Metabolic profiling of severe early childhood caries and black stain: pathway and biomarker discovery in 3-6-year-olds.

The present study aims to identify the key metabolic compounds of supragingival plaque and potential metabolic pathways associated with severe early childhood caries (S-ECC) in preschool children, and to provide a basis for screening candidate biomarkers that could be beneficial to clinical practice guidelines. A cohort of 93 children, including 32 healthy controls (HC group), 31 children with S-ECC (SECC group) and 30 children with S-ECC and black stain (SECCBS group), were recruited. Supragingival plaques were collected for UHPLC-MS analyses followed by bioinformatics analysis. Differential metabolites (DMs) were screened through Partial Least Squares Discriminatory Analysis (PLS-DA) and enriched in KEGG pathways. Ecological network analysis and WGCNA were performed to construct co-expression networks. Ten machine learning (ML) models were optimized by feature selection and hyperparameter tuning, and evaluated using the area under the ROC curve (AUC), sensitivity, specificity, precision, and predictability comparison. A total of 1,069 metabolites were detected, with 497 annotated into 40 KEGG pathways, primarily amino acid and lipid metabolism. We identified 134 DMs between the HC and SECC groups, and 49 DMs between the SECC and SECCBS groups. WGCNA identified 32 co-expression modules in each paired comparison. In the HC-SECC comparison, the top correlated module pathways (neuroactive ligand-receptor interaction and pyrimidine metabolism) aligned precisely with the KEGG enrichment results of the DMs. In the SECC-SECCBS comparison, biosynthesis of unsaturated fatty acids and fatty acid degradation were uniquely enriched. For clinical prediction, the random forest (RF) model achieved the highest AUC (98.38) in the HC-SECC group, while the SVM-Poly model was superior in the SECC-SECCBS group (AUC = 95.10). The supragingival plaque metabolome is discernibly different among healthy children, those with severe caries, and those with black stain. Key metabolites including uracil, uridine 5'-diphosphate, cytosine, epinephrine, uridine, and deoxycytidylic acid are closely associated with dental caries occurrence and represent promising targets for early diagnosis. Additionally, Glycerophospho-N-palmitoyl ethanolamine (GP-NPEA) and lipid metabolic adaptations (such as unsaturated fatty acid pathways) in the SECCBS group provide a novel mechanistic explanation for the lower caries susceptibility and the pigment formation associated with black stain. These biomarkers offer valuable insights for precision risk stratification and personalized therapeutic strategies in pediatric dentistry.