| 本文已被：浏览 831次 下载 576次 |
码上扫一扫！ |
| 基于铁死亡基因的乳腺癌预后模型构建 |
| 万坤铭1,谢玮1,袁梓珂1,蔡孟成1,金永生2,罗鹏飞3,俞超芹*4 |
|
|
(1. 海军军医大学(第二军医大学)基础医学院学员一大队, 上海 200433;
2. 海军军医大学(第二军医大学)药学系有机化学教研室, 上海 200433;
3. 海军军医大学(第二军医大学)第一附属医院烧创伤中心, 上海 200433;
4. 海军军医大学(第二军医大学)第一附属医院中医妇科, 上海 200433
*通信作者) |
|
| 摘要: |
| 目的 通过癌症基因组图谱（TCGA）数据库挖掘与乳腺癌预后相关的铁死亡基因，构建乳腺癌预后模型。方法 下载TCGA数据库中转录组与临床数据，获取与预后相关的在乳腺癌组织与癌旁正常组织中存在差异表达的铁死亡基因，利用最小绝对收缩和选择算子（LASSO）回归法构建风险评分模型。将TCGA数据库中获取的患者信息作为模型测试集数据，通过ROC曲线评估预后模型的效能，通过单因素与多因素Cox回归分析评价差异表达的铁死亡基因和风险评分能否作为预后因子。利用国际肿瘤基因组协作组（ICGC）数据库与基因表达汇编（GEO）数据库数据作为验证集对该模型进行验证。结果 共筛选出51个在乳腺癌组织与癌旁正常组织中存在差异表达的铁死亡基因，单因素Cox回归分析表明11个差异表达的铁死亡基因与乳腺癌预后相关。利用这11个基因构建乳腺癌预后风险评分模型：风险评分＝ALOX15×0.11＋CHAC1×0.07＋CISD1×0.15＋CS×0.24＋GCLC×0.04＋GPX4×（－0.07）＋NCOA4×0.17＋EMC2×0.30＋G6PD×0.19＋ACSF2×（－0.04）＋SQLE×0.12，ROC曲线分析表明该模型在测试集中预测乳腺癌患者术后2、4、6年生存率的AUC分别为0.678、0.680、0.612，多因素Cox回归分析结果显示预后模型风险评分可作为独立预后因子（HR＝3.104，P＜0.001）。根据预后模型风险评分是否≥4.277将患者分为高风险组和低风险组，在测试集与验证集中高风险组患者的生存率均低于低风险组患者（P均＜0.001）。结论 基于铁死亡基因的乳腺癌预后模型具有较好的预测效能，该模型中的铁死亡基因为乳腺癌靶向治疗提供了新的靶点。 |
| 关键词: 乳腺肿瘤 铁死亡 预后模型 靶向治疗 生物信息学 |
| DOI：10.16781/j.CN31-2187/R.20211256 |
| 投稿时间：2021-12-05修订日期：2022-01-15 |
| 基金项目: |
|
| Construction of breast cancer prognosis model based on ferroptosis-related genes |
| WAN Kun-ming1,XIE Wei1,YUAN Zi-ke1,CAI Meng-cheng1,JIN Yong-sheng2,LUO Peng-fei3,YU Chao-qin4 |
(1. The First Student Team, College of Basic Medical Sciences, Naval Medical University (Second Military Medical University), Shanghai 200433, China;
2. Department of Organic Chemistry, School of Pharmacy, Naval Medical University (Second Military Medical University), Shanghai 200433, China;
3. Burn and Trauma Center, The First Affiliated Hospital of Naval Medical University (Second Military Medical University), Shanghai 200433, China;
4. Department of Traditional Chinese Medicine (Gynaecology), The First Affiliated Hospital of Naval Medical University (Second Military Medical University), Shanghai 200433, China
*Corresponding author) |
| Abstract: |
| Objective To mine the ferroptosis-related genes associated with breast cancer prognosis through The Cancer Genome Atlas (TCGA) database, and construct a prognosis model.Methods The transcription group and clinical data were downloaded from TCGA database, differentially expressed ferroptosis-related genes in tumor tissues and adjacent normal tissues related to prognosis were obtained, and risk score model was constructed by least absolute shrinkage and selection operator (LASSO) regression. The patient information obtained from TCGA database was used as the model test set data. The effectiveness of the model was evaluated by receiver operating characteristic (ROC) curve, and univariate and multivariate Cox regression analyses were used to evaluate whether the differentially expressed ferroptosis-related genes and the risk scores could be used as prognostic factors. The model was verified by International Cancer Genome Consortium (ICGC) and Gene Expression Omnibus (GEO) databases.Results A total of 51 ferroptosis-related genes, which were differentially expressed in breast cancer tissues and adjacent normal tissues, were screened out. Univariate Cox regression analysis showed that 11 of them were associated with prognosis. A prognostic risk score model (risk score＝ALOX15×0.11＋CHAC1×0.07＋CISD1×0.15＋ CS×0.24＋GCLC×0.04＋GPX4×［ －0.07］＋NCOA4×0.17＋EMC2×0.30＋G6PD×0.19＋ACSF2×［ －0.04］＋ SQLE×0.12) for breast cancer was constructed with the 11 genes. ROC curve analysis showed that the area under curve (AUC) of the model in predicting the 2-, 4- and 6-year survival rates of breast cancer patients in the test set were 0.678, 0.680 and 0.612, respectively. The results of multivariate Cox regression analysis showed that the risk score could be used as an independent predictor (hazard ratio＝3.104, P＜0.001). The patients were divided into high-risk group (risk score≥4.277) and low-risk group (risk score＜4.277) according to the risk scores of the prognosis model. In the test set and validation set, the survival rate of highrisk patients was significantly lower than that of low-risk patients (both P＜0.001).Conclusion The prognosis model of breast cancer based on ferroptosis-related genes has better predictive performance. The ferroptosis-related genes in this model provides new targets for targeted therapy of breast cancer. |
| Key words: breast neoplasms ferroptosis prognostic model targeted therapy bioinformatics |
