Journal Article

Construction of a novel G2/M checkpoint-related gene signature for prognosis in hepatocellular carcinoma based on machine learning

Lin, Yucong, Guo, Kunxiong, Lin, Wanting

2025

Overview

Purpose Hepatocellular carcinoma (HCC) is recognized as one of the most aggressive cancers worldwide. This study aims to discover novel biomarkers that can effectively stratify patients at high risk for HCC and facilitate personalized management. Methods Differentially expressed genes analysis was conducted to identify marker genes associated with HCC based on transcriptomic profiles from the TCGA-LIHC database. And then, we utilized the weighted correlation network analysis (WGCNA) to explore the correlation between various gene expression clustering modules based on the the identified differentially expressed genes above and clinical phenotypes. Subsequently, we integrated 10 machine learning algorithms into 117 combinations to establish the optimal model for predicting survival in HCC patients. Then, immune landscape was evaluated. Additionally, single-cell RNA sequencing analysis revealed the expression patterns of genes involved in model at single-cell resolution. Furthermore, the role of DTYMK was examined in vitro in two liver cancer cell lines. Results Forty-eight G2/M checkpoint-related genes were identified as marker genes for HCC. Among these, STMN1, DTYMK, EZH2, PBK, and UCK2 were incorporated into a prognostic model (G2/MR), which demonstrated robust and reliable performance in predicting the survival rates of HCC patients. Individuals with high G2/MR scores exhibited shorter survival and higher regulatory T cells infiltration compared to those with low G2/MR scores. Silencing DTYMK arrested cell cycle and inhibited proliferation, migration, invasion, and colony formation in Huh-7 and PLC/PRF/5 cells. Conclusion Our study developed a five-gene risk score model that may serve as a valuable tool for prognostic assessment and personalized HCC management.

• Publisher: Springer US,Springer Nature B.V,Springer
• Subject: Algorithms / Antibodies / Biomarkers / Cancer Research / Cancer therapies / Cell cycle / Correlation analysis / G2/M checkpoint / Gene expression / Genomes / Hepatocellular carcinoma / Internal Medicine / Kinases / Liver cancer / Machine learning / Medical prognosis / Medicine / Medicine & Public Health / Membranes / Molecular Medicine / Oncology / Prognostic signature / Proteins / Radiotherapy / Reagents / Risk model / Surgical Oncology