Untargeted salivary metabolomics for early prediction of chemotherapy-induced oral toxicities in patients with oral squamous cell carcinoma.

Chemotherapy-induced oral toxicities, including oral mucositis, xerostomia, and dysgeusia, are common complications in patients with oral squamous cell carcinoma (OSCC), often leading to impaired quality of life, treatment interruptions, and poor therapeutic outcomes. Reliable non-invasive biomarkers capable of predicting these toxicities before severe clinical manifestations remain limited. This study investigated salivary metabolomic alterations associated with chemotherapy-induced oral toxicities using untargeted liquid chromatography-quadrupole time-of-flight mass spectrometry (LC-QTOF-MS). A prospective observational study was conducted among 60 patients with histopathologically confirmed OSCC receiving platinum-based chemotherapy. Unstimulated saliva samples were collected before and after chemotherapy and analyzed using untargeted LC-QTOF-MS. Metabolomic data were processed using Agilent MassHunter software and annotated against HMDB, METLIN, LipidMaps, and KEGG databases. Principal component analysis (PCA), partial least squares-discriminant analysis (PLS-DA), variable importance in projection (VIP) analysis, hierarchical clustering, volcano plot analysis, KEGG pathway enrichment, multivariable logistic regression, and receiver operating characteristic (ROC) curve analysis were performed to identify biomarkers associated with chemotherapy-induced oral toxicities. Oral mucositis occurred in 63.3% of patients, xerostomia in 55.0%, dysgeusia in 48.3%, and severe mucositis in 18.3%. Untargeted metabolomic profiling identified significant perturbations in lipid metabolism, oxidative stress, and amino acid metabolism. Ceramide, lysophosphatidylcholine (LPC), glutathione disulfide, arachidonic acid, taurine, and citrulline were among the most significantly dysregulated metabolites. PCA and PLS-DA demonstrated distinct metabolic separation between patients with and without oral toxicity (R² = 0.86; Q² = 0.74; permutation p < 0.001). KEGG pathway analysis revealed significant enrichment of glycerophospholipid, sphingolipid, glutathione, arachidonic acid, arginine, and taurine metabolic pathways. Ceramide (adjusted OR = 4.82, 95% CI: 2.11-10.96), LPC (adjusted OR = 3.64, 95% CI: 1.78-7.52), and glutathione (adjusted OR = 2.92, 95% CI: 1.39-6.12) were independent predictors of severe oral toxicity. ROC analysis demonstrated strong discriminatory performance for ceramide (AUC = 0.91), LPC (AUC = 0.88), and glutathione (AUC = 0.85), while the combined metabolomic model achieved an AUC of 0.95, with 90.9% sensitivity and 91.2% specificity. Untargeted salivary LC-QTOF-MS metabolomics identified distinct metabolic signatures associated with chemotherapy-induced oral toxicities in patients with OSCC. Dysregulation of lipid metabolism, oxidative stress pathways, and amino acid metabolism appears to play a central role in the development of oral mucositis, xerostomia, and dysgeusia. Ceramide, lysophosphatidylcholine, and glutathione emerged as promising non-invasive biomarkers with strong discriminatory potential for identifying patients at increased risk of severe oral toxicity. These findings support the potential application of salivary metabolomics for early risk stratification, treatment monitoring, and personalized supportive care; however, external validation in larger, independent multicentre cohorts is required before clinical implementation.