【作者】 吴昊；

【导师】 秦成勇；

【作者基本信息】 山东大学 ， 内科学（消化系病）， 2020， 博士

【摘要】 研究背景和目的肝癌是全球范围内最常见的恶性肿瘤之一,其发病率高居第六位,死亡率在所有癌症中排名第四位。肝细胞肝癌(Hepatocellularcarcinoma,HCC)是最常见的肝癌类型,占所有原发性肝癌的85-90%,每年有超过60万的人死于HCC,而中国则占其中的50%以上。虽然目前对于肝癌的危险因素、早期诊断、监测方法和治疗技术方面取得了重大进展。但由于其起病隐匿、生长迅速、恶性程度高、且易发生侵袭转移,超过70%的患者在被诊断为HCC时,其疾病的进展程度已经不适合再接受根治性肝切除或肝移植治疗,导致其预后极差,5年生存率低于15%。而判断肝癌患者的预后对于指导患者的个体化治疗,延长患者的总生存期、提高远期治疗效果意义重大。因此,临床上通过使用临床信息和病理学标准建立了各种分期系统,进而预测HCC预后并提供治疗指导。目前,已有的并开展临床使用的预测手段包括甲胎蛋白(Alpha fetoprotein,AFP)、肿瘤淋巴结转移(Tumor-node-metastasis,TNM)分期、Child-Pugh分级等。尽管这些分期系统已被证明有一定的应用价值,但由于未涉及HCC的生物学特征以及肿瘤的遗传和表观遗传学改变等,它们的预测效率仍然有限。长链非编码RNA(Longnon-coding RNA,lncRNA)是一系列转录本大于200个核苷酸且不具备蛋白质翻译功能的RNA分子。但其可在多层面上(表观遗传学、转录调控及转录后调控等)调控靶基因的表达,进而参与到多种关键的生物学过程。并且越来越清楚的是,lncRNA的功能取决于其独特的亚细胞定位。目前,大量的研究数据证实,lncRNA可以通过参与干扰肝癌细胞的细胞周期、侵袭转移、肿瘤微环境中的血管生成、免疫逃逸等多个生物学过程进而调控HCC的发生发展。此外,lncRNA具有重要的预测包括HCC在内的癌症患者的生存能力,提示lncRNA具有进一步指导个体化临床管理的潜在价值。因此,有必要进一步探讨lncRNA在HCC发生发展中的具体作用机制及临床意义。随着高通量技术的发展,公共数据库的出现使研究人员能够用来探究并鉴定潜在的生物标志物,以进行癌症的诊断和预后预测。癌症基因组图谱(The Cancer Genome Atlas,TCGA)计划旨在通过大规模的基因组测序和集成的多维分析,对人类肿瘤中主要的致癌基因组进行全面探索,为研究者提供了利用高通量数据和临床特征进行统计分析的巨大机会。目前,已有少量研究通过该数据库,获取HCC患者的RNA测序(RNA-Sequence,RNA-Seq)数据以及相应的临床资料,构建了有效的预测HCC患者生存的预测模型。但相关研究仅基于数据库资料,均未进行临床的实际验证和应用。此外,TCGA数据库中使用的转录谱分析和RNA测序方法昂贵且难以在医院执行。并且,现有研究也没有将LncRNA模型的预测效能与其他临床预测模型进行比较。因此,针对目前的研究现状,我们提出了以下问题:基于TCGA数据库筛选并用于构建模型的lncRNA在临床患者中的表达情况如何?RNA-Seq的结果是否可通过简单的实时定量聚合酶链反应(quantitative real-time polymerase chain reaction,qRT-PCR)检测所替代?模型在实际临床患者中的预测效能如何?模型的预测效能是否优于目前已有的临床模型?模型中lncRNA调控HCC的具体作用机制如何?基于以上问题,在本研究中,首先基于TCGA数据库,分析差异表达的lncRNA,并通过分析lncRNA与患者预后的关系,建立了基于lncRNA的预测模型,进而评估其在不同病因HCC患者中的预测效能。我们发现,该预测模型在乙肝相关的HCC患者中具有最佳的预测效率。基于此结果,进一步收集临床进行肝切除术的乙肝相关HCC患者的肝组织及临床资料,并进行随访,通过qRT-PCR检测了 HBV-HCC患者的lncRNA的表达水平,并首次建立了基于qRT-PCR数据的lncRNA预测模型。接下来,通过列线图(Nomogram)的方法拟对lncRNA模型进行优化。此外,进一步将模型的预测效能与其他临床分期系统:TNM,Child-Pugh,Okuda 分期系统(Okudastagingsystem),巴塞罗那分期(Barcelona-Clinic Liver Cancer,BCLC),意大利肝癌计划(Cancerof Liver Italian Program,CLIP)评分和香港中文大学预后指数(Chinese University Prognostic Index,CUPI)分级相比。随后,我们进行体外实验,明确了这构成模型的lncRNA在细胞系中的表达和亚细胞定位。并通过生物信息学分析得到lncRNA调控HCC发生发展的潜在作用机制。值得注意的是,lncRNA DDX11反义RNA1(DDX11 antisense RNA 1DDX11-AS1)在我们所构建的调控网络中处于“中心位置”,因此我们选取lncRNA DDX1 1-AS1为进一步研究对象,并进行基因富集分析、体内外实验来深入探究其调控HCC的发生发展的作用机制。本研究为HCC患者的生存预测和个体治疗,建立了有效的lncRNA模型。此外,对构成模型的lncRNA进行了深入探讨,lncRNA可作为HCC患者预后的生物标志物,有望成为临床治疗的有效靶点。第一部分通过TCGA数据库构建预测肝细胞肝癌患者生存的LncRNA模型目的通过对TCGA数据库中HCC患者RNA-Seq数据进行下载和分析,筛选并构建预测HCC患者生存的lncRNA预后模型。并对模型的预测效能及其对不同危险因素的HCC患者的预测效能进行比较和评估。方法1.通过TCGA数据库下载374例HCC患者肝组织及50例正常对照组的RNA-Seq表达谱数据,以及HCC患者相应临床资料,通过Perl以及R软件中“edgeR”包,提取并筛选出差异表达的lncRNA。2.基于差异表达的lncRNA以及患者的生存状态和生存时间,进行单因素及多因素Cox 比例风险回归分析,从中筛选对HCC患者有独立预后作用的lncRNA,并应用R语言构建预后风险预测模型。3.应用lncRNA模型,对370例具有临床预后信息的HCC患者进行风险评分计算,并应用中位值将患者分为高风险组和低风险组,进一步通过K-M分析及ROC曲线来评估风险预测模型的预测效能。4.对肝细胞癌患者进行基于病因的亚组分类,并根据风险评分将每个亚组患者进行分组,并通过K-M(Kaplan-Meier)分析及受试者工作曲线(Receiver operator characteristic curve,ROC),评估风险预测模型对不同危险因素患者预后的预测效能。结果1.通过TCGA数据库获取了 374例HCC患者肝组织及50例正常对照组的RNA-Seq表达谱数据及相关临床信息,并提取了 lncRNA表达信息,通过差异表达分析,得到1029个上调的lncRNA和58个下调的lncRNA。2.通过单因素和多因素Cox分析,筛选出了能独立预测HCC患者预后的lncRNA,并构建了一个基于5个lncRNA的风险预测模型:Risk score=(0.0448× LINC01116)+(0.2504 × DDX11-AS1)+(0.1001 × C10orf91)+(0.0722 × LUCAT1)+(0.1138 × FIRRE)。3.根据风险评分模型以及中位值,将370名具有生存时间的的HCC患者分为了高风险组和低风险组。K-M曲线显示两组患者生存显著不同。此外,上述模型预测患者3年生存的曲线下面积(Area Under Curve,AUC)为0.710,显示具有较好的预测效能。4.根据TCGA临床数据库中HCC患者的危险因素,对我们纳入的370名有预后信息的患者进行分组:包括饮酒导致的HCC,乙型肝炎病毒相关肝细胞癌(hepatitis B virus-associated hepatocellularcarcinoma,HBV-HCC),丙型肝炎病毒相关相关HCC,非酒精性脂肪性肝病相关HCC,无危险史相关HCC和其他因素因素HCC。然后在每组患者中,再根据风险评分和截点值将患者分为高风险组和低风险组。通过K-M及ROC分析,我们发现该模型对于乙肝相关HCC患者具有最好的预测效能(3-年AUC=0.811)。