临床普外科电子杂志 ›› 2022, Vol. 10 ›› Issue (2): 6-.

• • 上一篇    下一篇

基于衰老相关基因构建预测乳腺癌预后的模型


  

  1. 1. 江门市中心医院 乳腺科, 广东 江门 529000;2. 江门市中心医院 肿瘤内科, 广东 江门 529000; 3. 台山市都斛卫生院 外科,广东 台山 529243;4. 江门市中心医院 胃肠外科,广东 江门 529000
  • 出版日期:2022-04-01 发布日期:2022-07-15

A signature for predicting the prognosis of breast cancer based on aging-related genes

#br#   

  1. 1. Breast Department of Jiangmen Central Hospital, Guangdong Jiangmen 529000, China; 2. Oncology Department of Jiangmen Central Hospital, Guangdong Jiangmen 529000, China;3. Surgery Department of Taishan Duhu Health Center, Guangdong Taishan 529243, China;4. Gastrointestinal Surgery Department of Jiangmen Central Hospital, Guangdong Jiangmen 529000, China
  • Online:2022-04-01 Published:2022-07-15
  • Supported by:

    2021 年广东省医学科学基金(B2021016);2021 年广东省江门市科技计划项目(2021YL01123);

    2019 年江门市中心医院科研杰青项目(J201905)

摘要:

目的 通过生物信息学筛选影响乳腺癌患者预后的衰老基因,并构建预测乳腺癌预后模型。方法 从癌症基因组图谱(the cancer genome atlas,TCGA) 数据库下载2010 年9 月至2015 年6 月美国国家癌症中心收集的乳腺癌患者的临床资料及mRNA 转录组测序数据,利用Aging Atlas 数据库检索出衰老相关基因,比较并筛选出正常组织与乳腺癌组织的差异衰老基因。通过单因素Cox 回归和Lasso 回归筛选出预后相关的衰老基因并构建风险预测模型,以危险系数的中位数作为截值,将患者分为高风险组和低风险组。通过单因素及多因素分析,筛选影响患者预后的独立危险因素。通过纳入患者年龄、T 分期、N 分期及风险模型构建Nomogram,最后利用基因富集分析(gene set enrichment analysis,GSEA) 软件对预后相关的衰老基因进行功能富集分析。结果 共筛选出119 个表达差异的衰老基因,单因素Cox 回归筛选出10 个预后相关的衰老基因,其中包含2 个抑癌基因(NRG1、IL2RG)和8 个促癌基因(EIF4EBP1、MMP1、PLAU、MMP13、RAD51、FGF7、DLL3、IGFBP1)。通过Lasso 回归构建10 基因预测模型,发现高低风险组之间的预后存在显著差异(P < 0.001)。Nomogram 模型对乳腺癌患者3 年的预测准确性高。GSEA 发现高风险患者的基因显著富集在细胞周期、同源重组等信号通路等。而低风险患者的基因显著富集在JAK-STAT 信号通路、细胞因子- 受体- 相互作用等信号通路中。结论 基于衰老相关基因构建的模型对预测乳腺癌患者的预后有良好的效能。

关键词: 乳腺癌, 生物信息学, 衰老基因, 预后

Abstract:

Objective To screen out the aging genes that are significantly related to the prognosis of breast cancer through bioinformatics. Method Clinical data and mRNA sequencing data were downloaded from the cancer genome atlas (TCGA) database collected by the National Cancer Center from September, 2010 to June, 2015. Aging genes were downloaded from the Aging Atlas database. Differential express genes between the normal tissue and the cancer tissue were compared. Single factor Cox and Lasso regression were applied to obtain prognostic-related aging genes. Then the signature was constructed, the patients were divided into high-risk group and low-risk group with the median of risk coefficient as the cut-off value. Univariate and multivariate regression were used to identify the independent factor, then Nomogram was constructed. Gene set enrichment analysis (GSEA) software was used for the functional enrichment analysis of key genes. Result 119 differential aging genes and 10 prognostic-related genes were obtained, including 2 tumor suppressor genes(NRG1, IL2RG) and 8 cancer-promoting genes (EIF4EBP1,MMP1, PLAU, MMP13, RAD51, FGF7, DLL3, IGFBP1). The 10-aging-gene signature was constructed by Cox. The prognosis between the high risk and low risk group was significantly differently. The Nomogram showed good performance in predicting the overall survival of breast cancer patient. The GSEA showed the high-risk group was significantly enriched in signal pathways such as cell cycle and homologous recombination. The genes of lowrisk group was significantly enriched in the JAK-STAT signaling pathway, cytokine-receptor-interaction pathway. Conclusion The signature based on aging related genes had a good performance on predicting the prognosis of breast cancer patients.

Key words: Breast cancer, Bioinformatics, Aging genes, Prognosis