Construction and validation of a novel signature based on epithelial-mesenchymal transition–related genes to predict prognosis and immunotherapy response in hepatocellular carcinoma by comprehensive analysis of the tumor microenvironment

Abstract

Abstract Immunotherapy has yielded encouraging results in the treatment of advanced hepatocellular carcinoma (HCC). However, the relationship between epithelial-mesenchymal transition (EMT) and immunotherapy for HCC has not been adequately explained. In this study, we comprehensively analyzed a bulk RNA sequence dataset of 365 HCC patients in The Cancer Genome Atlas (TCGA) dataset. Subsequently, we constructed a prognostic signature based on 6 EMT-related genes and divided 365 HCC patients into high- and low-risk groups. The predictive efficacy of the signature was well validated in different clinical subgroups and in two independent external datasets. We further explored the relationship between prognostic signature and immunotherapy response in terms of immune cell infiltration, somatic mutations, tumor mutation burden (TMB), microsatellite instability (MSI), immune checkpoint–associated gene expression, single-nucleotide variants (SNV) neoantigens, cancer testicular antigens (CTA) scores, and tumor immune dysfunction and exclusion (TIDE) scores. We validated the predictive efficacy of prognostic signature for immunotherapy response using external independent immunotherapy data. Real-time quantitative polymerase chain reaction (qRT-PCR) was used to validate EMT-related gene overexpression in HCC tissue samples. Prognostic signature was an independent risk factor affecting the prognosis of HCC patients and has shown superiority in predicting patient survival compared to other clinical factors. Compared with the low-risk group, the proportion of Activated_CD4_T_cell, Type_2_T_helper_cel, and macrophages were higher in the tumor microenvironment of HCC patients in the high-risk group, while the Activated_CD8_T_cell and CD56bright_natural_killer_cell proportions were lower. The prognostic signature was positively correlated with TMB scores, MSI scores, SNV neoantigens scores, expression levels of immune checkpoint–related genes, and TIDE scores, and patients in the high-risk group were more suitable for immunotherapy. qRT-PCR confirms overexpression of 6 EMT-related genes in HCC tissues for the construction of prognostic signature. Our novel prognostic signature can effectively predict the prognosis and immunotherapy response of HCC patients. In the future, it will be an effective tool for physicians to screen suitable immunotherapy populations and improve response rates and overall survival (OS).