Abstract
BACKGROUND: Oral squamous cell carcinoma (OSCC) often recurs locally, reducing survival. The oral microbiome may influence tumor recurrence, but its prognostic role is unclear. This study investigated oral microbiomes associated with OSCC recurrence and their prognostic merit. MATERIALS AND METHODS: Saliva samples were collected from 133 patients with OSCC. 16S rRNA gene sequencing was performed, and microbial signatures were predicted via XGBoost. Functional metagenomic prediction was conducted using PICRUSt2. RESULTS: XGBoost identified Eubacterium, Lactobacillus, Kingella, Paludibacter, Parvimonas, Staphylococcus, and Veillonella as predictive for OSCC recurrence. Eubacterium and Lactobacillus were significantly enriched in recurrent disease and associated with poor survival. Staphylococcus and Veillonella were abundant in non-recurrent disease, correlating with a favorable prognosis. The microbiome-based model achieved superior predictive performance (AUC = 0.741) compared with the clinical N-stage model (AUC = 0.66). Eubacterium and Lactobacillus showed positive correlations with key genes, such as protein kinase B (AKT), fibroblast growth factor receptor 1 and guanine nucleotide-binding protein G subunit beta-2, within the phosphoinositide 3-kinase (PI3K)/AKT/mammalian target of rapamycin (mTOR) pathway. In contrast, Veillonella was negatively correlated with these genes. CONCLUSIONS: Oral saliva microbiome profiling reveals distinct microbial patterns associated with OSCC recurrence. Our correlation-based functional predictions indicated that the enrichment of Eubacterium and Lactobacillus along with a lower abundance of Veillonella may influence recurrence through oncogenic PI3K/AKT/mTOR, underscoring the prognostic potential of saliva-based microbial biomarkers.