Abstract
BACKGROUND: Hepatocellular carcinoma (HCC) is a lethal tumor. Its prognosis prediction remains a challenge. Meanwhile, cellular senescence, one of the hallmarks of cancer, and its related prognostic genes signature can provide critical information for clinical decision-making. METHOD: Using bulk RNA sequencing and microarray data of HCC samples, we established a senescence score model via multi-machine learning algorithms to predict the prognosis of HCC. Single-cell and pseudo-time trajectory analyses were used to explore the hub genes of the senescence score model in HCC sample differentiation. RESULT: A machine learning model based on cellular senescence gene expression profiles was identified in predicting HCC prognosis. The feasibility and accuracy of the senescence score model were confirmed in external validation and comparison with other models. Moreover, we analyzed the immune response, immune checkpoints, and sensitivity to immunotherapy drugs of HCC patients in different prognostic risk groups. Pseudo-time analyses identified four hub genes in HCC progression, including CDCA8, CENPA, SPC25, and TTK, and indicated related cellular senescence. CONCLUSIONS: This study identified a prognostic model of HCC by cellular senescence-related gene expression and insight into novel potential targeted therapies.