上海交通大学学报(医学版), 2024, 44(1): 1-12 doi: 10.3969/j.issn.1674-8115.2024.01.001

论著 · 基础研究

肝细胞癌相关的核编码线粒体基因及临床信息的综合预后模型

克德尔亚·艾山江,1,2, 傅怡2, 赖冬林2,3, 邬海龙,2,4, 龚伟,1,5

1.上海交通大学医学院附属新华医院普外科,上海 200092

2.上海健康医学院协同科研中心,上海 201318

3.南昌大学江西医学院第一附属医院,南昌 330006

4.上海健康医学院药学院,上海 201318

5.上海市胆道疾病研究重点实验室,上海交通大学医学院胆道疾病研究所,上海市胆道疾病研究中心,上海 200092

An integrated prognostic model of nuclear-encoded mitochondrial gene signature and clinical information for hepatocellular carcinoma

Aishanjiang Kedeerya,1,2, FU Yi2, LAI Donglin2,3, WU Hailong,2,4, GONG Wei,1,5

1.Department of General Surgery, Xinhua Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai 200092, China

2.Collaborative Innovation Center for Biomedicine, Shanghai University of Medicine and Health Sciences, Shanghai 201318, China

3.The First Affiliated Hospital, Jiangxi Medical College, Nanchang University, Nanchang 330006, China

4.School of Pharmacy, Shanghai University of Medicine and Health Sciences, Shanghai 201318, China

5.Shanghai Key Laboratory of Biliary Tract Disease Research, Research Institute of Biliary Tract Disease, Shanghai Jiao Tong University School of Medicine, Shanghai Research Center of Biliary Tract Disease, Shanghai 200092, China

通讯作者: 邬海龙,电子信箱:wuhl@sumhs.edu.cn龚 伟,电子信箱:gongwei@xinhuamed.com.cn

编委: 崔黎明

收稿日期: 2023-09-18   接受日期: 2023-12-06  

基金资助: 国家自然科学基金.  31870905.  82172628.  81974371
上海市卫生健康委员会基金.  201940352
上海市科学技术委员会基金.  22ZR1428100
上海交通大学医学院“双百人”项目.  20151001
上海交通大学医学院附属新华医院临床研究项目.  21XHDB10
上海市胆道疾病重点实验室研究基金.  17DZ2260200

Corresponding authors: WU Hailong, E-mail:wuhl@sumhs.edu.cnGONG Wei, E-mail:gongwei@xinhuamed.com.cn.

Received: 2023-09-18   Accepted: 2023-12-06  

作者简介 About authors

克德尔亚·艾山江(1995—),女,维吾尔族,硕士生;电子信箱:kadirya95@163.com。 E-mail:kadirya95@163.com

摘要

目的·建立一个基于线粒体基因和临床信息的肝细胞癌(hepatocellular carcinoma,HCC)总生存率(overall survival,OS)的预后模型。方法·从癌症基因组图谱(The Cancer Genome Atlas,TCGA)下载369例HCC患者和50例肝脏正常对照的基因表达谱和临床数据。核编码的线粒体基因(nuclear encoded mitochondrial gene,NEMG)从MitoCarta3.0数据库获得。使用“DEseq2”R包和单变量Cox分析选择与HCC患者OS相关并参与氧化磷酸化、三羧酸循环和细胞凋亡通路的NEMG[(泛素细胞色素C还原酶铰链蛋白(ubiquinol cytochrome C reductase hinge protein,UQCRH腺苷三磷酸柠檬酸裂解酶(ATP citrate lyase,ACLY磷酸烯醇式丙酮酸羧激酶2(phosphoenolpyruvate carboxykinase 2,PCK2Bcl-2同源拮抗剂1(Bcl-2 homologous antagonist/killer 1,BAK1Bcl-2相关X蛋白(Bcl-2-associated X protein,BAX)和Bcl-2/腺病毒E1B相互作用蛋白3样(Bcl-2/adenovirus E1B interacting protein 3-like,BNIP3L)]。应用多变量Cox回归来确定HCC OS的独立危险因素。建立一个基于独立危险因素(6个NEMG风险特征和TNM分期)的综合预后模型和预后列线图,计算中位预后评分。以中位预后评分作为分界点,将HCC患者分为高风险组和低风险组。进行Kaplan-Meier生存曲线分析,并进行对数秩检验来评估低风险组和高风险组患者OS的差异。使用“timeROC”软件包计算受试者操作特征(receiver operating characteristic,ROC)曲线下面积(area under the curve,AUC)。用基因表达数据库(Gene Expression Omnibus,GEO)下载HCC队列(GSE14520)验证综合预后模型对1、3、5年OS的预测性能。通过实时荧光定量聚合酶链反应(real-time quantitative polymerase chain reaction,qPCR)在来自上海交通大学医学院附属新华医院的34例HCC临床样本中验证6-NEMG的相对表达水平。结果·ROC分析结果显示,与仅6-NEMG风险特征(1、3、5年AUC分别为0.77、0.66、0.65)或仅TNM分期(1、3、5年AUC分别为0.66、0.67、0.63)相比,该综合预后模型对1年(AUC,0.78)、3年(AUC,0.73)和5年(AUC,0.69)HCC OS显示出更好的预测性能。Kaplan-Meier生存曲线分析结果显示高风险组患者的OS明显比低风险组差(P=0.001)。此外,在GEO外部队列中发现该预后模型的预测性能较好(1、3、5年AUC分别为0.67、0.66、0.74),高、低风险组患者的预后差异有统计学意义(P=0.001),与TCGA数据的结果一致。在临床HCC队列中,与癌旁肝脏组织相比,除BNIP3L外,其他5个NEMG在肿瘤组织的表达水平上调或者下调。相关性分析显示,在GSE14520与临床HCC队列中预后评分与HCC肿瘤的大小和数量呈正相关。结论·构建并验证了一个将6-NEMG风险特征与TNM分期相结合的HCC预后预测模型。该模型可能有助于HCC患者的预后预测。

