探讨去泛素化酶USP24及其共表达肿瘤代谢基因在肝细胞癌临床诊断和预后预测中的作用。

使用TCGA-LIHC队列研究USP24在肝细胞癌中的表达情况。采用生存分析评估USP24在肝细胞癌预后中的作用。采用GSEA软件预测USP24可能调控的肿瘤代谢途径，并通过共表达分析对肿瘤代谢通路主要贡献基因进行筛选。使用差异表达分析和单因素COX分析寻找与USP24相关的肿瘤代谢预后基因，并使用LASSO-COX算法构建预后模型。采用BP神经网络、支持向量机、随机森林和XGBoost算法构建肝细胞癌诊断模型。使用R4.2.2软件进行统计学分析，两组之间的比较使用Wilcoxon检验；生存分析采用Kaplan-Meier法。P<0.05为差异有统计学意义。

USP24在肝细胞癌中显著高表达，且USP24高表达组的患者生存率较低。GSEA分析显示USP24与11种肿瘤代谢途径相关。共表达分析、差异表达分析和单因素COX分析筛选出19个USP24相关肿瘤代谢预后基因，LASSO-COX算法选择8个基因用于构建预后模型。生存分析结果提示高风险组的预后较低风险组更差。训练组1年、2年、3年ROC曲线的AUC分别为0.793、0.706、0.696；验证组1年、2年、3年ROC曲线的AUC分别为0.701、0.684、0.728。肝细胞癌诊断模型中XGBoost模型具有较好的诊断能力，验证组ROC曲线的AUC为0.809。

USP24与肝细胞癌预后相关，可能是通过肿瘤细胞代谢调控肝细胞癌的进展。基于8个肿瘤代谢基因构建的诊断模型和预后模型可能分别有效诊断肝细胞癌以及预测患者的预后。

To investigate the role of deubiquitination enzyme USP24 and its co-expression of tumor metabolic genes in clinical diagnosis and prognosis prediction of hepatocellular carcinoma.

The expression of USP24 in hepatocellular carcinoma was studied using the TCGA-LIHC cohort. Survival analysis was used to evaluate the role of USP24 in the prognosis of hepatocellular carcinoma. GSEA software was used to predict the possible tumor metabolic pathways regulated by USP24, and the major contributing genes of tumor metabolic pathways were screened by co-expression analysis. Differential expression analysis and univariate COX analysis were used to identify the metabolic prognostic genes associated with USP24, and the prognostic model was constructed using LASSO-COX algorithm. The diagnosis model of hepatocellular carcinoma was constructed by BP neural network, support vector machine, random forest and XGBoost algorithm. R4.2.2 software was used for statistical analysis, Wilcoxon test was used for comparison between the two groups, and Kaplan-Meier method was used for survival analysis. P<0.05 was considered statistically significant.

USP24 is significantly overexpressed in hepatocellular carcinoma, and the survival rate of patients with high expression of USP24 is lower. GSEA analysis showed that USP24 was associated with 11 tumor metabolic pathways. Coexpression analysis, differential expression analysis and univariate COX analysis selected 19 USP24-related metabolic prognostic genes, and 8 genes were selected by LASSO-COX algorithm to construct prognostic model. Survival analysis results suggested that the high-risk group had a worse prognosis than the low-risk group. The AUC of 1-year, 2-year and 3-year ROC curves of the training group were 0.793, 0.706 and 0.696, respectively, while the AUC of 1-year, 2-year and 3-year ROC curves of the verification group were 0.701, 0.684 and 0.728, respectively. The XGBoost model had better diagnostic ability, and the AUC of ROC curve in the verification group was 0.809.

USP24 is associated with the prognosis of hepatocellular carcinoma, which may regulate the progression of hepatocellular carcinoma through tumor cell metabolism. The diagnostic model and prognostic model constructed based on 8 tumor metabolic genes may be effective in diagnosing hepatocellular carcinoma and predicting the prognosis of patients, respectively.