Abstract
BACKGROUND: Inflammatory bowel disease (IBD) commonly coexists with non-alcoholic fatty liver disease (NAFLD). Despite metabolic factors being less involved, IBD patients exhibit a higher risk of developing NAFLD compared to non-IBD individuals. Given the shared role of gut dysbiosis in the pathogenesis of both diseases, this study investigated the involvement of gut microbiota and associated metabolic pathways in IBD-associated NAFLD (COMO). METHODS: A retrospective analysis of clinical profiles from 490 IBD, 89 NAFLD, and 68 COMO patients was conducted. Fecal samples from 30 IBD, 32 NAFLD, 26 COMO patients and 29 healthy controls were prospectively collected and subjected to 16 S rRNA gene sequencing for microbial community analysis and functional pathway prediction. Subsequently, machine learning modeling was employed for feature importance analysis and identification of COMO patients. RESULTS: Demographic analysis revealed that COMO patients developed NAFLD earlier than NAFLD alone, with fewer metabolic associations with hypertension, hyperlipidemia and glucose dysregulation. Compared with IBD and NAFLD groups, COMO microbiota exhibited lower alpha diversity, with beta diversity aligning with IBD but distinct from NAFLD group. Shared microbial signatures included increased Lactococcus and decreased Coprococcus 3 and Ruminococcus 2, which was correlated with 11 metabolic pathways: five vitamin B pathways (thiamine, vitamin B6, biotin, folate and riboflavin), isoflavonoid, caffeine, phosphonate, cyanoamino acid, lipoic acid and ubiquinone pathways. Integrated microbial-metabolic machine learning models (logistic regression, random forest, support vector machine, and XGBoost) achieved AUC of 0.818-0.864 for COMO identification. CONCLUSIONS: Our findings implicate microbiota-mediated metabolic reprogramming in IBD-associated NAFLD pathogenesis, highlighting potential therapeutic targets for the treatment and prevention of NAFLD in IBD.