关键词: 核编码线粒体基因 ; 肝细胞癌 ; 癌症基因组图谱(TCGA) ; 基因表达数据库(GEO) ; 总生存率

Abstract

Objective ·To establish a prognostic model for the overall survival (OS) of hepatocellular carcinoma (HCC) based on mitochondrial genes and clinical information. Methods ·The gene expression and the clinical data of 369 HCC patients and 50 controls with normal liver were downloaded from The Cancer Genome Atlas (TCGA) database. The nuclear-encoded mitochondrial genes (NEMGs) were obtained from the MitoCarta3.0 database. The "DESeq2" R package and univariate Cox analysis were used to select NEMGs [ubiquinol cytochrome C reductase hinge protein (UQCRH),ATP citrate lyase (ACLY),phosphoenolpyruvate carboxykinase 2 (PCK2), Bcl-2 homologous antagonist/killer1 (BAK1), Bcl-2-associated X protein (BAX) andBcl-2/adenovirus E1B interacting protein 3-like (BNIP3L)] in HCC that were associated with OS of HCC and participated in dysregulation of oxidative phosphorylation, tricarboxylic acid cycle and cell apoptosis. Multivariate Cox analysis was applied to select independent risk factors for OS of HCC. A comprehensive prognostic model and a prognostic nomogram with 6-NEMG risk characteristics and TNM staging were established. By using the median of prognostic scores as a cut-off, HCC patients were classified into low-risk and high-risk group. Kaplan-Meier survival curve analysis was conducted and log-rank test was performed to evaluate the survival rates between the low-risk and high-risk group. The area under the curve (AUC) values of receiver operating characteristic (ROC) curve were calculated via using the "timeROC" package. The prognostic model for HCC was validated by using the GEO HCC cohort (GSE14520) for 1, 3 and 5 years. Finally, the relative expression level of 6-NEMG was validated in 34 clinical samples of HCC from Xinhua Hospital, Shanghai Jiao Tong University School of Medicine by using real-time quantitative polymerase chain reaction (qPCR) method. Results ·Compared to 6-NEMG risk signature only (AUCs for 1, 3 and 5 years were 0.77, 0.66 and 0.65, respectively) or TNM stage only (AUCs for 1, 3 and 5 years were 0.66, 0.67 and 0.63, respectively), ROC curve analysis showed that this integrated prognostic model displayed better predictive performance for 1-year (AUC, 0.78), 3-year (AUC, 0.73) and 5-year (AUC, 0.69) OS of HCC. The Kaplan-Meier survival curve analysis showed that the OS of HCC patients in the high-risk group was significantly worse than that in the low-risk group (P=0.001). In addition, predictive performance of the prognostic model (AUC for 1, 3 and 5 years is 0.67, 0.66 and 0.74, respectively) and prognostic differences between the high-risk and low-risk group (P=0.001) were further validated in GEO (GSE14520) external cohort, and these results were consistent with the TCGA data. In addition to BNIP3L, dysregulation of five other NEMGs in the clinical HCC cohort was validated. The correlation analysis in GSE14520 and HCC clinical cohort showed a positive correlation between prognosis score and the size and number of tumors. Conclusion ·A new prognostic model that combines 6-NEMG risk characteristics with TNM staging for predicting OS in HCC patients was constructed and validated. This model may help improve the prognosis prediction of HCC patients.

Keywords: nuclear encoded mitochondrial gene ; hepatocellular carcinoma ; TCGA ; GEO ; overall survival

PDF (3051KB) 元数据 多维度评价 相关文章 导出 EndNote| Ris| Bibtex  收藏本文

本文引用格式

克德尔亚·艾山江, 傅怡, 赖冬林, 邬海龙, 龚伟. 肝细胞癌相关的核编码线粒体基因及临床信息的综合预后模型. 上海交通大学学报(医学版)[J], 2024, 44(1): 1-12 doi:10.3969/j.issn.1674-8115.2024.01.001

Aishanjiang Kedeerya, FU Yi, LAI Donglin, WU Hailong, GONG Wei. An integrated prognostic model of nuclear-encoded mitochondrial gene signature and clinical information for hepatocellular carcinoma. Journal of Shanghai Jiao Tong University (Medical Science)[J], 2024, 44(1): 1-12 doi:10.3969/j.issn.1674-8115.2024.01.001

在全球范围内,原发性肝癌是最具侵袭性和难以治疗的恶性肿瘤之一1。肝细胞癌(hepatocellular carcinoma,HCC)约占所有原发性肝癌病例的90%2。尽管早期和晚期HCC的治疗方法都有所改善3-4,但近几十年来,许多国家的发病率和死亡率仍然很高5。由于预后不良,HCC目前是全球第四大常见的癌症相关死亡原因,5年生存率低于21%6。因此,建立一个能够准确预测HCC预后的模型对于改善HCC预后至关重要。虽然许多分类系统,如美国癌症联合委员会(American Joint Committee on Cancer,AJCC)的TNM分期(American Joint Committee on Cancer-Tumor Node Metastasis Staging)系统、巴塞罗那临床肝癌(Barcelona Clinic Liver Cancer,BCLC)分期和意大利肝癌评分(Cancer of the Liver Italian Program,CLIP)系统,已广泛应用于临床评价HCC患者的预后,但这些分类系统主要是基于临床病理特征,没有考虑分子标志物的重要预后作用7

