Paired metagenomic profiling reveals site-specific microbial composition and conserved diversity in the oral cavity of head and neck cancer patients

Background : The extent to which microbial diversity and composition vary across oral subsites within an individual remains poorly defined, limiting the reproducibility of microbiome-based biomarkers and the standardization of sampling strategies. We investigated compositional and diversity differences between paired tonsil and saliva samples to characterize microbial ecology across oral compartments in HNC patients. Methods : Paired tonsil and saliva samples were collected from 25 patients diagnosed with head and neck squamous cell carcinoma (NCT05142033) (median age = 63 years [IQR 53-73]; 76% male, 96% White). The cohort included 68% oral cavity, 20% oropharyngeal, 8% laryngeal, and 4% cutaneous primaries. Shotgun metagenomic sequencing was performed on all samples. Beta diversity was assessed using paired permutational multivariate analysis of variance (PERMANOVA) on Bray-Curtis dissimilarity matrices with restricted permutations to account for individual variation. Species-level differential abundance was tested using paired Wilcoxon signed-rank tests with false discovery rate (FDR) correction. Additionally, alpha diversity metrics were compared between sites. Results : Microbial composition differed significantly between tonsil and saliva samples (PERMANOVA: F=1.00, R²=2.0%, p=0.001). Twenty species showed significant abundance differences after FDR correction (q<0.05). Fifteen species were increased in saliva versus tonsils, including Rothia aeria (FDR=0.004), Actinomyces naeslundii (FDR=0.008), and Streptococcus sanguinis (FDR=0.009). Five species decreased in saliva, including Prevotella scopos , Prevotella jejuni , and Campylobacter concisus . Despite compositional differences, alpha diversity metrics were equivalent between sites (Shannon index: tonsil 3.73±0.59 vs saliva 3.71±0.70, p=0.565; observed species: 3135±1567 vs 3251±1228, p=0.708). Individual-level differences in evenness significantly correlated with compositional dissimilarity (Spearman rho=0.450, p=0.024), indicating that diversity and composition co-vary within individuals. Conclusions : Tonsil and saliva harbor distinct microbial communities yet maintain equivalent species diversity levels. This pattern demonstrates niche conservatism in the oral cavity. The paired study design was essential for detecting these differences, as massive individual variation dominated over site-specific effects. These findings have implications for understanding oral microbiome ecology in head and neck cancer and suggest that sampling site selection is critical for microbiome-based biomarker discovery.