【作者】 林敏；

【导师】 吴文捷；

【作者基本信息】 福建医科大学 ， 眼科学（专业学位）， 2020， 硕士

【摘要】 目的:转移性葡萄膜黑色素瘤恶性程度高、预后差且治疗方式有限。对转移性葡萄膜黑色素瘤患者进行预后评价至关重要。挖掘预后分子标志物,有助于临床做出准确的诊断分型、危险分层和预后评价。基于高通量测序技术的快速发展,海量的基因组等数据日益完善,为研究肿瘤发生、发展、转移和预后提供了良好的契机。在本研究中,利用生物信息学方法对公共数据库的资源进行挖掘,筛选转移性葡萄膜黑色素瘤(Uveal Melanoma,UM)的预后生物标志物。方法:从基因芯片数据库(gene expression omnibus,GEO)数据库和癌症基因组图谱数据库(The Cancer Genome Atlas,TCGA)数据库下载了葡萄膜黑色素瘤相关m RNA表达谱芯片数据和临床生存预后资料。利用R语言中的“affy包”将GEO数据库数据集GSE22138和GSE27831中的原始基因芯片数据进行预处理,并用“limma包”筛选出在非转移性葡萄膜黑色素瘤与转移性葡萄膜黑色素瘤样本中差异表达的基因(differentially expressed genes,DEGs)。通过STRING数据库构建蛋白-蛋白相互作用网络(protein-protein interaction,PPI),应用Cytoscape中的cyto Hubba插件筛选得到hub基因。采用Kaplan-Meier生存曲线法评估基因高、低表达组患者的总体生存率。对筛选出的差异基因通过LASSO回归模型分析与预后的关系,并用ROC曲线评价预后标志物的诊断价值。结果:通过对数据集GSE22138和GSE27831中的基因芯片数据进行差异表达分析,共得到837个差异表达基因,其中上调基因483个,下调基因344个。从PPI网络中分析得到了CXCL9、POMC、GNGT1、S1PR1、P2RY14、C5AR1、ANXA1、CXCR1、AGT和NPB等10个Hub基因。通过LASSO-COX回归模型得到WNT10B(P<0.001)、PCDHA10(P<0.001)、CFAP65(P<0.001)、KRTCAP2(P<0.001)、ACOT12(P<0.001)、PITX2(P=0.001)、GRAP2(P<0.001)和IGHMBP2(P<0.001)等与患者预后强相关的基因。Kaplan-Meier曲线则显示,ACOT12(HR=0.05,95%CI:0.02～0.12,P<0.001)、CFAP65(HR=0.18,95%CI:0.08～0.42,P<0.001)、IGHMBP2(HR=0.08,95%CI:0.03～0.18,P<0.001)和PCDHA10(HR=0.09,95%CI:0.04～0.2,P<0.001)的高表达与患者预后差相关。而GRAP2(HR=16.85,95%CI:7.39～38.43,P<0.001)、KRTCAP2(HR=9.8,95%CI:3.9～24.63,P<0.001)、PITX2(HR=4.38,95%CI:1.93～9.93,P=0.002)和WNT10B(HR=13.41,95%CI:5.63～31.95,P<0.001)的高表达提示患者预后较好。ROC曲线结果表明:KRTCAP2(AUC=0.79)、PCDHA10(AUC=0.795)和WNT10B(AUC=0.769)、ACOT12(AUC=0.71)和CFAP65(AUC=0.709)在转移性葡萄膜黑色素瘤中有良好的诊断价值。结论:本研究通过挖掘公共数据库内转移性葡萄膜黑色素瘤患者信息,采用生物信息学的方法,分析鉴定了CXCL9、POMC、GNGT1和S1PR1等可能参与葡萄膜黑色素瘤转移过程的10个关键基因,筛选出KRTCAP2、PCDHA10、WNT10B、ACOT12和CFAP65与患者预后强相关的生物标志物。该研究为进一步研究转移性黑色素葡萄膜瘤机制的提供新的思路,为进行大规模转移性葡萄膜黑色素基因组学研究提供了借鉴。更多还原

