摘要目的:利用基底膜相关基因(BMRG)构建一个新的预后风险模型，以探索乳腺癌与基底膜之间的关系。方法:从癌症基因组图谱(TCGA)和基因表达综合数据库(GEO)中收集转录组和临床数据，将TCGA数据库作为训练集，GEO数据库作为验证集，应用单因素Cox回归、最小绝对收缩和选择算子(LASSO)和多因素Cox回归分析建立BMRG预后模型。通过Kaplan-Meier方法和受试者操作特征(ROC)曲线进一步验证和评估风险模型。然后结合风险模型和临床特征构建列线图来预测乳腺癌的总生存率。通过基因集富集分析(GSEA)研究其可能参与的生物学途径。同时使用Wilcoxon秩和检验评估高风险组和低风险组患者对药物的敏感性差异。结果:共鉴定了193个差异表达基因，并构建了基于8个BMRG的风险模型，包括 COL6A2、 CTSA、 EVA1B、 ITGAX、 MMP-1、 ROBO3、 SDC1和 UNC5A。Kaplan-Meier生存曲线和ROC曲线分析表明，该模型可以很好地预测乳腺癌的预后，曲线下面积为0.779，表明准确度也很高。此外，列线图也显示出了良好的预测一致性和临床净收益。单因素和多因素Cox回归分析验证了BMRG模型是乳腺癌的独立危险因素。GSEA显示高风险组主要富集在细胞外基质受体相互作用通路。此外，高风险患者对紫杉类化疗药物和靶向治疗药物的敏感性更高，而低风险患者对吉西他滨和雷帕霉素的敏感性更高。 结论:基于8种BMRG构建的风险模型可作为乳腺癌的有效预后指标，可以提高临床医师对患者治疗反应的预测。

abstractsObjective:To construct a novel prognostic risk model using basement membrane-related genes (BMRG) to explore the relationship between breast cancer and basement membrane.Methods:Transcriptome and clinical data were collected from The Cancer Genome Atlas (TCGA) and the Gene Expression Omnibus (GEO) database, the TCGA data was used as the training set and the GEO database as the validation set. Then univariate Cox regression, least absolute shrinkage and selection operator (LASSO) and multivariate Cox regression analyses were applied to build a BMRG prognostic model. The risk model was further validated and evaluated by Kaplan-Meier method and receiver operating characteristic (ROC) curve. The risk model and clinical characteristics were then combined to construct a nomogram to predict the overall survival of breast cancer. The biological pathways that may be involved were investigated by gene set enrichment analysis (GSEA). In addition, the differences in drug sensitivity between high-risk and low-risk groups of patients by the Wilcoxon rank sum test.Results:A total of 193 differentially expressed genes were identified, and risk models based on eight BMRG was constructed, including COL6A2, CTSA, EVA1B, ITGAX, MMP-1, ROBO3, SDC1, and UNC5A. Kaplan-Meier and ROC analyses showed that the model could well predict the prognosis of breast cancer, with an area under the curve of 0.779, indicating a high degree of accuracy as well. In addition, the nomogram showed good predictive consistency and net clinical benefit. Univariate and multivariate Cox regression analyses validated the BMRG model as an independent risk factor for breast cancer. GSEA analysis showed that the high-risk group was predominantly enriched in the extracellular matrix receptor interaction pathway. In addition, high-risk patients were more sensitive to taxanes chemotherapeutic agents and targeted therapeutic agents, while low-risk patients were more sensitive to gemcitabine and rapamycin. Conclusion:The risk model constructed based on eight BMRG can be used as a valid prognostic indicator for breast cancer and can improve the prediction of patient response to treatment.