线粒体是一种双层膜细胞器,在真核细胞中作为动力源,在细胞存活、凋亡、代谢和信号转导中起着至关重要的作用8-9。近99%的线粒体蛋白质是由位于细胞核上的核编码线粒体基因(nuclear encoded mitochondrial gene,NEMG)编码的10。这些NEMG参与多种线粒体途径,包括氧化磷酸化(oxidative phosphorylation,OXPHOS)、细胞凋亡、三羧酸(tricarboxylic acid,TCA)循环、β氧化、酮体生成等11-13,对线粒体稳态至关重要。越来越多的证据表明,线粒体的功能障碍和失调参与癌症的发生和进展。线粒体DNA的存在对癌细胞的生长和致瘤性有着重要作用14。在化学治疗耐药的癌症干细胞中观察到增强的OXPHOS15-16。在许多癌症类型中,抑制OXPHOS可使癌细胞对化学治疗和靶向治疗重新敏感17-18。此外,TCA循环中一些酶的显性突变,如异柠檬酸脱氢酶2(isocitrate dehydrogenase 2,IDH2)19、琥珀酸脱氢酶(succinate dehydrogenase,SDH)20和延胡索酸水合酶(fumarate hydratase,FH)21与肿瘤的起始和进展有关。此外,既往研究22-23表明,NEMG参与HCC的发病和进展。尽管NEMG在HCC的发生发展中很重要,但基于NEMG的预后模型研究有限且不够成熟。因此,建立一个基于NEMG的预后模型可能对HCC的预后有很高的预测价值。

本研究基于生物信息学数据库和HCC临床样本,构建并验证了一个基于NEMG风险基因和临床特征的HCC预后模型,以期改善HCC患者的风险分级和生存预测。

1 资料与方法

1.1 数据收集

分别从癌症基因组图谱(The Cancer Genome Atlas,TCGA)数据库及基因表达数据库(Gene Expression Omnibus,GEO)中下载419例(369例HCC和50例非HCC样本)TCGA-LIHC(liver hepatocellular carcinoma)和247例HCC患者的基因表达和相应的临床信息。在移除没有随访记录和完整临床信息的样本后,把TCGA数据库中339例HCC样本和50例正常肝脏样本作为训练队列,GEO数据集(GSE14520)中219例HCC患者作为验证队列。从MitoCarta 3.0数据库(https://www.broadinstitute.org/mitocarta)下载哺乳动物线粒体蛋白和通路清单,共包含1 123个线粒体基因。

1.2 HCC中差异表达的核编码线粒体基因的分析

根据“DEseq2”R包筛选差异表达基因(differentially expressed gene,DEG),筛选标准:∣log2fold change(FC)∣>0.5,QP adjusted)<0.05。对TCGA数据库中369例HCC组织与50例正常肝脏组织的差异表达NEMG(differentially expressed nuclear encoded mitochondrial gene,deNEMG)进行分析,并将TCGA数据库中的deNEMG与验证队列GSE14520数据集中包含的基因取交集,共获得264个deNEMG以供进一步分析。

1.3 基于参与OXPHOSTCA和凋亡通路的deNEMG风险模型构建

采用单变量比例风险回归(Cox)方法识别与HCC总生存率(overall survival,OS)相关的deNEMG,筛选出参与3种重要途径(OXPHOS、TCA循环和细胞凋亡)并与OS相关的deNEMG,来构建基于NEMG的风险特征模型。该6-NEMG特征模型的风险评分采用线性公式计算:风险评分=∑coefficienti×genei(genei,个体HCC患者中deNEMG的相对表达水平;coefficienti,NEMG的多变量Cox回归系数)。

1.4 HCC的预后模型及列线图的建立及验证

对包括年龄、性别、甲胎蛋白(α-fetoprotein,AFP)和TNM分期等重要的临床病理特征以及deNEMG的风险特征通过“forestplot”和“survival”包进行单因素和多因素Cox回归分析,获得独立的预后因素。使用“rms”和“nomogramEx”软件包建立了一个基于独立预后因素的预后模型和预后列线图。根据该预后模型中涉及的独立预后因素的系数,制定公式来计算每个患者的预后评分。以预后评分的中位数作为临界值,将HCC患者分为高风险组和低风险组。采用“survival”和“survminer”包进行Kaplan-Meier生存曲线分析,并采用log-rank检验评估低风险组和高风险组之间的OS。使用“timeROC”软件包计算受试者操作特征(receiver operating characteristic,ROC)曲线下面积(area under the curve,AUC)。

1.5 临床样本收集

于2018年至2020年期间在上海交通大学医学院附属新华医院肝移植科收集34例行原发性肝切除术或肝移植术后HCC患者的癌和癌旁组织。所有病例均经病理确诊为HCC,术前均未接受放射治疗或化学治疗。

1.6 实时荧光定量.聚合酶链反应

用TRIzol试剂(Invitrogen,美国)从34对肿瘤和癌旁组织中提取总RNA。使用PrimeScript RT试剂盒(Takara,日本)进行反转录。采用SYBR PrimeScript RT-PCR试剂盒(Takara,日本)进行实时荧光定量聚合酶链反应(real-time quantitative polymerase chain reaction,qPCR)分析。引物序列见表1。采用公式2-ΔΔCT计算基因的相对表达水平。

