Salivary Metabolomics Discloses Metabolite Signatures of Oral Leukoplakia with and Without Dysplasia

Abstract

Leukoplakia is a condition marked by white patches on the inner surfaces of the oral cavity. Its potential to progress to oral squamous cell carcinoma underscores the need for effective screening and early diagnosis procedures. We employed NMR-based salivary and tissue metabolomics to identify potential biomarkers for leukoplakia and dysplastic leukoplakia. Univariate and multivariate methods were used to evaluate the NMR-derived metabolite concentrations. The salivary metabolite profile of leukoplakia exhibited specific alterations compared to healthy controls. These metabolic changes were more pronounced in cases of dysplastic lesions. Multivariate ROC curve analysis, based on a selection of salivary metabolites, ascribed high diagnostic accuracy to the models that discriminate between dysplastic and healthy cases. However, NMR analysis of tissue biopsies was ineffective in extracting metabolic signatures to differentiate between lesional, peri-lesional, and healthy tissues. Our pilot study employing a metabolomics-based approach led to the development of salivary models that represent a complementary strategy for clinically detecting leukoplakia. However, larger-scale validation is required to fully evaluate their diagnostic potential and to effectively stratify leukoplakia patients according to dysplasia status.

Keywords: dysplasia; leukoplakia; malignant transformation; salivary diagnostics; salivary metabolomics.

Figure 1

Partial least squares discriminant analysis (PLS-DA) scores plot of salivary metabolite datasets from the NDLK, DLK, and HC groups (A). The coloured ellipses represent the 95% confidence region of each cluster. Variable’s importance in projection (VIP) plot (B) ranks the most discriminative metabolites resulting from PLS-DA component 1. A metabolite with a higher VIP score discriminates more effectively between the groups. The blue and red boxes illustrate the relative metabolite abundance in the specified groups.

Figure 2

Partial least squares discriminant analysis (PLS-DA) scores plot of the salivary metabolite datasets from the NDLK, DLK and HC groups. (A) NDLK vs. HC, (B) DLK vs. HC. The coloured ellipses represent the 95% confidence region of each cluster. PLS-DA score plots are flanked by their corresponding variable’s importance in projection (VIP) plots, (C,D), which rank the most discriminative metabolites resulting from PLS-DA component 1. A metabolite with a higher VIP score discriminates more effectively between the two groups. The blue and red boxes illustrate the relative abundance of metabolites in the specified groups.

Figure 3

Volcano plot analysis of the salivary metabolite datasets from (A) the patients with leukoplakia vs. healthy controls (NDLK/HC) and (B) the patients with dysplastic leukoplakia vs. healthy controls (DLK/HC). For each metabolite compared, the plots combine its fold change (log₂ (FC) on the x-axis) and −log₁₀ p-value from the t-test, on the y-axis. Metabolites that satisfy the p < 0.05 and |FC| > 1.5 are considered discriminant and represented by coloured circles in the upper left and right quadrants.