基于贝叶斯超参数优化后随机森林回归模型探讨SPP2在女性肺腺癌中的作用机制  

The role of SPP2 in female lung adenocarcinomas explored through a random forest regression model with Bayesian hyperparameter optimization

作　　者：何杰[1] 陈济[2] 周奇银 李妮 黄娜[1] 黄蓉飞[1] 李佳 He Jie;Chen Ji;Zhou Qiyin;Li Ni;Huang Na;Huang Rongfei;Li Jia(Department of Respiratory and Critical Care Medicine,the First Affiliated Hospital of Chengdu Medical College,Chengdu 610500,China;Department of Oncology,Chengdu Seventh People′s Hospital,Chengdu 610041,China)

机构地区：[1]成都医学院第一附属医院呼吸与危重症医学科,成都610500 [2]成都市第七人民医院肿瘤科,成都610041

出　　处：《国际呼吸杂志》2025年第3期192-206,共15页International Journal of Respiration

基　　金：四川省自然科学基金(2022NSFSC0725);成都医学院临床科研基金资助科研项目(2021LHJYPJ-03);成都医学院第一附属医院高层次引进人才科研启动基金(CYFY-GQ59)。

摘　　要：目的通过生物信息学和随机森林算法筛选核心基因SPP2,结合细胞实验探讨SPP2在女性肺腺癌中的调控机制。方法本研究为实验研究。从TCGA数据库中提取女性肺腺癌患者的临床信息和基因表达数据,使用"limma"包分析肿瘤样本和正常样本之间基因差异表达。采用Perl软件将差异基因的表达量和每一个女性肺腺癌患者的生存时间和预后结局一对一匹配。采用"Survival"包对Perl软件处理好的矩阵数据进行单因素Cox生存分析,随后,利用"randomforestSRC"包建立回归模型,贝叶斯算法优化模型的超参数,筛选排列第一位的基因并将其作为hub基因。应用CIBERSORT算法分析女性肺腺癌样本中SPP2表达量与肿瘤22种免疫细胞丰度的关系。选取女性肺腺癌数据集GSE19188和单细胞测序数据集GSE117570,验证SPP2的表达。以SPP2 mRNA表达量的中位值作为截断值,将276例女性肺腺癌样本分为SPP2高表达组(138例)和SPP2低表达组(138例),采用Kaplan-Meier和Log-rank检测,以及Cox回归分析探索SPP2基因表达对女性肺腺癌的预后价值,并对2组进行差异基因分析。差异基因导入GSEA预测SPP2所涉及的信号通路。采用非随机抽样的方法选取2021年1月至2022年12月在成都医学院第一附属医院行肺腺癌切除术的女性患者20例,收集肺腺癌的组织及配对的癌旁正常肺组织。免疫组织化学染色分析SPP2在肺腺癌组织中的表达,实时定量聚合酶链反应和蛋白质印迹法检测女性肺腺癌细胞系中SPP2 mRNA和蛋白表达情况。选择HCC515细胞为研究对象,分别转染SPP2 shRNA(LV-shSPP2)和阴性对照(LV-control),得到sh-SPP2组和sh-NC组。蛋白质印迹法检测沉默SPP2对HCC515细胞上皮间质转化和雌激素受体的影响,运用Transwell实验检测沉默SPP2后对HCC515细胞迁移、侵袭的影响,噻唑蓝试验用于检测沉默SPP2对细胞增殖的影响。结果TCGA数据库中提取了34例女性肺腺癌患者的�ObjectiveTo investigate the regulatory mechanism of the core gene secreted phosphoprotein-2(SPP2)in female lung adenocarcinomas through an integration of bioinformatics,random forest algorithms and in vitro experiments.MethodsThis was an experimental study.Clinical information and gene expression data of female lung adenocarcinomas were extracted from The Cancer Genome Atlas(TCGA)database.The"limma"package was used to analyze differentially expressed genes(DEGs)between tumor and normal samples.Perl software was used to match the expression levels of DEGs with the survival time and prognosis outcomes of each female lung adenocarcinoma patient on a one-to-one basis.The"Survival"package was used for univariate Cox survival analysis on the matrix data processed by Perl.Subsequently,the"randomforestSRC"package was used to establish a regression model,with the Bayesian algorithm optimizing the hyperparameters in the model.The first ranked gene was screened as a hub gene.The CIBERSORT algorithm was applied to analyze the relationship between SPP2 expression and the abundance of 22 types of immune cells in female lung adenocarcinomas.The expression of SPP2 was validated using the female lung adenocarcinoma dataset GSE19188 and the single-cell sequencing dataset GSE117570.Using the median value of mRNA expression of SPP2 as the cut-off value,276 samples of female lung adenocarcinomas were divided into the SPP2 high-expression group(138 cases)and the SPP2 low-expression group(138 cases).Kaplan-Meier plots and Log-rank tests as well as Cox regression analyses were used to explore the prognostic value of SPP2 in female lung adenocarcinomas and identify DEGs between the two groups.DEGs were imported into GSEA to predict the signaling pathways enriched in SPP2.Twenty female patients who underwent lung adenocarcinoma resection at the First Affiliated Hospital of Chengdu Medical College from January 2021 to December 2022 were selected using non-random sampling methods.Lung adenocarcinoma tissues and paired adjacent normal lung t

关 键 词：肺腺癌 计算生物学 受体 雌激素 SPP2 女性 上皮间质转化 

分 类 号：R734.2[医药卫生—肿瘤]