Abstract
AIMS: To investigate the associations between oral-rinse microbiota and distinct oral conditions, and further evaluate its potential ability to distinguish periodontitis severity. METHODS: Oral-rinse-sourced microbiota with 16S ribosomal RNA sequencing from 3770 adults in US National Health and Nutrition Examination Survey 2009-2012 were analysed across oral health, caries, periodontitis, co-existing caries and periodontitis and edentulism. Diagnostic potential of the oral-rinse microbiota for periodontitis severity was evaluated using multi-class random forest (RF) model with internal validation and external validation in an independent cohort (n = 392). RESULTS: Oral condition accounted for substantial variance in oral-rinse microbiota, revealing disease or tooth loss-associated shifts. Increasing acidogenic/aciduric taxa (Veillonella, Lactobacillus, Atopobium) or periodontitis-associated taxa (Filifactor, Treponema, Tannerella) were identified in caries-only or periodontitis-only groups, respectively, while the co-existing disease group showed overlapping shifts. Taxa shifted dose-dependently with increasing periodontitis severity. The RF model achieved moderate performance in identifying severe periodontitis, with the area under the receiver operating characteristic curve (AUROC) of 0.81 (0.75-0.87) internally and 0.83 (0.77-0.88) externally. Key contributing taxa aligned with established periodontitis-associated genera, supporting model interpretability. CONCLUSION: Based on our results, oral-rinse microbiota captures disease-specific signatures across oral conditions, supporting its potential as a non-invasive tool to monitor oral microbial ecology and assess periodontitis severity at the population level.