结论通过分析TCGA数据库中HCC患者的lncRNA表达及临床信息,构建了一个新型的lncRNA风险评估模型,可有效预测HCC患者的生存。lncRNA预测模型对HBV-HCC患者具有最好的预测效能。第二部分应用LncRNA模型预测乙肝相关肝细胞肝癌肝切除术后患者的生存目的上一步结果显示模型对于HBV-HCC具有最好的预测效能。接下来的研究我们将选取临床HBV-HCC患者作为研究对象,收集临床HBV-HCC患者的组织、临床及生存信息。qRT-PCR检测并明确LINC01 116,DDX11-AS1,LUCAT1,C10orf91和FIRRE的表达。由于前期该模型的建立是基于RNA-Seq数据基础上的,而考虑到临床可行性,我们基于qRT-PCR结果,并结合临床患者总生存时间(Overall survival,OS)及生存状态,重新构建基于qRT-PCR数据的预测临床HBV-HCC患者预后的lncRNA模型,并评估模型的预测效能。分析模型与临床病理资料的相关性,并通过列线图的方法拟优化lncRNA预测模型。最后,比较lncRNA 模型与 TNM、Child-Pugh、Okuda、BCLC、CLIP 及 CUPI 分期系统对HBV-HCC患者预后的预测效能。方法1.收集50例肝切除术后HBV-HCC患者以及10例正常对照者的肝脏组织,通过qRT-PCR检测构成模型的5个lncRNA的表达。2.通过单因素及多因素COX回归分析,重新构建基于qRT-PCR结果的针对临床HBV-HCC患者的lncRNA预后预测模型。3.根据lncRNA模型计算患者的风险评分,根据中位值将50例HBV-HCC患者分为高风险组和低风险组,通过K-M分析及ROC曲线来评估风险预测模型的预测效能。4.收集并统计纳入临床研究队列的50例经肝切除术后的HBV-HCC患者相应的实验室检查及手术病理数据。通过单因素和多因素Cox风险比例分析,评估lncRNA模型是否可作为预测HBV相关HCC患者生存的独立预测因子。5.在4-lncRNA预测模型的基础上,联合患者的临床数据,构建列线图,试图优化目前模型的预测效能,并通过C指数(C-index)和校准曲线来评估列线图的预测效能。6.通过K-M分析及ROC曲线来评估,并比较lncRNA模型、TNM、Child-Pugh、Okuda、BCLC、CLIP及CUPI分期系统对HBV相关HCC患者预后的预测效能。结果1.LINC01116,LUCAT1,DDX11-AS1,C10orf91 和 FIRRE 在 HBV 相关HCC患者肝组织中的表达水平,均明显高于正常对照肝组织。与我们前期从TCGA数据库中所得到的结果一致。2.单因素回归分析显示,5个lncRNA均与HBV相关HCC患者生存相关。但是在COX多因素回归分析中,lncRNAC10orf91的风险比(Hazardratio,HR)从单因素的>1变成了在多因素COX分析中HR<1,提示该lncRNA在模型中不稳定,并可能与其它lncRNA之间存在较大的相互关系,进而会影响模型的稳定性。因此,我们将C10orf91剔除,重新进行多因素COX分析,重新构建了基于4个lncRA((LINC01116,LUCAT1,DDX11-AS1和FIRRE)的,适用于HBV相关HCC肝切除术后患者生存预测的模型。3.K-M分析显示,低风险组患者的总生存时间(3.67±0.09年)明显长于高风险的HBV-HCC患者(2.01±0.25年)。此外,通过ROC曲线分析显示4-lncRNA预测模型的预测效能优于DDX11-AS1,LUCAT1,LINC01116和FIRRE,3年生存预测的AUC为0.875(95%置信区间,95%Confidence interval,95%(CI=0.769-0.981)4.单多因素分析显示,4-lncRNA标记可作为临床术后HBV相关HCC患者的独立预测因子。此外,高风险组患者白蛋白水平明显低于低风险组(P=0.004)。高风险组与低风险组患者的Child-Pugh分级有明显差异(P=0.042)。5.基于来自多元分析的独立因素,进一步建立了列线图。除模型外,我们纳入的因素还包括ALP和肿瘤最大直径。我们发现虽然列线图的C指数略高于单纯的lncRNA预测模型,但两者相比较并没有统计学差异(P=0.175)。此外,nomogram模型的校准曲线在1年、2年及3年中均显示与参考线偏差较大,反应其预测与实际关系一致性较低。进一步通过ROC分析表明,lncRNA预后模型对2年和3年生存率要优于列线图模型,具有最佳预测效率。综上所述,结合临床指标所构建的列线图模型,其预测效能并不优于lncRNA模型。6.通过比较 lncRNA 模型与 TNM、Child、Okuda、BCLC、CLIP 及 CUPI 分期系统在预测HBV相关HCC患者生存的预测效能。结果显示,基于qRT-PCR的lncRNA风险评分模型具有最佳预测效能(AUC=0.897,95%CI=0.733-0.96)。结论通过TCGA数据库及临床数据库,建立了一种有效的基于qRT-PCR的4-lncRNA模型,该模型的预测效能优于TNM、Child、Okuda、BCLC、CLIP及CUPI分期系统,可用于术后HBV-HCC患者的生存预测和治疗指导。第三部分LncRNA模型预测肝细胞肝癌患者生存的机制研究目的通过前两部分的研究,我们基于TCGA数据库,同时联合临床队列,建立了有效的lncRNA模型来预测HCC患者的生存,接下来将针对lncRNA模型调控HCC的机制进行进一步探究。方法1.通过qRT-PCR检测lncRNA在正常肝细胞系LO2和HCC细胞系(Huh7,HepG2,MHCC97-h,LM3)中的表达情况。并采用荧光原位杂交(Fluorescent in situ hybridization,FISH)实验,明确 LINC01116,LUCAT1,DDX11-AS1 和FIRRE的亚细胞定位。2.通过TCGA数据库下载RNA-Seq以及微小RNA测序(microRNA-Sequence,miRNA-Seq)数据,通过Perl以及R软件提取并筛选差异表达的mRNA以及差异表达的miRNA,用于机制分析。3.采用 miRcode 数据库、miRTarBase 和 TargetScan 数据库筛选 lncRNA-miRNA-mRNA 关系,拟构建内源竞争 RNA(competingendogenous RNA,ceRNA)机制。此外,采用皮尔逊相关系数用于评估lncRNA和mRNA之间的共表达关系。筛选分别与LINC01116,LUCAT1,DDX11-AS1和FIRRE具有相关性的mRNA(R>0.4,并且P<0.05),接下来从以上的mRNA中进一步筛选,选取与至少两个lncRNA的表达变化高度相关的mRNA。最后使用Cytoscape软件来构建和可视化的ceRNA以及lncRNA-mRNA共表达网络。4.通过基因功能注释(Gene Ontology,GO)和京都市基因与基因组百科全书(Kyoto Encyclopedia of Genes and Genomes,KEGG)信号通路来分析并显示lncRNA拟调控基因的富集通路。结果1.与正常肝细胞系相比,LINC01116,LUCAT1,DDX1 1-AS1 和 FIRRE 在4种HCC细胞系中均表达升高。FISH结果显示,LINC01116和DDX11-AS1在细胞核和细胞质中均表达,lncRNALUCAT1和FIRRE主要富集在细胞质中,少数富集在细胞核中。2.通过选取P<0.05和log2差异倍数(log2 foldchange,log2FC)绝对值≥1为纳入条件,共筛选出来差异表达的mRNA共4851个,其中有表达水平高于正常对照组的mRNA有3769个,表达水平低于于正常对照组的差异mRNA有1082个。此外,分析并筛选了差异表达的miRNA共250个(|log2FC|≥1,P<0.05),其中表达水平高于正常对照组的miRNA共222个,表达水平低于于正常对照组的miRNA共28个。3.LINC01116,LUCAT1,DDX11-AS1 和 FIRRE 这 4 个 lncRNA 均没有筛选出潜在的ceRNA网络。接下来通过计算皮尔逊相关系数相关系数以及维恩图(Venndiagram)分析表明,筛选出97个至少与两个lncRNA具有共表达关系的蛋白编码基因。