The objective of this study is to form a cancer stem cell index-based model to stratify HCC risk and predict survival. After screening the Tumor Genome Atlas (TCGA) of liver and normal liver tissue samples, we obtained differentially expressed genes (DEGs). We employed a weighted correlation network analysis (WGCNA) and differentially expressed genes were studied in HCC to find the modules most associated with cancer stem cells (mRNAsi). At the same time, gene ontology and Kyoto Genome Encyclopedia (KEGG) were used for functional annotation and combined with LASSO, univariate, and multivariate COX regression analyses, a prediction model of key module genes of cancer stem cells was developed. The model's clinical efficacy was measured using the C index, calibration curve, multiindex ROC curve, and clinical decision curve. WGCNA found that black modules were most correlated with tumour stem cell index. Seven genes (CSDC2, GNA14, LGI2, MMRN1, PDE2A, SELP, and STK32B) were filtered by univariate, LASSO, and multivariate Cox regression analyses to establish the primary HCC model. The survival analysis and ROC curve in the TCGA training and validation cohort showed good performance. The independent prognostic factor of primary HCC was risk score, according to univariate and multivariate Cox regression analyses. It is found that the stem cell index model of 7 genes could predict factors independently, indicating that signatures of the stem cell will play a significant role in liver cancer survival prediction and risk stratification.
Keyphrases