表1   qPCR引物序列

Tab 1  Primer sequences for qPCR

PrimerForward sequence (5′→3′)Reverse sequence (5′→3′)
UQCRHGCTCTGTGATGAGCGTGTATCCGTTGTTAAAGAGTTTGTGGGCCAC
ACLYGCTCTGCCTATGACAGCACCATGTCCGATGATGGTCACTCCCTT
PCK2TAGTGCCTGTGGCAAGACCAACGAAGCCGTTCTCAGGGTTGATG
BAK1TTACCGCCATCAGCAGGAACAGGGAACTCTGAGTCATAGCGTCG
BAXTCAGGATGCGTCCACCAAGAAGTGTGTCCACGGCGGCAATCATC
BNIP3LTGTGGAAATGCACACCAGCAGGCTACTGGACCAGTCTGATACCC
18SGGAGAGGGAGCCTGAGAAACGTTACAGGGCCTCGAAAGAGTCC

Note:UQCRH—ubiquinol cytochrome C reductase hinge protein; ACLY—ATP citrate lyase; PCK2—phosphoenolpyruvate carboxykinase 2; BAK1—Bcl-2 homologous antagonist/killer1; BAX—Bcl-2-associated X protein; BNIP3L—Bcl2/adenovirus E1B interacting protein 3-like; 18S—18S ribosomal RNA.

新窗口打开| 下载CSV


1.7 统计学分析

所有数据均使用R软件包(4.2.0版)或GraphPad Prism软件(9.4.1版)进行统计分析。定量资料以x±s表示。定性资料采用n(%)表示。采用“corrplot”包进行预后评分与HCC临床信息之间的相关性分析。采用Fisher检验比较分类变量(性别、年龄)。采用对数秩检验的Kaplan-Meier曲线分析,评价不同风险组OS的差异。P<0.05表示差异具有统计学意义。

2 结果

2.1 参与3种重要途径且与预后相关的NEMG筛选

为了鉴定与HCC进展相关的NEMG,使用“DEseq2”软件包在TCGA-LIHC队列中鉴定了369个deNEMG(|log2FC|>0.5,P<0.05)(图1A)。在369个deNEMG中,在GEO验证队列GSE14520中可检测到264个deNEMG(102个上调,162个下调)(图1B、C)。TCGA-LIHC队列中HCC患者OS的单变量Cox回归分析显示,95个deNEMG与OS显著相关(P<0.05)。考虑到OXPHOS、TCA循环和细胞凋亡途径对HCC进展很重要,取95个deNEMG和参与3个通路的NEMG的交集。选择6个deNEMG[泛素细胞色素C还原酶铰链蛋白(ubiquinol cytochrome C reductase hinge protein,UQCRH腺苷三磷酸柠檬酸裂解酶(ATP citrate lyase,ACLY磷酸烯醇式丙酮酸羧激酶2(phosphoenolpyruvate carboxykinase 2,PCK2Bcl-2同源拮抗剂1(Bcl-2 homologous antagonist/killer 1,BAK1Bcl-2相关X蛋白(Bcl-2-associated X protein,BAX)和Bcl-2/腺病毒E1B相互作用蛋白3样(Bcl-2/adenovirus E1B interacting protein 3-like,BNIP3L)]来构建基于NEMG的风险特征模型(图1D)。通过多变量Cox回归分析得到6个基因的相关系数(图1E)。经计算,6-NEMG特征的风险评分=0.517 3×UQCRH的基因表达水平+0.307 4×ACLY的基因表达水平+0.135 3×PCK2的基因表达水平+0.153 3×BAK1的基因表达水平+0.020 1×BAX的基因表达水平+0.193 7×BNIP3L的基因表达水平。

图1

图1   预后相关NEMG的识别和基于6-NEMG的风险特征模型的构建

Note: A. DEGs between HCC and adjacent paired normal tissues. Red dots represent significant upregulation and blue dots represent significant downregulation of DEGs in HCC tissues. B. Heatmap of 264 deNEMGs in TCGA cohort. N—adjacent paired normal tissues; T—HCC tumor tissues. C. Venn diagram showing differently expressed deNEMGs between tumor and adjacent tissues. D. Univariate Cox regression analysis of 6 significant prognostic NEMGs associated with OXPHOS, TCA cycle and cell apoptosis. E. Multivariate Cox analysis of 6-NEMG risk signature.

Fig 1   Identification of prognosis-related NEMGs and construction of a 6-NEMG-based risk signature


2.2 基于TCGA队列的HCC预后模型构建

通过对HCC预后相关NEMG进行单因素和多因素Cox分析,发现TNM分期和6-NEMG特征的风险评分是HCC OS的独立预后因素(图2A、B)。基于此本研究构建了一个整合了TNM分期和6-NEMG特征的预后模型。采用如下公式计算该综合预后模型的预后评分:预后评分=∑风险评分×系数R+TNMi×系数T(风险评分指的是6-NEMG风险特征在个体HCC患者的风险评分;TNMi为个体HCC患者的TNM分期;系数R为风险评分的多变量Cox回归系数;系数T为TNM分期的多变量Cox回归系数)。经计算,预后评分=0.888×风险评分+0.418×TNM。

图2

图2   6-NEMG风险特征与TNM分期相结合的预后模型的构建

Note: A/B. Univariate (A) and multivariate (B) Cox analysis of risk score and clinicopathological features for OS of HCC. C. ROC analysis for the prognostic model at 1-, 3- and 5-year survival in the TCGA-LIHC cohort. D. Kaplan-Meier survival analysis of OS in the high- and low-risk group in the TCGA-LIHC cohort.