并最终通过Cytoscape构建了可视化的共表达网络,结果显示,lncRNA DDX1 1-AS1处于调控网络的中心位置,提示其在HCC进展中发挥重要作用。4.GO分析的结果表明,lncRNA相关的基因显着富集了 9个GO途径,主要为调控染色体和细胞周期相关。KEGG富集分析证实,这些lncRNA调控的蛋白编码基因富集在细胞周期相关途径,叉头转录因子(Forkheadbox O,FOXO)信号传导通路,以及与癌症发生发展密切相关的过氧化物酶体增殖物激活受体(Peroxisomeproliferator-activated receptor,PPAR)信号传导途径以及经典的 p53信号通路。此外,富集的通路还集中在与肝脏功能关的初级胆汁酸的生物合成通路,以及多种细胞代谢通路。结论构建了 lncRNA模型调控HCC进展的作用网络,并对lncRNA影响HCC进展进而预测肝细胞肝癌患者生存的作用机制进行了全面分析。此外,发现lncRNA DDX11-AS1处于调控网络的中心位置,提示其在HCC进展中发挥重要作用。该部分的研究为进一步深入探究lncRNA在HCC中的作用提供了方向。第四部分模型中LncRNA DDX11-AS1在肝细胞肝癌中的作用机制研究目的第三部分的研究表明,lncRNA DDX11-AS1处于调控网络的“中心位置”,提示其在调控HCC进展中发挥最为重要作用。因此,在该部分的研究中我们将选择DDX11-AS1为研究对象,进一步通过体内外实验探究DDX11-AS1调控HCC的具体作用机制。方法1.将TCGA数据库中的HCC患者,根据DDX11-AS1的表达水平,将患者分为高表达组和低表达组,通过基因集富集分析(Gene Set Enrichment Analysis,GSEA)分析,预测DDX1 1-AS1高表达组中所富集的通路。2.收集15例HCC患者及8例正常对照组患者的肝组织,采用免疫组织化学染色以及FISH实验,分别检测CD31又称血小板-内皮细胞粘附分子(Platelet endothelial cell adhesion molecule-1,PECAM-1/CD31)和 DDX11-AS1 在肝组织中的表达水平,并分析肝组织中DDX11-AS1与CD31表达的关系。3.收集患者临床资料,通过卡方检验,比较分析DDX11-AS1的表达与HCC患者临床病理特征的相关性。4.筛选DDX11-AS1表达高的HCC细胞系,通过构建小干扰RNA(Small interfering RNA,siRNA),来敲降HCC系中DDX11-AS1的表达。荧光显微镜下观察转染效率,并通过qRT-PCR检测敲降效能。5.采用共培养小室,构建敲减DDX11-AS1的HCC细胞与人脐静脉内皮细胞(Human umbilical vein endothelial cell,HUVEC)的共培养体系。通过细胞划痕实验和Transwell实验、检测与DDX11-AS1敲减共培养体系中HUVEC的迁移能力。6.通过基质胶小管形成实验,检测与lncRNA DDX1 l-AS1敲减共培养体系中HUVEC的迁移及官腔形成能力。7.通过 qRT-PCR及酶联免疫吸附试验(Enzyme Linked Immunosorbent Assay,ELISA),检测敲减lncRNADDX11-AS1的HCC细胞及上清液中主要血管生成因子的表达情况。结果1.GSEA分析显示,DDX11-AS1高表达组,富集在包含“血管内皮生长因子(vascular endothelial growth factor,VEGF)信号通路”在内的与血管再生相关通路,因此我们将研究重心放在其是否可调控HCC的血管再生方向。2.HCC患者肝组织内DDX11-AS1表达与CD31表达均明显高于正常对照组。此外,DDX11-AS1表达与反应血管再生程度的CD31表达呈正相关。3.IncRNA DDX11-AS1 的表达除与患者谷丙转氨酶(Alanine aminotransferase,ALT)及碱性磷酸酶(Alkalinephosphatase,ALP)的表达相关外,与患者的肿瘤个数及是否有有血管侵犯密切相关。4.通过检测DDX11-AS1在HCC细胞系内的表达情况,我们选取Huh7和MHCC97h两个细胞系进行接下来的干扰实验。此外,选取siRNA-2用于敲降lncRNA DDX11-AS1 的表达。5.细胞功能学实验显示,干扰lncRNA DDX11-AS1可以抑制Huh7以及MHCC97h细胞共培养体系中HUVEC细胞的迁移能力。6.此外,干扰lncRNA DDX11-AS1也可明显可以抑制Huh7和7MHCC97h细胞共培养体系中HUVEC细胞的血管形成能力。7.qRT-PCR实验结果发现,干扰Huh7中lncRNADDX11-AS1表达后,血管内皮生长因子 A(vascularendothelial growth factorA,VEGFA)、血管生成素1(Angiogenin-1,Ang-1)和血管生成素 2(Angiogenin-2,Ang-2)表达明显降低(P<0.05)。此外,我们进一步通过ELISA的方法检测细胞上清中VEGFA、Ang-1和Ang-2的含量,干扰DDX11-AS1的表达后,Huh7肝癌细胞上清的VEGFA表达和Ang-2的表达水平明显下降。结论DDX11-AS1通过影响肿瘤细胞产生并释放VEGFA等血管再生因子,从而间接调控HCC的血管生成并影响HCC的发生发展。更多还原

【Abstract】 Globally,liver cancer is the sixth most commonly diagnosed cancer and ranked fourth in terms of deaths in 2018,with a steadily rising incidence rate.Hepatocellular carcinoma(HCC)is the most frequent type of liver cancer,accounting for 75-80%of all primary liver cancer cases.Despite the major progress in risk factors,early diagnosis,surveillance methods,and treatment techniques,prognose for HCC patients remains poor.It is vital to investigate patients with a high risk of poor outcomes to guide effective clinical management.Thus,considerable effort has been devoted to establishing various staging systems for HCC using clinical information and pathological criteria.Currently,several classification models,including the clinical tumor-node-metastasis(TNM)staging,Barcelona-Clinic Liver Cancer(BCLC),Okuda staging system,Cancer of Liver Italian Program(CLIP),and Chinese University Prognostic Index(CUPI),have been developed and used clinically.Although these staging systems have proven useful,their predictive efficiency remains limited and fails to involve biological characteristics of HCC that might account for clinical heterogeneity when establishing novel predictive modelsLong