Gut Clostridium aminophilum as a protective microbe in never-smoker lung cancer: insights from metagenomics and machine learning

Lung cancer in never-smoking patients (LCINS) has gradually gained attention in recent years, and emerging evidence points to a potential role of the gut microbiota in LCINS by modulating host immunity and metabolism. In this study, metagenomics was performed on stool samples from 24 LCINS patients and 18 healthy individuals. LCINS patients exhibited reduced microbial richness in α-diversity compared with healthy controls, and differences in principal component distribution were observed. Pathway enrichment analyses revealed that LPS biosynthesis capacity was significantly increased, whereas the fatty acid biosynthesis pathway was significantly decreased in LCINS. A total of 317 differentially abundant species were identified, including 29 assigned to the module most strongly associated with LCINS based on weighted gene co-expression network (WGCNA). Then, 36 machine learning–based diagnostic models were constructed based on these 29 species, with the glmBoost + Ridge model showing the best performance for LCINS discrimination (AUC = 0.855). SHAP (SHapley Additive exPlanations) identified Clostridium aminophilum as a protective factor and the top contributor to model performance (SHAP value = 0.068), with a single-species AUC of 0.743. In vitro , the culture supernatant of Clostridium aminophilum markedly suppressed A549 cell proliferation and migration. In vivo , oral administration of Clostridium aminophilum (5 × 10 ⁶ CFU) for 21 days significantly suppressed tumor growth in subcutaneous A549-derived xenograft models. These findings suggest a significant association between LCINS and the gut microbiota, characterized by a marked depletion of Clostridium aminophilum . Notably, metabolites derived from Clostridium aminophilum were shown to inhibit lung cancer progression.