【Abstract】 Objective: Metastatic uveal melanoma(uveal melanoma,UM)has the disadvantages of high-grade malignancy,poor prognosis and limited therapy method.Prognostic evaluation in patients with metastatic uveal melanoma is critical.Exploring molecular markers of prognosis is helpful for accurate diagnosis classification,risk stratification and prognosis evaluation clinically.The rapid development of high-throughput sequencing technology,massive genome and other complete data provide a good opportunity to investigate the genesis,development,metastasis and prognosis of tumors.In this study,resources from public databases were mined using bioinformatics methods to screen prognostic biomarkers for metastatic uveal melanoma.Methods: The uveal melanoma-associated m RNA expression profile chip data and clinical survival prognosis data were downloaded from the gene expression omnibus(GEO)and the cancer genome atlas(TCGA).The original gene chip data in GEO database dataset GSE22138 and GSE27831 were pretreated with the “affy”package in R language,and the genes differentially expressed(DEGs)in non-metastatic uveal melanoma and metastatic uveal melanoma samples were screened by “limma” package.The protein-protein interaction network(PPI)was constructed by STRING database,and hub genes were obtained by screening cyto Hubba plug-ins in the Cytoscape.Kaplan-Meier survival curve was used to evaluate the overall survival rate of patients with high and low gene expression.The correlation of screened DEGs with the prognosis was analyzed with LASSO regression model,and ROC curve was used to evaluate the diagnostic value of prognostic markers.Results: The results showed that 837 DEGs were obtained by analyzing the microarray data of GSE22138 and GSE27831.Ten Hub genes such as CXCL9,POMC,GNGT1,S1PR1,P2RY14,C5AR1,ANXA1,CXCR1,AGT and NPB were analyzed from PPI network.By LASSO-Cox regression model,WNT10B(P<0.001)、PCDHA10(P<0.001)、CFAP65(P<0.001)、KRTCAP2(P<0.001)、ACOT12(P<0.001)、PITX2(P=0.001)、GRAP2(P<0.001),and IGHMBP2(P<0.001)were found to be closely related to the prognosis of patients.Kaplan Meier curve showed that high expression of ACOT12(HR=0.05,95%CI: 0.02～0.12,P<0.001)、CFAP65(HR=0.18,95%CI: 0.08～0.42,P<0.001)、IGHMBP2(HR=0.08,95%CI: 0.03～0.18,P<0.001),and PCDHA10(HR=0.09,95%CI: 0.04～0.2,P<0.001)were related to poor prognosis,while high expression of GRAP2(HR=16.85,95%CI: 7.39～38.43,P<0.001)、KRTCAP2(HR=9.8,95%CI: 3.9～24.63,P<0.001)、PITX2(HR=4.38,95%CI: 1.93～9.93,P=0.002),and WNT10B(HR=13.41,95%CI: 5.63～31.95,P<0.001)suggested better prognosis.ROC curve showed that KRTCAP2(AUC=0.79)、 PCDHA10(AUC=0.795)和 WNT10B(AUC=0.769)、 ACOT12(AUC=0.71),and CFAP65(AUC=0.709)had good diagnostic value in metastatic uveal melanoma.Conclusion: This study analyzed and identified 10 key genes,such as CXCL9,POMC,GNGT1 and S1PR1,that may be involved in the metastatic process of uveal melanoma by mining the information of patients with metastatic uveal melanoma in a public database and using bioinformatics and screened KRTCAP2、PCDHA10、WNT10B、ACOT12,and CFAP65 biomarkers strongly associated with patient prognosis.This study provides a new idea for further research on the mechanism of metastatic uveal melanoma,and provides a reference for study of large-scale metastatic uveal melanoma genomics.更多还原