A model based on Immunogenic cell death-related genes can predict prognosis and drug sensitivity in patients with hepatocellular carcinoma

Abstract

Background: Immunogenic cell death (ICD) can activate the immune system against cancer in host with immunocompetent. However, the prognostic role of ICD-related genes (IRGs) in hepatocellular carcinoma (HCC) is unknown.We aimed to establish a prognostic model based on IRGs, and to verify the relationship between this model and the immune microenvironment of HCC, and whether this model can predict the prognosis of patients with HCC. Methods: The Cancer Genome Atlas-Liver Hepatocellular Carcinoma(TCGA-LIHC), and The Liver Cancer, Riken Japan (LIRI-JP) were downloaded via public databases, and 34 IRGs were extracted from the literature. First, consensus clustering analysis was applied in TCGA-LIHC to classified LIHC samples into different clusters based on IRGs. Differentially expressed genes (DEGs) between LIHC and normal samples in TCGA-LIHC, and DEGs among clusters were respectively sifted out through differential expression analysis, and they were overlapped to obtain IRGs-DEGs. Next, univariate Cox and least absolute shrinkage and selection operator (LASSO) analyses were implemented on IRGs-DEGs for identifing prognosis related genes, and HCC samples were divided into high/low risk groups via risk score. Moreover, independent prognostic factors were yielded via univariate and multivariate Cox analyses, and enrichment analysis was applied for exploring biological pathways related to the prognosis model. Finally, the role of prognostic model in immune microenvironment and chemotherapy of HCC were analyzed. Results: HCC samples were classified into two clusters in the light of IRGs, and cluster2 had a poorer survival state than that of cluster1. Totally 2197 DEGs in TCGA-LIHC and 112 DEGs between clusters were yielded, and they were intersected to get 72 IRGs-DEGs. Six prognosis-related genes, namely KRT20, MMP12, AGR2, CXCL5, CYP3A4, and MAGEA8 were finally identified via univariate Cox and LASSO analyses. Besides, the risk score and grade were found to be correlated with LIHC prognosis. Obviously, the prognostis model was related to immune, and metabolism related pathways like nitrogen metabolism, and adaptive immune response. Moreover, tumor immune dysfunction and exclusion (TIDE) score was sensibly lowly expressed in low risk group, suggesting that low risk group patients were more susceptible to immunotherapy. Ultimatly, high risk group was more sensitive to Camptothecin, Sorafenib and others, while low risk group was more responsive to Veliparib and Dabrafenib. Conclusion: Through bioinformatic analysis, a prognosis model (contained KRT20, MMP12, AGR2, CXCL5, CYP3A4, and MAGEA8) was constructed in HCC, contributing to studies related to prognosis and treatment of HCC.