Fig 2   Construction of a prognostic model integrating the 6-NEMG risk signature with TNM stage


时间依赖的ROC曲线显示,该综合预后模型对术后1年(AUC=0.78)、3年(AUC=0.73)和5年(AUC=0.69)预后表现出良好的预测性能(图2C)。利用预后评分的中位数作为临界值,将TCGA-LIHC队列(表2)中339例临床信息完整的HCC患者分为高风险组(n=169)和低风险组(n=170)。Kaplan-Meier曲线显示,高风险组患者的OS明显低于低风险组患者(P=0.001,图2D)。

表2   不同风险组患者的基本特征

Tab 2  Baseline characteristics of the patients in different risk groups

FeatureTCGA (n=339)GEO (n=219)
High-risk groupLow-risk groupP valueHigh-risk groupLow-risk groupP value
Age/n(%)0.6810.969
≤50 years36 (21.30)39 (22.94)53 (11.93)58 (52.73)
>50 years133 (78.73)131 (77.06)56 (88.07)52 (47.27)
Gender/n(%)0.2540.156
Male107 (63.31)124 (72.94)96 (88.07)93 (84.55)
Female62 (36.69)46 (27.06)13 (11.93)17 (15.45)
Stage/n(%)0.0000.000
Ⅰ+Ⅱ89 (52.66)165 (97.06)65 (59.63)105 (95.45)
Ⅲ+Ⅳ80 (47.34)5 (2.94)44 (40.37)5 (4.55)
AFP/n(%)0.0310.045

≤20 ng·mL-1 (TCGA)

or ≤300 ng·mL-1 (GEO)

118 (69.82)111 (65.29)57 (52.29)61 (55.45)

>20 ng·mL-1 (TCGA)

or >300 ng·mL-1 (GEO)

51 (30.18)59 (34.71)52 (47.71)49 (44.55)
Event/n(%)
Survival92 (54.44)131 (77.06)57 (52.29)78 (70.91)
Death77 (45.56)39 (22.94)52 (47.71)32 (29.09)

新窗口打开| 下载CSV


2.3 基于TCGA队列的HCC预后预测模型列线图的建立

在基于TCGA队列进行基因筛选以及建立预后预测模型之后,进一步验证该模型在预测HCC患者预后的可行性。ROC曲线分析结果表明,在预测HCC预后方面,预后评分优于仅风险评分或仅TNM分期(图3A~C)。

图3

图3   基于TCGA队列的HCC预后预测模型列线图的建立

Note: A‒C. ROC analysis of the prognostic score, risk score and TNM stage for 1-year OS (A), 3-year OS (B) and 5-year OS (C). D. Nomogram constructed by the prognostic score to predict OS of HCC. E. Calibration curves of the prognostic score-derived nomogram for 1-year, 3-year and 5-year OS of HCC.

Fig 3   Establishment of the prognostic score-derived nomogram based on TCGA cohort


为了验证该预后模型的预测效能,根据预后评分构建了预后列线图(C指数为0.689),见图3D。校准图显示,预测列线图预测的预后与实际OS非常接近(图3E)。这也进一步表明基于NEMG和TNM分期建立的HCC预后模型具有较高的可信度。

2.4 建立的HCC综合预后模型在GEO队列和临床样本中的验证

为了验证预测模型的稳定性,在GEO数据集GSE14520(表2)中对其进行进一步验证。根据预后评分的中位数,GSE14520中的HCC患者被分为高风险组(n=109)和低风险组(n=110)。时间ROC曲线分析证实了该预后模型对GSE14520中HCC OS良好的预测性能,在1、3和5年OS的AUC值分别为0.67、0.66和0.74(图4A)。Kaplan-Meier曲线还显示,与低风险组患者相比,高风险组患者的预后明显较差(图4B,P=0.001)。

图4

图4   GEO和临床HCC队列中预后模型的验证

Note: A. ROC analysis of the prognostic model for OS of HCC in GSE14520. B. Kaplan-Meier analysis of OS in HCC patients of GSE14520 which were stratified into high- and low-risk groups by this prognostic model. C. Correlation analysis between clinicopathological features and prognostic score in HCC patients of GSE14520. The correlation coefficient in red indicates P<0.05. D. Correlation analysis between clinicopathological features and prognostic score in 34-pair HCC and matched normal tissues. The correlation coefficient in red indicates P<0.05. E. Gene expression levels of the 6 NEMGs involved in this prognostic model in 34-pair HCC tumor tissues (T) and matched normal tissues (N).

Fig 4   Validation of the prognostic model in GEO and clinical HCC cohort


相关性分析显示,预后评分与一些重要的临床病理特征如主要肿瘤大小和CLIP分期呈正相关(P=0.000,P=0.000,图4C)。此外,在34对HCC癌和癌旁组织的队列(表3)中,患者临床信息与预后评分之间的相关性分析也表明,预后评分与肿瘤大小和肿瘤数量呈正相关(图4D),与GEO数据集中的分析结果相一致。随后,在上述的34对HCC肿瘤样本中验证了预后模型中6种NEMGs(UQCRHACLYPCK2BAK1BAXBNIP3L)的相对表达水平。与癌旁组织相比,HCC组织中的UQCRHP=0.010)、ACLYP=0.000)、BAXP=0.001)和BAK1P=0.000)的表达水平显著上调(图4E),而PCK2的表达水平显著下调(P=0.007,图4E),这与TCGA-LIHC队列中的各基因的表达趋势相一致。但BNIP3L的表达水平在癌和癌旁组织之间差异没有统计学意义(P=0.974,图4E)。

表3   临床队列中HCC患者的临床病理特征

Tab 3  Clinicopathological features of HCC patients in the clinical cohort

FeatureTotal/n(%)
Age
≤50 year8 (23.53)
>50 year26 (76.47)
Gender
Male24 (70.59)
Female10 (29.41)
TNM stage
15 (44.12)
5 (14.71)
11 (32.35)
3 (8.82)
AFP
≤20 ng·mL-121 (61.76)
>20 ng·mL-113 (38.24)
Size
≤5 cm20 (58.82)
>5 cm14 (41.18)
Tumor number
=122 (64.71)
>112 (35.29)
HBsAg
-7 (20.59)
+27 (79.41)
GPT
≤40 U·L-125 (73.53)
>40 U·L-19 (26.47)
Cirrhosis
-16 (40.06)
+18 (59.94)

Note: HBsAg—hepatitis B surface antigen; GPT—glutamic pyruvic transaminase.

