摘要Background and aims:Hepatocellular carcinoma(HCC)is one of the leading causes of cancer death.Multi-pathway combination therapy is used to treat HCC,and immunotherapy is also a routine part of treatment.As a major component of the tumor microenvironment(TME),cancer-associated fibroblasts(CAFs)actively participate in cancer progression through complex functions.However,because CAFs dynamically change during cancer development,most of the current treatment strategies targeting CAFs fail.We created a prognostic CAF marker gene signature(CAFMGS)to investigate the utility of CAFs as a prognostic factor and therapeutic target.Methods:Gene Expression Omnibus(GEO)single-cell RNA sequencing(Sc-RNA-seq)data were analyzed to identify CAF marker genes in HCC.The Cancer Genome Atlas(TCGA)database was used as a training cohort to construct the CAFMGS model and the International Cancer Genome Consortium(ICGC)dataset was used to validate the CAFMGS.Results:Marker genes in the CAFMGS model were(0.0001-SPP1),(0.0084-VCX3A),(0.0015-HMGA1),(0.0082-PLOD2),and(0.0075-CACYBP).The CAFMGS_score was separated into high-risk and low-risk groups based on the median of the patients'OS.Univariate and multivariate analyses confirmed that CAFMGS_score was an inde-pendent prognostic factor in the training group.CAFMGS_score was a more accurate prognostic indicator compared with clinicopathological score and tumor mutational burden score.Conclusion:CAFMGS offers a fresh perspective on stromal cell marker genes in HCC prognosis and expands our knowledge of CAF heterogeneity and functional diversity,perhaps paving the way for CAF-targeted immuno-therapy in HCC patients.