A cuproptosis random forest cox score model-based evaluation of prognosis, mutation characterization, immune infiltration, and drug sensitivity in hepatocellular carcinoma.

Abstract

Hepatocellular carcinoma is the third most deadly malignant tumor in the world with a poor prognosis. Although immunotherapy represents a promising therapeutic approach for HCC, the overall response rate of HCC patients to immunotherapy is less than 30%. Therefore, it is of great significance to explore prognostic factors and investigate the associated tumor immune microenvironment features. By analyzing RNA-seq data of the TCGA-LIHC cohort, the set of cuproptosis related genes was extracted via correlation analysis as a generalization feature. Then, a random forest cox prognostic model was constructed and the cuproptosis random forest cox score was built by random forest feature filtering and univariate multivariate cox regression analysis. Subsequently, the prognosis prediction of CRFCS was evaluated via analyzing data of independent cohorts from GEO and ICGC by using KM and ROC methods. Moreover, mutation characterization, immune cell infiltration, immune evasion, and drug sensitivity of CRFCS in HCC were assessed. A cuproptosis random forest cox score was built based on a generalization feature of four cuproptosis related genes. Patients in the high CRFCS group exhibited a lower overall survival. Univariate multivariate Cox regression analysis validated CRFCS as an independent prognostic indicator. ROC analysis revealed that CRFCS was a good predictor of HCC (AUC =0.82). Mutation analysis manifested that microsatellite instability (MSI) was significantly increased in the high CRFCS group. Meanwhile, tumor microenvironment analysis showed that the high CRFCS group displayed much more immune cell infiltration compared with the low CRFCS group. The immune escape assessment analysis demonstrated that the high CRFCS group displayed a decreased TIDE score indicating a lower immune escape probability in the high CRFCS group compared with the low CRFCS group. Interestingly, immune checkpoints were highly expressed in the high CRFCS group. Drug sensitivity analysis revealed that HCC patients from the high CRFCS group had a lower IC50 of sorafenib than that from the low CRFCS group. In this study, we constructed a cuproptosis random forest cox score (CRFCS) model. CRFCS was revealed to be a potential independent prognostic indicator of HCC and high CRFCS samples showed a poor prognosis. Interestingly, CRFCS were correlated with TME characteristics as well as clinical treatment efficacy. Importantly, compared with the low CRFCS group, the high CRFCS group may benefit from immunotherapy and sorafenib treatment.