noncoding RNAs(IncRNAs)are defined as RNA transcripts longer than 200 nucleotides that have no significant protein-coding capacity.lncRNAs exhibit diverse subcellular localization patterns,and it is becoming increasingly clear that the functions of lncRNAs depend on their unique subcellular localization.The potential functions of lncRNAs in regulating the malignancy of tumor cells have been previously examined.To date,many studies have highlighted the indispensable molecular mechanisms and biological characteristics of lncRNA in HCC occurrence and progression.Moreover,an increasing number of aberrantly expressed lncRNAs can also serve as valuable prognostic gene expression signatures for HCC patients.However,only a few studies have attempted to identify lncRNA prognostic signatures for HCC prognosis prediction,resulting in varying efficiencies and unclear mechanisms.Additionally,none of these studies have compared the predictive efficiencies of lncRNA models and other clinical classification systems.Thus,conducting more convincing clinical validation and finding potential mechanisms of lncRNA-based signature in HCC prognosis prediction are necessary.With the development of high-throughput technologies,public databases have enabled researchers to investigate a variety of approaches for identifying potential biomarkers for HCC diagnosis and prognosis prediction.The Cancer Genome Atlas(TCGA)is a project aimed at generating multi-dimensional maps of the key genomic changes in 33 types of cancer to provide great opportunities for analyzing the high throughput data and clinical features using statistical analysis.All of the existing studies that have reported HCC prognostic models based on lncRNAs only used such online databases.However,the transcriptional profiling and RNA sequencing testing methods used in TCGA database are expensive and difficult to perform at hospital.Moreover,the IncRNA signature developed using tumor data from the online database might not be available for the different testing methods and real clinical patients.Therefore,an easier method for the detection of IncRNAs and a new method-based lncRNA signature are urgently needed.In this study,the five-lncRNA formula was shown to have the best prediction efficiency in HBV-HCC patients based on the TCGA database.Based on this result,the expression level of each lncRNA was detected by qRT-PCR in 50 primary HBV-HCC patients who underwent surgical hepatectomy as their primary treatment,and a qRT-PCR data-based lncRNA signature was first established.Our lncRNA model had the highest prediction efficiency compared to other clinical staging systems(TNM,Child-Pugh,Okuda staging system,BCLC,CLIP,and CUP1)in these resected HBV-HCC patients.Finally,we investigated the expression levels and locations of these five lncRNAs in vitro.In summary,we developed a useful innovative five-lncRNA model for survival prediction and individual treatment guidance of resected HBV-HCC patients.High-risk patients should have shorter follow-up times and combined treatments such as TACE,chemotherapy,and targeted agents.Additionally,our study has provided a deeper understanding of the five lncRNAs,which could be considered prognostic biomarkers,as well as clinical treatment guidance and therapeutic targets.Part 1.Constructing a lncRNA model for predicting the survival of patients with hepatocellular carcinoma through TCGA database AimThe aim of the current study was to identify potential prognostic long non-coding RNA(lncRNA)biomarkers