a) Heatmap showing the quantitative values of each compound; (b) Volcano plot displaying DEMs (fold change (FC) = Early (n = 22)/late (n = 12), p−value < 0.05, |log2(FC)| > 1).

Figure 3

ROC curves with area under the ROC curve (AUC) and confidence interval (CI) values on selected DEMs using the least absolute shrinkage and selection operator (LASSO) feature selection algorithm. Boxplots of relative concentrations for selected DEMs between Early (red) and Late (green) groups. The black dots represent the concentrations of the selected feature from all samples. Horizontal red lines on the boxplot indicate the optimal cutoff. Yellow diamonds represent the mean concentration of each group.

Figure 4

Identification and prediction of key markers between Early group (n = 22) and Late group (n = 12) using multivariate ROC curve-based exploratory analysis. (a) Overview of all ROC curves from six distinct predictive models, highlighting their respective AUC values and CI; (b) A chart depicting the predictive performance of each of the six models, with the highest accuracy indicated by a red dot for 50−feature panel of model 5; (c) The ROC curve specific to the chosen model 2; (d) A list of the top 10 significant metabolites, ranked by their average importance of being selected during cross−validation.

Figure 5

Box–whisker plots of three upregulated metabolites in HM group (n = 14) compared to the FM (n = 20). Boxplots of relative concentrations for selected DEMs between FM (red) and HM (green) groups. Black dots denote the concentration levels of the chosen feature across all samples. Horizontal red lines mark the optimal cutoff, while the average concentration for each group is symbolized by yellow diamonds.