ObjectiveTo investigate the postnatal changes in urinary metabolic amino acid levels in infants with retinopathy of prematurity (ROP) and their effect on ROP, and to analyze the amino acid metabolic pathways that may be involved in the development of ROP. MethodsA retrospective cohort study. From January 2020 to December 2023, 65 premature infants with severe ROP (ROP group) who were hospitalized, born with gestational age ＜32 weeks in Children's Hospital Affiliated to Zhengzhou University were included in the study. Fifty premature infants with matched sex and gestational age and no ROP were selected as the control group. Urine amino acids and their derivatives were detected by gas chromatography-mass spectrometry. The two groups were compared by independent sample t test. The metabonomics of urinary amino acids was analyzed by orthogonal partial least squares discriminant analysis (OPLS-DA) model. The variable projection importance (VIP) score ＞1 suggested that the substance was two groups of differentially expressed amino acids. The predictive value of urinary amino acids for severe ROP was compared by using the receiver's operating characteristic (ROC) curve and the area under the curve. After t test and metabolomics analysis, the two groups of amino acids with large differences were normalized and compared by Pearson correlation analysis. The Kyoto Encyclopedia of Genes and Genomes database was used to analyze the metabolic pathways of differentially expressed amino acids involved in ROP. ResultsCompared with the control group, the concentrations of oxalic acid -2 and thiodiacetic acid-2 in urine metabolites of children in ROP group were significantly decreased, while the concentrations of 4-hydroxybutyric acid-2, 3-methylpentadienoic acid-2(1), 2-ketoglutarate-ox-2(2) and 3, 6-epoxy-dodecanedioic acid-2 were significantly increased, with statistical differences (t=0.036, 0.005, 0.038, 0.032, 0.022, 0.011; P＜0.05). The results of OPLS-DA analysis showed that amino acids of urinary metabolites in ROP group and control group were distributed in the left and right regions of the scatter plot, and there was a satisfactory separation trend between the two groups (R2Ycum=0.057 4, Q2cum=0.025 7, P＜0.05). As shown in the S-Plot, the amino acids biased towards two stages are glycolic acid-2, phosphoric acid-3, oxalic acid-2, thiodiacetic acid-2, 4-hydroxybutyric acid-2, 3-methylcrotonylglycine-1, 3-methylpentadienoic acid-2(1), 2-ketoglutarate-ox-2(2) and 3, 6-epoxy- dodecanedioic acid-2, respectively. Eleven differentially expressed amino acids with VIP score ＞1 were screened, among which the highest VIP score was oxalate-2, glycerate-3, phosphoric acid-3, 3-methylcrotonylglycine-1, uranoic acid -3 and thiodiacetic acid-2. The difference of amino acid concentration between the two groups was the highest in 4-hydroxybutyric acid-2 and thiodiacetic acid-2. The correlation between oxalic acid-2 and glycerate-3 was the highest (r=0.830, P＜0.001), and most amino acids were positive correlated. ROC curve fitting analysis showed that the combined prediction of 11 differenly-expressed amino groups had the largest area under the curve (0.816), the cutoff value was 0.531, and the sensitivity and specificity were 83.1% and 70.0%, respectively. The enrichment analysis of these 11 amino acids with significant differences suggested that the main pathways involved included butyrate metabolism, glyoxylic acid and dicarboxylic acid metabolism and lipoic acid metabolism. ConclusionAbnormal amino acid metabolism of 4-hydroxybutyrate-2, 3-methylpentadienoic acid-2(1), thiodiacetic acid-2, 2-ketoglutarate-ox-2(2), 3, 6-epoxy-dodecanedioic acid-2 may have a certain effect on the occurrence of ROP.