for predicting survival in patients with hepatocellular carcinoma(HCC)using The Cancer Genome Atlas(TCGA)dataset.Subsequently,compare and evaluate the predictive power of the lncRNA signature in HCC patients with different risk factors.Methods1.Downloaded the raw RNA-Seq data and the corresponding clinical information of 374 HCC tumor and 50 normal liver tissue specimens from the TCGA database.The differentially expressed lncRNAs were calculated using the Perl and package "edgeR"from R.2.Univariate and multivariate Cox regression analyses were used to assess the prognostic contribution of these lncRNAs and construct the predictive risk score model.3.The cut-off point for risk score was identified with the median to stratify 370 HCC patients with survival data into the high-risk group and low-risk group.K-M analysis and the receiver operating characteristic(ROC)curve analyses was plotted to evaluate the prognostic effacay of the IncRNA model.4.Stratification analysis was performed for different risk factors of HCC.K-M analysis and the receiver operating characteristic(ROC)curve analyses were performed to compare and evaluate the prognostic effacay of the lncRNA model in the different subgroup.Results1.The raw RNA-Seq data of 374 HCC tissues and 50 normal liver tissues were obtained from TCGA database.After differential expression analysis,1029 upregulated and 58 downregulated differentially expressed lncRNAs were screened out by comparing HCC and normal liver tissues(absolute logFC>2.0,adjusted P<0.05)2.LINC01116,LUC AT1,DDX11-AS1,C10orf91,and FIRRE were independent predictors in the multivariate Cox regression analysis(P<0.05).Finally,a lncRNA-based model was developed as follows:risk score=(0.0882×LINC01116)+(0.2504×DDX11AS1)+(0.1001×C10orf91)+(0.0722×LUCAT1)+(0.1138×FIRRE)3.Based on the risk score,370 HCC patients with known survival times were classified as high-risk group and low-risk group according to the cut-off value.The expression pattern of each lncRNA in every HCC patient is shown through heatmap.The K-M curves of the two groups were significantly different(P<0.0001).Furthermore,the areas under the curves(AUCs)of the above model were 0.710 for the 3-year survival times4.We performed stratification analysis for six HCC groups(alcohol consumption,HBV,HCV,NAFLD,no risk history,and other factors)according to the factors provided by TCGA clinical database.This five-lncRNA signature has the highest prediction efficiency in HB V-HCC patients,,the risk score could largely predict the 3-year survival,as the AUC was 0.81 1.ConclusionsBy analyzing the lncRNA expression and clinical information of HCC patients in the TCGA database,a new lncRNA prognostic model was constructed,which can effectively predict the survival of HCC patients.The lncRNA signature has the best prediction performance for HB V-HCC patients.Part 2.LncRNA signature for prognosis prediction in HBV-HCC patients after resectionAimHBV infection accounts for about 80%of all HCC patients in China,and has the best prediction efficiency according to the risk score model in TCGA database.To better guide clinical treatment,we aimed to develop a new 5-lncRNA prognostic score based on qRT-PCR data and compared its predictive efficiency with that of other clinical staging systems,to further assess the robustness and extend the utility of this signature.Methods1.We included 50 tumor tissues from primary HBV-HCC patients and 10 normal liver tissues collected from patients with hepatic trauma.Subsequently,we determined