目的:胃癌是一种具有异质性的高度侵袭性恶性肿瘤,这使得预后的预测变得困难.本研究旨在探讨多胺代谢与胃癌预后的关系,并建立基于多胺代谢相关基因(polyamine-related gene,PMRG)的预后模型.方法:从癌症基因组图谱(The Cancer Genome Atlas,TCGA)数据库和基因表达综合(Gene Expression Omnibus,GEO)数据库中下载胃癌患者的临床病理信息和基因表达数据,并根据PMRG的表达模式将其划分为2类不同的分子分型(PMRG cluster A和PMRG cluster B).依据分子分型之间的差异基因,基于LASSO(least absolute shrinkage and selection operator)回归和多因素COX回归分析建立风险预后模型,并评估患者预后、受试者操作特征(receiver operating characteristic,ROC)曲线、免疫细胞浸润、肿瘤微环境、微卫星不稳定性等.应用肿瘤免疫单细胞中心2数据库从单细胞水平分析模型基因的表达模式.通过人类蛋白图谱数据库验证模型基因的蛋白表达情况.结果:在PMRG表达的2个分型中,PMRG cluster B型的生存结果好于PMRG cluster A型(P<0.05),PMRG cluster B型的中位生存期为7.81年,PMRG cluster A型的中位生存期为4.93年.筛选了7个特征基因[ZW10相互作用着丝粒蛋白(ZW10 interacting kinetochore protein,ZWINT)、醛脱氢酶 1 家族成员A3(aldehyde dehydrogenase 1 family member A3,ALDH1A3)、RNA结合蛋白2(RNA binding protein with multiple splicing 2,RBPMS2)、酮酸脱氢酶激酶4(pyruvate dehydrogenase kinase 4,PDK4)、黑色素瘤相关抗原-A3(melanoma antigen family A3,MAGEA3)、溶质载体家族27成员2(solute carrier family 27 member 2,SLC27A2)和细胞分裂周期相关7(cell division cycle associated 7,CDCA7)]构建风险预后模型,并进行了验证.ROC曲线评估显示:模型整体预测性能良好,1、3和5年的曲线下面积分别为0.642、0.680和0.701.免疫浸润分析结果表明:高风险患者主要浸润的免疫细胞类型为静息肥大细胞、单核细胞、初始B细胞、M2巨噬细胞、静息树突状细胞和静息CD4+记忆性T细胞.肿瘤微环境分析结果表明:高风险患者基质评分和ESTIMATE评分较高,说明高风险患者有较高程度的肿瘤细胞浸润.单细胞转录患者测序分析结果显示:ZWINT、PDK4、SLC27A2、CDCA7基因主要在小凹细胞中表达,ALDH1A3基因主要在成纤维细胞中表达.人类蛋白图谱(Human Protein Atlas,HPA)数据库表明在蛋白水平上,ALDH1A3在正常组织中表达强于胃癌组织,PDK4在胃癌组织中的阳性表达强于正常胃组织.结论:本研究探讨了PMRG的表达模式及其与肿瘤特征的关系,鉴定了2个分子亚型,并为预测胃癌患者的预后提供了一种新的模型.新的预后模型可为胃癌患者的临床预后预测和治疗提供新的参考.
Identification of molecular subtypes and construction of a prognostic model for gastric cancer based on polyamine metabolism-related genes
Objective:Gastric cancer is a highly heterogeneous and aggressive malignancy,making prognostic prediction challenging.This study aims to explore the relationship between polyamine metabolism and the prognosis of gastric cancer,and to construct a prognostic model based on polyamine metabolism-related genes(PMRG).Methods:Clinicopathological information and gene expression data of gastric cancer patients were downloaded from The Cancer Genome Atlas(TCGA)and Gene Expression Omnibus(GEO)databases.Based on the expression patterns of PMRG,patients were divided into two different molecular subtypes(PMRG cluster A and PMRG cluster B).A risk prognostic model was established using least absolute shrinkage and selection operator(LASSO)regression and multivariate COX regression analysis based on the differential genes between the molecular subtypes,and patient prognosis,receiver operating characteristic(ROC)curves,immune cell infiltration,tumor microenvironment,and microsatellite instability were evaluated.The expression patterns of the model genes at the single-cell level were analyzed using the Tumor Immune Single-cell Hub 2 database.The protein expression of model genes was validated through the Human Protein Atlas database.Results:Among the two PMRG expression subtypes,the survival outcome of the PMRG cluster B type was better than that of the PMRG cluster A type(P<0.05),with a median survival time of 7.81 years for the PMRG cluster B type and 4.93 years for the PMRG cluster A type.Seven characteristic genes[ZW10 interacting kinetochore protein(ZWINT),aldehyde dehydrogenase 1 family member A3(ALDH1A3),RNA binding protein with multiple splicing 2(RBPMS2),pyruvate dehydrogenase kinase 4(PDK4),melanoma antigen family A3(MAGEA3),solute carrier family 27 member 2(SLC27A2),and cell division cycle associated 7(CDCA7)]were selected to construct and validate the risk prognostic model.The ROC curve indicated good overall predictive performance of the model,with areas under the curve of 0.642,0.680,and 0.701 for 1,3,and 5 years,respectively.Immune infiltration analysis showed that high-risk patients predominantly had infiltration by resting mast cells,monocytes,naïve B cells,M2 macrophages,resting dendritic cells,and resting CD4+T memory cells.Tumor microenvironment analysis indicated that high-risk patients had higher stromal and ESTIMATE scores,suggesting a higher degree of tumor cell infiltration.Single-cell RNA sequencing analysis showed that ZWINT,PDK4,SLC27A2 and CDCA7 genes were mainly expressed in foveolar cells,while ALDH1A3 was mainly expressed in fibroblasts.The Human Protein Atlas(HPA)database revealed that ALDH1A3 was more strongly expressed in normal tissues than in gastric cancer tissues,while PDK4 showed positive expression in gastric cancer tissues compared to normal gastric tissues.Conclusion:The study explores the expression patterns of PMRG and their relationship with tumor characteristics,identifies 2 molecular subtypes,and provides a novel model for predicting the prognosis of gastric cancer patients.This new prognostic model offers a novel reference for clinical prognosis prediction and treatment of gastric cancer patients.
gastric cancerpolyamine metabolismmolecular subtypesprognostic model
梁路、唐旭鹏、张立、申娟宁、李宁
展开 >
山西医科大学第一临床医学院,太原 030001
山西医科大学第一医院病理科,太原 030001
胃癌 多胺代谢 分子分型 预后模型
山西省基础研究计划(自由探索类)项目Basic Research Project of Shanxi Province(Free exploration class)
万方数据
维普
