铜死亡相关基因在肝细胞癌中的表达及其临床意义

Expressions of cuproptosis-related genes in hepatocellular carcinoma and their clinical significance

背景与目的:肝细胞癌(HCC)是一种全球常见的恶性肿瘤,具有高复发率和高病死率。铜死亡是一种新型的程序性细胞死亡,涉及肿瘤细胞的增殖和生长、血管生成和转移。因此,本研究探讨铜死亡相关基因(CRGs)在HCC中的表达与预后的关系,并建立预后相关的列线图模型以及分析CRGs与HCC免疫细胞浸润的关系。方法:使用R语言“limma”包对TCGA数据库下载的HCC组织与正常组织的数据中CRGs进行差异表达分析;“clusterProfiler”包进行GO和KEGG分析;单因素Cox回归分析筛选与预后相关的CRGs,LassoCox回归分析构建HCC中CRGs相关预后评分模型;“ggsurvplot”包以总生存(OS)为结局绘制Kaplan-Meier生存曲线;“survival ROC”包绘制ROC曲线评估预后评分的准确性;“regplot”和“rms”包绘制列线图和校准曲线;利用TIMER数据库分析CRGs的表达与6种免疫细胞丰度之间的关系。结果:与正常组织相比,HCC组织19个CRGs中的16个有差异表达(上调:PDHB、PDHA1、MTF1、LIPT1、LIPT2、LIAS、GLS、DLD、DLST、DLAT、CDKN2A、ATP7A;下调:SLC31A1、GCSH、DBT、NLRP3);NLRP2的突变频率最高(12%)。GO和KEGG分析表明,CRGs富集于三羧酸循环、碳代谢作用、丙酮酸代谢、糖酵解/糖异生和铂类药物耐药性等信号通路。基于单因素Cox回归分析和Lasso-Cox回归分析筛选出影响HCC预后OS的3个CRGs (CDKN2A、GLS、DLAT)作为预后生物标志物构建预后模型,并使用回归系数构建预后评分:风险评分=0.22×DLAT (表达水平)+0.11×CDKN2A (表达水平)+0.03×GLS (表达水平)。Kaplan-Meier曲线显示,高风险评分HCC患者预后较差(P<0.05),用风险模型的时间依赖ROC曲线评价模型预测性能,1、3、5年的AUC分别为0.741,0.657,0.633。将年龄、性别、T分期、N分期、M分期、病理分型、CDKN2A、GLS和DLAT纳入构建列线图,校准图显示列线图预测和实际观察之间有良好的一致性。GLS、DLAT和CDKN2A与HCC免疫细胞浸润呈正相关,并与免疫检查点PDCD1、 CD274、 HAVCR2明显相关(均P<0.05)。进一步分析表明,HCC组织中CDKN2A、GLS和DLAT表达越高,患者巴塞罗那病理分期越晚,组织学分级越差(均P<0.05)。结论:与铜死亡相关的基因特征可以作为HCC患者潜在的预后预测因子,并可能为HCC治疗提供新的途径。

Background and Aims: Hepatocellular carcinoma(HCC) is a common malignancy with a high recurrence and mortality rate. Cuproptosis is a new type of programmed cell death involved in tumor cells’ proliferation, growth, angiogenesis, and metastasis. Therefore, this study aims to investigate the relationship between the expression of cuproptosis-related genes(CRGs) and the prognosis in HCC, establish a prognosis-related nomogram model, and analyze the association of CRGs with the immune cell infiltration in HCC.Methods: Differential expression analysis of CRGs in the TCGA database was performed using the R language "limma" package;the "clusterProfiler" package was used for GO and KEGG analysis;the prognostic CRGs were screened by univariate Cox regression analysis;the prognostic scoring model based on CRGs for HCC was constructed by Lasso-Cox regression analysis;the "ggsurvplot" package drew the Kaplan-Meier survival curve draws using overall survival(OS) as the outcome variable;the "survival ROC" package created the ROC curve for assessing the accuracy of the prognostic score;the nomogram and the calibration curves were drawn by the ’regplot’ and ’rms’ packages;the associations between the expression of CRGs and the abundance of six immune cells were analyzed using the TIMER database.Results: Among the 19 CRGs, there were 16 differentially expressed ones in HCC tissue compared with normal tissue(up-regulation: PDHB, PDHA1, MTF1, LIPT1, LIPT2, LIAS, GLS, DLD, DLST, DLAT, CDKN2A, and ATP7A;down-regulation: SLC31A1, GCSH, DBT, and NLRP3), and NLRP 2 had the highest mutation frequency of 12%. GO, and KEGG analyses showed that CRGs were enriched in signaling pathways such as the tricarboxylic acid cycle, carbon metabolism, pyruvate metabolism, glycolysis/gluconeogenesis, and platinum drug resistance. Three CRGs(CDKN2A, GLS, and DLAT) that affected the OS were screened by univariate Cox regression analysis and LASSO Cox regression analysis for the construction of the prognostic model, and the prognostic score was constructed using regression coefficient: risk score=0.22×DLAT(expression level) + 0.11×CDKN2A(expression level) + 0.03×GLS(expression level). The Kaplan-Meier curve analysis showed that the HCC patients with highrisk scores had a poor prognosis(P<0.05), and the model prediction performance was evaluated by the time-dependent ROC curve of the risk model, and the AUC at 1, 3, and 5 years was 0.741, 0.657 and 0.633, respectively. The nomogram was constructed by incorporating age, sex, T stage, N stage, M stage, pathological classification, CDKN2A, GLS, and DLAT. The calibration map showed good consistency between the nomogram prediction and the actual observation. There were positive correlations of GLS, DLAT, and CDKN2A with HCC immune cell infiltration and a significant correlation with immune checkpoints PDCD 1, CD274, and HAVCR2(all P<0.05). Further analysis indicated that the higher CDKN2A, GLS, and DLAT expression in HCC tissue, the later the Barcelona pathological stage, the worse the histological grade in patients(all P<0.05).Conclusion: Gene signatures associated with cuproptosis can be used as potential prognostic predictors for HCC patients and may provide new insights into the treatment of HCC.