the expression levels of these five lncRNAs in our prospectively collected samples using qRT-PCR.2.To assess the prognostic formula according to the experimental data,K-M and Cox analyses were performed to recalculate the regression coefficients of each lncRNA.Thus,a new risk score model was constructed derived from qRT-PCR results of the lncRNAs to predict the clinical HBV-HCC patients.2.Univariate and multivariate Cox regression analyses were used to assess the prognostic contribution of these lncRNAs and construct the predictive risk score model.3.To testify whether the qRT-PCR data based lncRNA signature is an independent prognostic factor for OS,we performed univariate and multivariate Cox regression analyses with the signature and clinicopathological factors in HBV-HCC patients after surgical resection.4.All independent prognostic factors which based on multivariate Cox regression analysis were used to construct the nomogram(combined prognostic model)for predicting the probability of 1-,2-and 3-year OS of HCC patients.Then,nomogram validation which consist of discrimination and calibration was performed.C-index was calculated to evaluate the discrimination of the nomogram.5.K-M analysis and ROC curve were used to evaluate and compare the predictive power of lncRNA model,TNM,Child-Pugh,Okuda,BCLC,CLIP and CUPI staging system on the prognosis of HBV-HCC patients.Results1.All lncRNAs-LINC01116,LUCAT1,DDX11-AS1,C10orf91 and FIRRE-displayed highly expression patterns in HBV-HCC tumor tissues when comparing with normal samples,which was consistent with the findings in the TCGA cohort.2.The results of univariate and multivariate Cox regression analyses showed that all five lncRNAs were all significantly correlated with HBV-HCC patients’ OS However,the HR of C10orf91 was>1 in the univariate analysis but<1 in the multivariate analysis.We thought that the lncRNA C10orf91 may have interactions with other variables.Then we excluded C10orf91 and rebuilt the signature.Finally,we obtained a new four-lncRNA model which was stable and suitable for the data of qRT-PCR method using our clinical samples.The model was described as follows:risk score=0.353×LINC01116+0.037xDDX11-AS1+0.032×LUCAT1+0.136×FIRRE3.HBV-HCC patients were divided into high risk and low risk groups using the median score as the cut-off value.According to K-M analysis,patients in the low risk groups had significantly longer OS(3.67±0.09 years)than patients in the HRG(2.01 ±0.25 years).In addition,the prognostic capacities of individual lncRNAs and the four-lncRNA signature were assessed by the ROC curve analysis.The 3-year AUCs were 0.875(95%CI=0.769-0.981),0.735(95%CI=0.592-0.878),0.770(95%CI=0.612-0.928),0.852(95%CI=0.741-0.964),and 0.809(95%CI=0.674-0.944)for lncRNA signature,DDX11-AS1,LUCAT1,LINC01116,and FIRRE,respectively,indicating that the risk score has the best predictive efficiency.4.In univariate Cox analysis,the prognostic model,total bilirubin,alkaline phosphatase(ALP),and maximum diameter of the tumor were all correlated with the survival of resected HBV-HCC patients.Then,those factors were selected to the multivariate analysis,which showed only our signature(P<0.001),ALP(P=0.011)and maximum diameter of the tumor(P=0.017)to be the independent prognostic factor Thus,our results proved that this lncRNA signature could be used as an independent predictor in clinical postoperative HBV-HCC patients.5.A nomogram was established to predict HCC patients’ survival on the basis of the independent factors from the