基于GEO数据库建立上皮性卵巢癌分子预后风险评分模型

Construction of a molecular prognostic risk score model of epithelial ovarian cancer based on GEO datebase

目的探索上皮性卵巢癌预后特征基因并构建分子预后风险评分模型。方法从GEO数据库中下载GSE26712数据集,其中包括185例上皮性卵巢癌患者和10例正常对照组织的表达数据和相应的临床数据。将上皮性卵巢癌患者分为训练组(129例)和验证组(56例)。在GSE26712数据集中筛选差异表达基因并从训练组中识别预后相关基因。对共有的基因进行LASSO和逐步回归,筛选出最佳预后基因,通过多因素Cox比例风险回归确定预后特征基因的回归系数,并构建风险评分模型。受试者工作特征曲线(ROC)、Kaplan-Meier(KM)生存分析和十折交叉验证用于评估其预测能力。结果筛选出IGFBP4、IGF2、TLR2、DIAPH2、AADAC共5个与卵巢癌预后显著相关的基因。根据5个基因的Cox系数构建风险评分模型进行预后预测分析,KM生存曲线显示高风险组的患者预后更差(P<0.0001)。ROC曲线显示1、3、5年的曲线下面积分别为0.813、0.876、0.895。结论本研究构建的风险评分模型可以较好地预测上皮性卵巢癌患者的预后,为临床医生提供可靠的预后评估工具并辅助临床治疗决策。

Objective To explore the prognostic characteristic genes of epithelial ovarian cancer and construct a molecular prognosis risk score model.Methods The GSE26712 dataset was downloaded from the GEO database and included expression data and corresponding clinical data from 185 patients with epithelial ovarian cancer and 10 normal control tissues.Epithelial ovarian cancer patients were divided into training group(129 cases)and validation group(56 cases).Differentially expressed genes were screened in the GSE26712 dataset and prognosis-related genes were identified from the training group.LASSO and stepwise regression were performed on the common genes to screen out the best prognostic genes.Multivariate Cox proportional risk regression was used to determine the regression coefficients of prognostic characteristic genes,and the risk score model was constructed.Receiver operating characteristic curve(ROC),Kaplan-Meier(KM)survival analysis and 10-fold cross-validation were used to assess their predictive ability.Results Five genes including IGFBP4,IGF2,TLR2,DIAPH2 and AADAC were screened to be significantly related to the prognosis of ovarian cancer.According to the Cox coefficients of five genes,a risk score model was constructed to predict the prognosis.The KM survival curve showed that the prognosis of the high risk group was worse(P<0.0001).ROC curve showed that the 1-,3-,and 5-year areas under the curve were 0.813,0.876,0.895,respectively.Conclusion The risk score model constructed in this study can better predict the prognosis of patients with epithelial ovarian cancer,providing a reliable prognostic assessment tool for clinicians and assisting clinical treatment decision-making.