Streptomyces-linked exercise gene networks orchestrate immune exclusion and determine immunotherapy response in hepatocellular carcinoma

Exercise-related genes (ERGs) have emerged as potential modulators of tumor biology, yet their systematic characterization in hepatocellular carcinoma (HCC) remains incomplete. Here we integrated 320 ERGs from MSigDB with TCGA-HCC cohort to construct an interaction-perturbation network, identifying two distinct subtypes with divergent prognostic outcomes. Cluster1 (44.4% of patients) exhibited significantly higher network perturbation scores, activated proliferation and epithelial-mesenchymal transition pathways, and elevated immune checkpoint gene expression, collectively contributing to poorer survival. This subtype also demonstrated enhanced immune evasion potential and lower predicted immunotherapy response rates. Conversely, Cluster2 (55.6%) was characterized by metabolic pathway enrichment, increased CTNNB1 mutations, higher tumor mutational burden, and more favorable immunotherapy prediction. Notably, integrated analysis of intratumoral microbiota revealed that Streptomyces abundance was significantly associated with immune exclusion features and negatively correlated with both exercise-related and lactate metabolic pathways. Mediation analysis further suggested that elevated lactate metabolism was statistically associated with adaptive microbial changes, which in turn showed directional associations with exercise-related pathway activity and clinical outcomes. Using Streptomyces-associated host genes, we developed an 8-gene prognostic signature that effectively stratified patient outcomes and predicted immunotherapy response across multiple independent cohorts. Our findings delineate exercise-related molecular subtypes with distinct immune-microbiota-metabolic crosstalk and provide a clinically applicable signature for risk stratification in HCC.