multivariate analysis.The validation of the nomogram was consisted of discrimination and calibration.Calibration curve for the probability of overall survival at 3-year did not indicated optimal agreement between the nomogram’prognostic value and actual observation.However,even this nomogram’ C-index was greater than that of other models,the nomogram did not have better prognostic efficiency than the four-lncRNA signature(P=0.175).In addition,the ROC curve analysis indicated that our prognostic model had the best predictive efficiency for survival for the 2-year and 3-year survival times.6.K-M analysis and ROC curve analysis were conducted to assess the prognostic capacity of lncRNA model,TNM,Child-Pugh,Okuda,BCLC,CLIP and CUPI systems.The results showed that the qRT-PCR-based 4-lncRNA formula had the best predictive efficiency for survival(AUC=0.875,95%CI=0.769-0.981)compared to other six clinical staging systems in postoperative HBV-HCC patients.ConclusionsThrough the TCGA database and clinical database,an effective qRT-PCR-based 4-lncRNA model was established.This formula had the best predictive efficiency for survival compared to TNM,Child,Okuda,BCLC,CLIP and CUPI staging systems.And it can be used for postoperative HBV-HCC patietns’ survival prediction and treatment guidance.Part 3.The mechanism of LncRNA model predicting the survival of hepatocellular carcinoma patientsAimThe aim of the current study was to performed comprehensive bioinformatics analyses to analysis the deeper potential regulatory mechanisms of these lncRNAs in HCC.Methods1.The qRT-PCR and FISH analyses were used to detect the expression levels and locations of LINC01116,DDX11-AS1,LUCAT1 and FIRRE in LO2 and HCC cell lines(Huh7,HepG2,MHCC97-h,LM3)2.The RNA-Seq data and miRNA-Seq data of 374 HCC samples and 50 normal liver tissue specimens were obtained from the TCGA.The differentially expressed mRNAs and miRNAs were filtered out with |log2FC|≥1.0 and adjusted P value<0.05.3.The interactions between the differentially expressed lncRNAs and miRNAs were predicted by the miRcode,miRDB,miRTarBase and TargetScan.Pearson correlation to calculate the relationship between these lncRNAs and the differently expressed mRNAs.Cytoscape software was used to visualize the network.4.Biological functions of the lncRNA target genes were predicted using functional enrichment analysis of Gene Ontology(GO)and The Kyoto Encyclopedia of Genes and Genomes(KEGG)pathways.Results1.All of the HCC cell lines indicated higher expression levels of each lncRNA compared to the normal hepatocyte cell line LO2.LINC01116,DDX11-AS1 and FIRRE are expressed in both nuclear and cytoplasmic compartments in Huh7 cell lines.However,LUCAT1 is mainly expressed in the cytoplasm.2.Using P<0.05 and absolute logFC≥1.0 as cutoffs,we found that 4851 mRNAs and 250 miRNAs were differentially expressed between HCC and normal samples according to TCGA database.2.A total of 97 protein-coding genes were selected that correlated with at least two of these four lncRNAs.The visualization of co-expression networks of these lncRNAs and mRNAs was constructed.3.GO terms and KEGG pathways were performed to explore the potential molecular mechanisms of the target genes including in co-expression networks.The KEGG pathways were significantly enriched in "Cell cycle","Progesterone-mediated oocyte maturation","Valine,leucine and isoleucine degradation","Oocyte meiosis","Propanoate metabolism" and "FoxO signaling pathway".ConclusionsThis study has shed new light on the