Construction and Analysis of Hepatocellular Carcinoma Prognostic Model Based on Random Forest.

Abstract

Methods: Transcriptome data and clinical data of HCC were downloaded from the TCGA database. Screen important genes based on the random forest method, combined with differential expression genes (DEGs) to screen out important DEGs. The Kaplan‒Meier curve was used to evaluate its prognostic significance. Cox regression analysis was used to construct a survival prognosis prediction model, and the ROC curve was used to verify it. Finally, the mechanism of action was explored through GO and KEGG pathway enrichment and GeneMANIA coexpression analyses.

Results: Seven important DEGs were identified, three were highly expressed and four were lowly expressed. Among them, GPRIN1, MYBL2, and GSTM5 were closely related to prognosis (P < 0.05). After the survival prognosis prediction model was established, the survival analysis showed that the survival time of the high-risk group was significantly shortened (P < 0.001), but the ROC analysis indicated that the model was not superior to staging. Twenty coexpressed genes were screened, and enrichment analysis indicated that glutathione metabolism was an important mechanism for these genes to regulate HCC progression.

Conclusion: This study revealed the important DEGs affecting HCC progression and provided references for clinical assessment of patient prognosis and exploration of HCC progression mechanisms through the construction of predictive models and gene enrichment analysis.