Abstract
Cirrhosis remains a significant global health burden, causing approximately 1.4-1.5 million deaths each year and contributing to nearly 46 million disability-adjusted life years (DALYs) worldwide. Increasing evidence identifies the gut-liver axis as a central driver of disease progression, wherein intestinal dysbiosis, barrier disruption, and microbe-derived metabolites collectively exacerbate inflammation, fibrogenesis, and related complications. Across more than 40 recent studies, gut microbial α-diversity declined by 30-60%, and over 80% reported a marked depletion of short-chain fatty acid (SCFA)-producing taxa, particularly Lachnospiraceae and Ruminococcaceae. Meta-analyses indicate that fecal butyrate levels decrease by 40-70%, accompanied by a two- to fourfold increase in endotoxin concentrations. Bile acid profiling demonstrates an approximately 50% reduction in secondary bile acids and significant suppression of FXR/TGR5 signaling, whereas tryptophan metabolism shifts toward the kynurenine pathway, weakening epithelial defense and exacerbating portal hypertension. Clinically, dysbiosis and microbial translocation are associated with higher MELD scores, and patients in the lowest quartile of microbial diversity have a threefold increased risk of hepatic encephalopathy or spontaneous bacterial peritonitis. Microbiome-targeted interventions-including lactulose, rifaximin, probiotics or synbiotics, fecal microbiota transplantation, and bile acid modulators-restore community balance in 70-85% of clinical trials, although efficacy and safety vary by etiology and baseline microbiota composition. Integrated microbiome-metabolome models achieve areas under the curve (AUCs) of 0.82-0.90 for noninvasive classification and early detection of cirrhosis. Collectively, these findings underscore reproducible, quantitative microbiome-metabolite alterations and outline a roadmap for microbiome-informed precision care that connects mechanistic insight with clinical application, emphasizing the need for longitudinal and multi-ethnic validation.