prognostic lncRNAs by identifying their expression patterns and the regulatory mechanism of HCC progression,which could be considered as prognostic biomarkers and therapeutic targets.In addition,it was found that lncRNA DDX11-AS1 is at the center of the regulatory network,suggesting that it plays an important role in the progression of HCC.Part 4.The mechanism of LncRNA DDX11-AS1 in regulating hepatocellular carcinomaAimIn this part of the study,we will further explore the specific mechanism of DDX11-AS1 in regulating the progress of HCC through in vivo and in vitro experiments.Methods1.Divide the HCC patients in the TCGA database into high expression group and low expression group according to the expression level of DDX11-AS1.Gene Set Enrichment Analysis(GSEA)analysis was performed to analysis the enriched pathways in the high expression group.2.Liver tissues of 15 HCC patients and 8 normal control patients were collected.And immunohistochemical staining and FISH were used to detect the expression levels of CD31 and DDX11-AS1.Then analyzed the relationship of DDX11-AS1 and CD31.3.Analyze the correlation between the expression of DDX11-AS1 and the clinicopathological characteristics of HCC patients.4.Small interfering RNA(siRNA)was used to knock down the expression of DDX11-AS1 in HCC lines.The transfection efficiency was observed under a fluorescence microscope,and the knockdown efficiency was detected by qRT-PCR.5.Detect the migration ability of HUVEC in the co-culture system by cell scratch experiment and Transwell experiment.6.Detect the migration ability of HUVEC in the co-culture system with DDX11-AS1 knockdown by cell scratch experiment and Transwell experiment7.qRT-PCR and enzyme linked immunosorbent assay(ELISA)were performed to detect the expression of major angiogenic factors in HCC cells and supernatants of HCC cells.Results1.We found that the DDX11-AS1 high expression group is enriched in pathways related to angiogenesis including the "vascular endothelial growth factor(VEGF)signaling pathway",so we will focus our research on whether it can regulate the angiogenesis in HCC.2.The results showed that the expression of DDX11-AS1 and CD31 in the liver tissue of HCC patients were significantly higher than those of the normal control group.In addition,the expression of DDX1 1-AS1 is positively correlated with the expression of CD31,which reflects the degree of angiogenesis.3.The expression of DDX11-AS1 is closely related to the patient’s Alanine aminotransferase(ALT)and alkaline phosphatase(ALP),the number of tumors,and the vascular invasion.4.Two cell lines Huh7 and MHCC97h were selected for the next interference experiment.In addition,siRNA-2 was selected to knock down the expression of lncRNA DDX11-AS1.5.Interfering with lncRNA DDX11-AS 1 can inhibit the migration ability of HUVEC cells in Huh7 and MHCC97h cell co-culture system.6.In addition,interfering with IncRNA DDX11-AS1 can also obviously inhibit the angiogenesis ability of HUVEC cells in the co-culture system of Huh7 and 7MHCC97h cells.7.qRT-PCR showed that after interfering with the expression of lncRNA DDX 11-AS1 in Huh7,the expression of VEGFA,Ang-1 and Ang-2 were significantly reduced.In addition,the expression of VEGFA and Ang-2 in the supernatant of Huh7 decreased significantly.ConclusionsLncRNA DDX11-AS1 affects tumor cells to produce and release vascular regeneration factors such as VEGFA.Thereby indirectly regulating the angiogenesis of HCC and affecting the occurrence and development of HCC.更多还原