Identification of Oral Microbiome Biomarkers Associated with Lung Cancer Diagnosis and Radiotherapy Response Prediction

Background The oral cavity serves as the anatomical entrance of the respiratory tract, sharing microbiological and pathophysiological connections with the lower airways. Despite radiotherapy being a cornerstone treatment for lung cancer, the field lacks robust oral microbiome biomarkers capable of predicting therapeutic outcomes. Methods In this study, we analyzed the oral microbiome of 136 lung cancer patients and 199 healthy controls from a discovery cohort and two validation cohorts using 16S rRNA sequencing. Differential microbial taxa were subsequently identified through an integrative analysis combining Wilcoxon rank-sum tests, LEfSe, and ANCOM-BC2. Random forest analysis was employed to construct models for lung cancer diagnosis and treatment response prediction. The prognostic value of discriminatory features was then assessed using Kaplan-Meier analysis. Results Healthy controls exhibited significantly higher Streptococcus abundance than patients. Microbial community structure shifted substantially during treatment. Responders showed enrichment of Rothia aeria and Prevotella salivae , associated with prolonged OS and PFS, whereas non-responders exhibited elevated Porphyromonas endodontalis correlating with shorter OS and PFS. ANCOM-BC2 analysis indicated that Akkermansia and Alistipes were virtually absent in non-responders, whereas Desulfovibrio and Moraxella were near-complete absence in responders. In the independent validation cohorts, the Streptococcus -based diagnostic signature exhibited outstanding discriminatory capacity, attaining AUCs of 0.85 and 0.99, respectively. Meanwhile, the response prediction model based on Prevotella salivae and Neisseria oralis yielded an AUC of 0.74. Regression analysis demonstrated that oral microbiota richness and diversity were inversely associated with ECOG score and ProGRP level in small cell lung cancer patients. Conclusions Lung cancer patients harbor distinct oral microbiota signatures whose dynamics correlate significantly with therapeutic response and survival outcomes. The developed diagnostic and predictive models demonstrate robust performance, supporting oral microbiota as non-invasive biomarkers for lung cancer management.