Metabolomics-based predictive biomarkers of oral cancer and its severity in human patients from North India using saliva

Abstract

Oral squamous cell carcinoma (OSCC) is frequently the outcome of oral submucous fibrosis (OSMF), a common possibly premalignant disease. In our study, a cohort of 50 patients with OSCC and OSMF, along with 50 healthy controls, was analyzed to identify significant metabolic differences between the patient and control groups through multivariate statistical analysis using NMR-based metabolomics in saliva samples. The 2D scatter plot of PC1 versus PC2 scores clearly show a distinction between the groups, with the principal component analysis (PCA) explaining 24.6% of the variance. Partial least-squares discriminant analysis (PLS-DA) demonstrated R² and Q² values of 0.94 and 0.90, respectively, indicating a robust model fit. A total of 20 distinct metabolites were identified, including 5 that were up-regulated and 5 that were down-regulated. Univariate ROC curve analysis identified nine salivary metabolites with AUC values exceeding 0.70, including acetone, tryptophan, 5-aminopentanoic acid, betaine, aspartic acid, ethanol, acetoacetate, adipic acid, and citrate. Notably, the distinct presence of three metabolites—acetone, tryptophan, and 5-aminopentanoic acid—yielded AUC values of 0.98123, 0.95358, and 0.91506, respectively. The refined statistical model was subjected to metabolic pathway analysis, revealing interconnected pathways. We were also able to predict the severity of the disease, specifically distinguishing between stage I and stage II OSCC. This differentiation was highlighted by the PCA score plot, which explained 28.6% of the variance. These results were further confirmed by PLS-DA. These insights pave the way for early diagnosis and predicting severity in patients with oral cancer, which will enable better management of the disease.