Integrated microbiome-metabolome profiling unveils a predictive signature for early recurrence in hepatocellular carcinoma

Hepatocellular carcinoma (HCC) exhibits high recurrence rates post-resection, yet predictive biomarkers remain elusive. Emerging evidence implicates intratumoral microbiota in cancer progression, but its role in HCC recurrence is unexplored. Here, we characterized microbial and metabolic profiles in 90 HCC patients [49 with early recurrence (RFS ≤ 2 years), 41 non-recurrent controls] using 16S rRNA sequencing and LC–MS metabolomics. Recurrent tumors showed reduced microbial diversity (Shannon index, p < 0.05) and distinct compositional shifts, including enrichment of Proteobacteria (LEfSe LDA > 4) and depletion of commensals like Akkermansia. A 20-microbial-genus signature predicted recurrence (AUC = 0.81, 95% CI: 0.72–0.91), while a 20-metabolite panel (e.g., resolvin D5, γ -glutamylthreonine) achieved superior accuracy (AUC = 0.958, CI: 0.950–0.966). Functional analyses linked recurrence-associated microbiota with disrupted lipid/amino acid metabolism and pro-inflammatory pathways (KEGG, p < 0.01). Microbial-metabolite correlation networks revealed strong associations between dysbiotic taxa (e.g., Cyanobacteria) and immunomodulatory metabolites (* r > 0.6, p < 0.05). This study identifies intra-tumoral microbiome-metabolome signatures as novel biomarkers for HCC recurrence, offering mechanistic insights into microbial regulation of the tumor microenvironment and clinical tools for post-surgical risk stratification.