新窗口打开| 下载CSV


3 讨论

由于HCC患者的异质性结果以及管理和治疗策略选择的困难性,迫切需要一个有效的HCC患者预后预测指标。临床前研究的证据支持线粒体功能障碍是代谢性肝病和癌症发病机制的关键因素,这进一步表明开发线粒体基因靶向治疗是抑制HCC进展的一种有吸引力的策略24。研究25证明,异常的线粒体能量状态有助于癌症的发生和发展。线粒体TCA循环酶缺陷可产生肿瘤代谢产物,促进肿瘤发生。线粒体功能障碍是由mtDNA突变、TCA循环酶缺陷、电子呼吸链泄漏和随后的氧化应激引起的;致癌和抑癌信号可以改变细胞代谢途径,破坏氧化还原的平衡,从而导致人类癌症的发生26。NEMG的突变或异常表达可能导致与肿瘤发生相关的异常能量产生。因此建立基于NEMG的HCC患者预后模型对HCC患者的进展及预后至关重要。

本研究着重研究OXPHOS、TCA循环和细胞凋亡这3个通路中的预后相关基因,对95个与HCC预后相关的deNEMG和参与3种有丝分裂通路的NEMG取交集,发现仅6个参与3种有丝分裂通路的deNEMG(UQCRHACLYPCK2BAK1BAXBNIP3L)与HCC的预后密切相关。因此本研究利用这6个基因来构建基于NEMG的风险模型。在这6个NEMG中,UQCRH编码一个定位于线粒体和细胞核的铰链蛋白。UQCRH作为泛素-细胞色素C还原酶复合物(也称为OXPHOS通路的复合物Ⅲ)的主要亚基之一,负责复合物Ⅲ的细胞色素C和C1之间的电子转移。事实上,在肺腺癌27和HCC28中已经有关于UQCRH表达上调的报道。此外,UQCRH的表达与肿瘤大小、分化程度和有无血管侵袭相关,并与HCC患者的不良预后密切相关28。ACLY和PCK2参与TCA循环,其中ACLY形成一种细胞质同源四聚体酶,催化TCA循环中的中间体柠檬酸盐转化为草酰乙酸和乙酰辅酶A29。由于乙酰辅酶A是新脂肪生成和胆固醇生物合成的重要底物之一,因此ACLY是一种重要的脂肪生成酶,在结直肠癌、非小细胞肺癌、乳腺癌、膀胱癌和HCC等癌症类型中高表达30-32。同时,在HCC患者中,ACLY的高表达与不良预后、肿瘤干性和转移相关33PCK2编码一种线粒体磷酸烯醇式丙酮酸羧激酶,该酶催化TCA循环衍生的草酰乙酸转化为磷酸烯醇式丙酮酸,并伴随GTP水解。PCK2在某些癌症类型,如肺癌、前列腺癌、乳腺癌、宫颈癌和睾丸癌中经常上调和激活34-35,可能以此对抗肿瘤微环境中的代谢饥饿状态36。此外,既往研究37表明,PCK2的下调与HCC的预后和患癌时间有关,并且PCK2的异位过表达可能通过诱导HCC细胞的葡萄糖缺乏而降低细胞活力。BAXBAK1BNIP3L是凋亡相关基因,其中BAX和BAK1是线粒体外膜通透性的决定性关键因素38-40。BNIP3L是线粒体吞噬的双重调节因子,而线粒体吞噬是清除受损线粒体的过程41。据报道BNIP3L与BAX和BAK相互作用,导致线粒体外膜通透性增强和凋亡加快42。同时,BNIP3L介导的线粒体吞噬通过糖酵解代谢重编程,促进乙型肝炎病毒X蛋白(hepatitis B virus X protein,HBx)诱导的HCC的肿瘤干性42

TCGA-LIHC队列的单因素和多因素Cox分析结果表明,6-NEMG风险特征和TNM分期是HCC的独立危险因素。因此,将6-NEMG风险特征与TNM分期相结合,建立了一个新的HCC综合预后模型。该模型在TCGA-LIHC和GSE14520队列中预测HCC预后方面均表现出良好的性能。利用预后评分的中位数作为临界值,将TCGA-LIHC和GSE14520队列患者分为高风险和低风险组,进行了预后差异分析,结果表明在2个队列中高风险组的OS均明显低于低风险组患者。此外,在34对HCC临床样本(癌和癌旁组织)中验证这6个NEMG的表达情况。在HCC肿瘤样本中UQCRHACLYBAK1BAX相对高表达,而PCK2相对低表达,这符合在TCGA-LIHC数据库中的分析结果;但BNIP3L在癌组织和癌旁组织的表达差异没有统计学意义,可能跟临床样本量有限有关。这些验证进一步表明6个NEMG在HCC的预后预测中起着重要的作用。

本研究用TCGA数据建模,用GEO数据验证的原因在于:TCGA数据库收录的基因表达数据多为通过测序手段获得的,而GEO数据库收录的大部分数据是用芯片技术获得的;前者测序技术挖掘到的基因比后者多,并且临床信息比较全,用能测到更多基因的数据库(TCGA)去训练,可避免模型的重要基因被遗漏。此外,本研究存在以下局限性。首先,TCGA和GEO数据库中有些病例的临床特征不完整,因此仅选择了与肿瘤预后相关并且在2个数据库中均包含的临床特征进行分析。其次,临床HCC队列的样本量不足,临床信息不完整,未能充分验证该预后模型在临床HCC样本中的预测性能。因此,在将该HCC预后模型扩展到临床应用之前,还需要进行进一步的临床试验。此外,模型中包含的基因的功能也需要通过细胞实验和动物实验模型来进一步验证。

