Construction a six-gene prognostic model for hepatocellular carcinoma based on WGCNA co-expression network

Abstract

Objective: Currently, the incidence of hepatocellular carcinoma remains high, and the prognosis of patients is poor. Prognostic biomarkers are still worth exploring. Methods: Based on The Cancer Genome Atlas (TCGA) database, the differentially expressed genes (DEGs) were screened. Subsequently, a modular analysis of these DEGs was performed using the weighted gene co-expression network analysis (WGCNA). A prognostic model for liver cancer patients was constructed employing the Cox proportional hazards model. Through univariate and multivariate Cox regression analyses, we developed a Cox proportional-hazards model specifically for hepatocellular carcinoma. Subsequently, International Cancer Genome Consortium (ICGC) cohort data were used to validate the accuracy of the Cox proportional-hazards model. Following this, we conducted further analyses of prognostic genes, encompassing functional enrichment analysis and survival analysis. Additionally, we utilized the BBcancer database to investigate whether these prognostic genes have the potential to serve as blood markers. Notably, in this six-gene prognostic model, we also analyzed the genes' drug susceptibility. Results: Leveraging the candidate genes identified from the WGCNA analysis, we constructed a Cox proportional-hazards model with an AUC value greater than 0.7. This model incorporates HMMR, E2F2, WDR62, KIF11, MSH4, and KCNF1, revealing that patients with low expression levels of these genes had significantly better survival prognosis compared to those with high expression levels (P ​< ​0.05). The enrichment analysis revealed that these prognostic genes are enriched in pathways related to hepatitis B, hepatitis C, and hepatocellular carcinoma. Furthermore, we observed a strong association between HMMR, E2F2, WDR62, KIF11, MSH4, and KCNF1 with overall survival (OS) in hepatocellular carcinoma (HCC) patients, among which HMMR, E2F2, WDR62 and KIF11 genes were significantly differentially expressed in extracellular vesicles. Additionally, this six-gene prognostic model demonstrated sensitivity to drugs such as VX-680, TAE684, Sunitinib, S-Trityl-L-cysteine, Paclitaxel, and CGP-60474. Conclusion: The Cox risk prognostic model based on HMMR, E2F2, WDR62, KIF11, MSH4, and KCNF1 represents a valuable tool for predicting the prognosis of HCC patients and may serve as a target for drug development. In particular, HMMR, E2F2, WDR62, and KIF11 have potential as blood biomarkers for hepatocellular carcinoma, though their precise biological functions require further exploration.

Keywords

• Hepatocellular carcinoma
• WGCNA
• Prognostic model
• Bioinformatics
• Methylation analysis
• Drug sensitivity analysis