Correlation analysis of blood metabolites and periventricular leukomalacia in preterm infants two weeks postnatally

Periventricular leukomalacia (PVL) is a neonatal brain injury disease that may cause neurodevelopmental disorders. In this study, our objective was to compare the differences in blood metabolites between preterm infants with and without PVL. We also aimed to screen for potential metabolic markers related to the early identification of PVL. Preterm infants diagnosed with periventricular leukomalacia (PVL) by cranial imaging were assigned to the PVL group (n = 32), while those without a PVL diagnosis were assigned to the control group (n = 68). Blood metabolite levels were measured at 2 weeks postnatally using ultra-performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS). Most blood metabolite concentrations were statistically different between the two groups (P < 0.05). The results of the OPLS-DA model identification analysis and validation show that the model is valid. Of the 12 blood metabolites with predicted variable importance (VIP) > 1, those with the highest VIP scores included octadecadienoylcarnitine (C18:2), 3-hydroxy-isovalerylcarnitine (C4DC), 3-oH-octadecylcarnitine (C18OH), 3-hydroxy-cetacylcarnitine (C16OH), Adipylcarnitine (C6DC), and Tyrosine (Tyr). The S-plot shows that the blood metabolites C4DC, Ornithine (Orn), C16OH, and C18:2 are biased towards the two poles. The univariate analysis for cross-validation showed that C18:2, C4DC, C18OH, and C16OH concentrations in blood metabolites were significantly different between the two groups (P < 0.001). The results of multivariate logistic regression analysis showed that a high concentration of Tyr was negatively associated with the occurrence of PVL (P = 0.031), while a longer duration of parenteral nutrition (PN) and a higher Orn concentration were positively correlated with PVL (all P < 0.05). Receiver Operating Characteristic (ROC) curve analysis indicated that C18:2 had a relatively high discriminatory ability for PVL among the detected metabolites, but it did not show independent statistical significance in the multivariate logistic regression analysis. The most significant difference in the blood metabolite concentrations between the two groups was observed for C18:2. Enrichment analysis showed that fatty acid degradation is the most likely major mechanistic pathway involved in the development of PVL. There is a certain correlation between early blood metabolite levels and PVL occurrence in premature infants. Although C18:2 shows a certain discriminatory ability, the value of it as an early detection indicator still needs further research.