作者贡献声明

本研究涉及的所有实验均已通过上海交通大学医学院附属新华医院医学伦理委员会审核批准(文件号XHEC-D-2023-021)。受试对象或其亲属已经签署知情同意书。

AUTHOR's CONTRIBUTIONS

All experimental protocols in this study were reviewed and approved by Ethics Committee of Xinhua Hospital, Shanghai Jiao Tong University School of Medicine (Approval Letter No. XHEC-D-2023-021). Consent letters have been signed by the research participants or their relatives.

利益冲突声明

所有作者声明不存在利益冲突。

COMPETING INTERESTS

All authors disclose no relevant conflict of interests.

参考文献

MILLER K D, FIDLER-BENAOUDIA M, KEEGAN T H, et al. Cancer statistics for adolescents and young adults, 2020[J]. CA Cancer J Clin, 2020, 70(6): 443-459.

[本文引用: 1]

SIEGEL R L, MILLER K D, FUCHS H E, et al. Cancer statistics, 2021[J]. CA A Cancer J Clinicians, 2021, 71(1): 7-33.

[本文引用: 1]

SANGRO B, SAROBE P, HERVÁS-STUBBS S, et al. Advances in immunotherapy for hepatocellular carcinoma[J]. Nat Rev Gastroenterol Hepatol, 2021, 18(8): 525-543.

[本文引用: 1]

RIMASSA L, PERSONENI N, CZAUDERNA C, et al. Systemic treatment of HCC in special populations[J]. J Hepatol, 2021, 74(4): 931-943.

[本文引用: 1]

BERTUCCIO P, TURATI F, CARIOLI G, et al. Global trends and predictions in hepatocellular carcinoma mortality[J]. J Hepatol, 2017, 67(2): 302-309.

[本文引用: 1]

YANG J D, HAINAUT P, GORES G J, et al. A global view of hepatocellular carcinoma: trends, risk, prevention and management[J]. Nat Rev Gastroenterol Hepatol, 2019, 16(10): 589-604.

[本文引用: 1]

ZHU J Y, TANG B F, LI J, et al. Identification and validation of the angiogenic genes for constructing diagnostic, prognostic, and recurrence models for hepatocellular carcinoma[J]. Aging, 2020, 12(9): 7848-7873.

[本文引用: 1]

DING Z B, ERICKSEN R E, LEE Q Y, et al. Reprogramming of mitochondrial proline metabolism promotes liver tumorigenesis[J]. Amino Acids, 2021, 53(12): 1807-1815.

[本文引用: 1]

MARQUEZ J, FLORES J, KIM A H, et al. Rescue of TCA cycle dysfunction for cancer therapy[J]. J Clin Med, 2019, 8(12): 2161.

[本文引用: 1]

SONG B S, MOON J S, TIAN J W, et al. Mitoribosomal defects aggravate liver cancer via aberrant glycolytic flux and T cell exhaustion[J]. J Immunother Cancer, 2022, 10(5): e004337.

[本文引用: 1]

LI M, WANG L, WANG Y J, et al. Mitochondrial fusion via OPA1 and MFN1 supports liver tumor cell metabolism and growth[J]. Cells, 2020, 9(1): 121.

[本文引用: 1]

LI X J, LI Y M, XU A J, et al. Apoptosis-induced translocation of centromere protein F in its corresponding autoantibody production in hepatocellular carcinoma[J]. Oncoimmunology, 2021, 10(1): 1992104.

CAI Y L, LIN Y X, XIONG X Z, et al. Knockdown expression of MECR, a novel gene of mitochondrial FAS Ⅱ inhibits growth and colony-formation, promotes apoptosis of hepatocelluar carcinoma cells[J]. Biosci Trends, 2019, 13(3): 234-244.

[本文引用: 1]

WALLACE D C. Mitochondria and cancer[J]. Nat Rev Cancer, 2012, 12(10): 685-698.

[本文引用: 1]

SANCHO P, BARNEDA D, HEESCHEN C. Hallmarks of cancer stem cell metabolism[J]. Br J Cancer, 2016, 114(12): 1305-1312.

[本文引用: 1]

KUNTZ E M, BAQUERO P, MICHIE A M, et al. Targeting mitochondrial oxidative phosphorylation eradicates therapy-resistant chronic myeloid leukemia stem cells[J]. Nat Med, 2017, 23(10): 1234-1240.

[本文引用: 1]

VAZQUEZ F, LIM J H, CHIM H, et al. PGC1α expression defines a subset of human melanoma tumors with increased mitochondrial capacity and resistance to oxidative stress[J]. Cancer Cell, 2013, 23(3): 287-301.

[本文引用: 1]

FARGE T, SALAND E, DE TONI F, et al. Chemotherapy-resistant human acute myeloid leukemia cells are not enriched for leukemic stem cells but require oxidative metabolism[J]. Cancer Discov, 2017, 7(7): 716-735.

[本文引用: 1]

ABBAS S, LUGTHART S, KAVELAARS F G, et al. Acquired mutations in the genes encoding IDH1 and IDH2 both are recurrent aberrations in acute myeloid leukemia: prevalence and prognostic value[J]. Blood, 2010, 116(12): 2122-2126.

