摘要
肝癌是常见的恶性肿瘤之一,晚期肝癌的预后极差。鉴于线粒体氧化应激在肝癌发生发展中的重要作用,选用线粒体氧化应激相关基因构建肝细胞癌(hepatocellular carcinoma,HCC)预后风险模型。首先,结合单因素Cox回归分析与支持向量机、随机森林分析、LASSO回归分析3种机器学习方法筛选预后关键基因,并基于多因素Cox回归分析构建模型;其次,在数据库中对模型的预后价值进行验证;再次,利用基因富集分析探讨高低风险组间预后差异的可能机制,并比较两组间的免疫微环境及治疗反应;最后,使用实时荧光定量逆转录PCR(reverse transcription quantitative real-time PCR;RT-qPCR)验证关键基因在HCC组织中的表达。结果共筛选出PDE2A、TREM2、BMP6、NQO1、CPS1、EPO、MAPT、G6PD、SFN、HMOX1十个基因。与低风险组比较,高风险组HCC患者预后较差(P<0.0001)。富集分析表明,过氧化物酶体增殖物激活受体(peroxisome proliferator-activated receptor,PPAR)信号通路等在高低风险组间存在显著差异。肿瘤免疫分析表明,肿瘤免疫浸润、免疫检查点相关基因、免疫治疗反应等在高低风险组间也存在显著差异。RT-qPCR的验证结果表明,相比正常肝组织,HCC组织中CPS1、PDE2A、BMP6的表达显著降低(P<0.05),而G6PD、SFN的表达显著升高(P<0.05)。总之,本研究建立的线粒体氧化应激相关HCC预后风险模型具有良好的预测效能及准确度,可用于HCC的精准治疗,有较高的临床应用价值。
Liver cancer is one of the common malignant tumors,and the prognosis for advanced liver cancer is extremely poor.In view of the important role of mitochondrial oxidative stress in the development of hepatocellular carcinoma(HCC),mitochondrial oxidative stress-related genes were selected to construct a prognostic risk model for HCC.Firstly,the prognostic key genes were screened by using univariate Cox regression analysis and three machine learning methods,namely support vector machine;Random forest analysis,and LASSO regression analysis,and a model was constructed based on multivariate Cox regression.Secondly,the prognostic value of the model was further validated in the database.Thirdly,the possible mechanisms of the prognostic differences between the high-and low-risk groups were explored using gene enrichment analysis,and the immune microenvironment and treatment response between the two groups were compared.Finally,the expression of key genes in liver cancer tissues was verified by reverse transcription quantitative time PCR(RT-qPCR).Results showed that a total of 10 genes including PDE2A,TREM2,BMP6,NQO1,CPS1,EPO,MAPT,G6PD,SFN and HMOX1 were chosen out.Compared with the low-risk group,the highrisk group of HCC patients had a worse prognosis(P<0.0001).Enrichment analysis showed that peroxisome proliferator-activated receptor(PPAR)signaling pathway was significantly different between the high-and low-risk groups.And the tumor immunity analysis showed that the tumor immune infiltration,immune checkpoint-related genes,and immunotherapy response were also significantly different between the two groups.Validation results using RT-qPCR indicated that,compared with normal liver tissues,expressions of CPS1,PDE2A and BMP6 were lower in HCC tissues(P<0.05),while expressions of G6PD and SFN were higher in HCC tissues(P<0.05).In conclusion,the mitochondrial oxidative stress-related prognostic risk model established in this study has good predictive efficacy and accuracy,and can be used for the precise treatment of HCC.It would have a high clinical application value.
作者
陈柯宇
张暕
伍次春
蒋川
彭仕芳
傅蕾
CHEN Keyu;ZHANG Jian;WU Cichun;JIANG Chuan;PENG Shifang;FU Lei(Department of Infectious Diseases,Xiangya Hospital,Central South University,Changsha 410008,Hunan,China)
出处
《生命科学研究》
CAS
2024年第4期336-347,363,共13页
Life Science Research
基金
国家自然科学基金面上项目(81974080,82170640)
湖南省自然科学基金项目(2022JJ30954)。
关键词
肝细胞癌(HCC)
线粒体氧化应激
机器学习
预后
风险模型
hepatocellular carcinoma(HCC)
mitochondrial oxidative stress
machine learning
prognosis
risk model
