Ecological shifts and microbiome reorganization across oral precancer-cancer spectrum

Introduction: Oral cavity squamous cell carcinoma (OCSCC) and their precursor lesions, oral potentially malignant diseases (OPMD), arise within a complex mucosal microenvironment where the microbiome may contribute to carcinogenesis. In this cross-sectional observational study, we aimed to compare the bacterial communities of normal oral mucosa, OPMD, and OCSCC to determine whether specific compositional shifts or diversity patterns are associated with progressive disease. Methods: Eighty-one patients with OPMD (n=55) or OCSCC (n=26) undergoing clinical evaluation had site-matched swabs from lesions and contralateral clinically normal mucosa. Twelve patients without OPMD or any malignancy served as controls. Amplicon sequence variants (ASVs) generated from 16S rRNA gene amplicon sequencing data were taxonomically assigned to genus level. Alpha diversity (richness, Shannon, inverse Simpson) was compared within patients using paired Wilcoxon tests and across diagnostic groups using nonparametric tests. Beta diversity (Bray-Curtis) was assessed by principal coordinates analysis (PCoA) with PERMANOVA to estimate effect size (R²) and p values. Alluvial plots and Venn diagrams were generated to evaluate overlap between groups and inter and intra group diversity. Results: Cross-sectionally, alpha diversity did not differ significantly among controls, dysplasia lesions, and cancer lesions across all metrics. In contrast, PCoA of all groups showed modest but statistically significant compositional separation by disease status (PERMANOVA F=2.191, R²=0.052, p=0.001); within cancer patients, lesion versus contralateral normal sites showed a trend (F=1.159, R²=0.028, p=0.055), and within dysplasia patients, separation was minimal but nominally significant (F=0.647, R²=0.009, p=0.025). Overall, 49% of genera were shared among cancer lesions, dysplasia lesions, and controls, while 20% of genera were unique to dysplasia lesions and 10% were unique to cancer lesions. Lastly, both dysplasia and cancer lesions demonstrated higher intragroup Bay-Curtis distances compared with controls (FDR-adj. p<0.001), consistent with a progressive ecological diversification in malignant transformation. Conclusion: In this comparatively large, site-matched cross-sectional cohort spanning normal mucosa, OPMD, and OCSCC, oral carcinogenesis was characterized by shifts in beta diversity. Key findings include significantly increased evenness in dysplasia lesions relative to paired mucosa, small yet significant compositional differences by disease status (R² ≈ 5%), and an increase in unique ASVs within dyplasia and cancer lesions. These results support a model in which oral carcinogenesis is associated with subtle ecological reorganization rather than gross disruption of the microbiome and highlight candidate bacterial taxa for mechanistic and biomarker studies.