[本文引用: 1]

PASINI B, MCWHINNEY S R, BEI T, et al. Clinical and molecular genetics of patients with the Carney-Stratakis syndrome and germline mutations of the genes coding for the succinate dehydrogenase subunits SDHB, SDHC, and SDHD[J]. Eur J Hum Genet, 2008, 16(1): 79-88.

[本文引用: 1]

LEHTONEN H J, BLANCO I, PIULATS J M, et al. Conventional renal cancer in a patient with fumarate hydratase mutation[J]. Hum Pathol, 2007, 38(5): 793-796.

[本文引用: 1]

FU J, LIU G X, ZHANG X, et al. TRPM8 promotes hepatocellular carcinoma progression by inducing SNORA55 mediated nuclear-mitochondrial communication[J]. Cancer Gene Ther, 2023, 30(5): 738-751.

[本文引用: 1]

HUANG Q C, WU D, ZHAO J, et al. TFAM loss induces nuclear actin assembly upon mDia2 malonylation to promote liver cancer metastasis[J]. EMBO J, 2022, 41(11): e110324.

[本文引用: 1]

CHELLA KRISHNAN K, KURT Z, BARRERE-CAIN R, et al. Integration of multi-omics data from mouse diversity panel highlights mitochondrial dysfunction in non-alcoholic fatty liver disease[J]. Cell Syst, 2018, 6(1): 103-115.e7.

[本文引用: 1]

WALLACE D C. Mitochondria and cancer[J]. Nat Rev Cancer, 2012, 12(10): 685-698.

[本文引用: 1]

LUO Y D, MA J J, LU W Q. The significance of mitochondrial dysfunction in cancer[J]. Int J Mol Sci, 2020, 21(16): 5598.

[本文引用: 1]

GAO F, LIU Q C, LI G P, et al. Identification of ubiquinol cytochrome c reductase hinge (UQCRH) as a potential diagnostic biomarker for lung adenocarcinoma[J]. Open Biol, 2016, 6(6): 150256.

[本文引用: 1]

PARK E R, KIM S B, LEE J S, et al. The mitochondrial hinge protein, UQCRH, is a novel prognostic factor for hepatocellular carcinoma[J]. Cancer Med, 2017, 6(4): 749-760.

[本文引用: 2]

SUN H, WANG F C, HUANG Y Q, et al. Targeted inhibition of ACLY expression to reverse the resistance of sorafenib in hepatocellular carcinoma[J]. J Cancer, 2022, 13(3): 951-964.

[本文引用: 1]

GRANCHI C. ATP citrate lyase (ACLY) inhibitors: an anti-cancer strategy at the crossroads of glucose and lipid metabolism[J]. Eur J Med Chem, 2018, 157: 1276-1291.

[本文引用: 1]

WEI J, LEIT S, KUAI J, et al. An allosteric mechanism for potent inhibition of human ATP-citrate lyase[J]. Nature, 2019, 568(7753): 566-570.

ICARD P, WU Z R, FOURNEL L, et al. ATP citrate lyase: a central metabolic enzyme in cancer[J]. Cancer Lett, 2020, 471: 125-134.

[本文引用: 1]

HAN Q, CHEN C A, YANG W, et al. ATP-citrate lyase regulates stemness and metastasis in hepatocellular carcinoma via the Wnt/β-catenin signaling pathway[J]. Hepatobiliary Pancreat Dis Int, 2021, 20(3): 251-261.

[本文引用: 1]

LEITHNER K, HRZENJAK A, TRÖTZMÜLLER M, et al. PCK2 activation mediates an adaptive response to glucose depletion in lung cancer[J]. Oncogene, 2015, 34(8): 1044-1050.

[本文引用: 1]

HYROŠŠOVÁ P, ARAGÓ M, MORENO-FELICI J, et al. PEPCK-M recoups tumor cell anabolic potential in a PKC-ζ-dependent manner[J]. Cancer Metab, 2021, 9(1): 1.

[本文引用: 1]

BLUEMEL G, PLANQUE M, MADREITER-SOKOLOWSKI C T, et al. PCK2 opposes mitochondrial respiration and maintains the redox balance in starved lung cancer cells[J]. Free Radic Biol Med, 2021, 176: 34-45.

[本文引用: 1]

LIU M X, JIN L, SUN S J, et al. Metabolic reprogramming by PCK1 promotes TCA cataplerosis, oxidative stress and apoptosis in liver cancer cells and suppresses hepatocellular carcinoma[J]. Oncogene, 2018, 37(12): 1637-1653.

[本文引用: 1]

COSENTINO K, HERTLEIN V, JENNER A, et al. The interplay between BAX and BAK tunes apoptotic pore growth to control mitochondrial-DNA-mediated inflammation[J]. Mol Cell, 2022, 82(5): 933-949.e9.

[本文引用: 1]

LUO X, O'NEILL K L, HUANG K. The third model of Bax/Bak activation: a Bcl-2 family feud finally resolved?[J]. F1000Research, 2020, 9: 935.

HAO B B, LI X J, JIA X L, et al. The novel cereblon modulator CC-885 inhibits mitophagy via selective degradation of BNIP3L[J]. Acta Pharmacol Sin, 2020, 41(9): 1246-1254.

[本文引用: 1]

LI Y, ZHENG W Q, LU Y Y, et al. BNIP3L/NIX-mediated mitophagy: molecular mechanisms and implications for human disease[J]. Cell Death Dis, 2021, 13(1): 14.

[本文引用: 1]

CHEN Y Y, WANG W H, CHE L, et al. BNIP3L-dependent mitophagy promotes HBx-induced cancer stemness of hepatocellular carcinoma cells via glycolysis metabolism reprogramming[J]. Cancers, 2020, 12(3): 655.

[本文引用: 2]

/