Department of Urology, Southern Medical University, Guangzhou, China.
Department of Urology, The Cancer Hospital of the University of Chinese Academy of Sciences (Zhejiang Cancer Hospital), Hangzhou, China.
J Immunol Res. 2022 Apr 12;2022:1793005. doi: 10.1155/2022/1793005. eCollection 2022.
Bladder cancer (BLCA) is one of the most common cancers and ranks ninth among all cancers. Extracellular matrix (ECM) genes activate a number of pathways that facilitate tumor development. This study is aimed at providing models to predict BLCA survival and recurrence by ECM genes.
Expression data from BLCA samples in GSE32894, GSE13507, GSE31684, GSE32548, and TCGA-BLCA cohorts were downloaded and analyzed. The ECM-related genes were obtained by differentially expressed gene analysis, stage-associated gene analysis, and random forest variable selection. The ECM was constructed in GSE32894 by the hub ECM-related genes and validated in GSE13507, GSE31684, GSE32548, and TCGA-BLCA cohorts. The correlations of the ECM score with cells (T cells, fibroblasts, etc.) and the response to immunotherapeutic drugs were investigated. Four machine learning models were selected and used to construct models to predict the recurrence of BLCA. A total of 15 paired BLCA and normal tissue specimens, human immortalized uroepithelial cell lines, and bladder cancer cell lines were selected for the validation of the difference in expression of FSTL1 between normal tissues and BLCA.
Six ECM genes (CTHRC1, MMP11, COL10A1, FSTL1, SULF1, and COL5A3) were recognized to be the hub ECM-related genes. The ECM score of each BLCA patient was calculated using these six selected ECM-related genes. BLCA patients with a high ECM score group had significantly lower overall survival rates than patients in the low ECM score group. We found that the ECM score was positively associated with immune cells and fibroblasts and negatively correlated with tumor purity. When treated with immunotherapy, BLCA patients with a high ECM score presented a high response rate and better prognosis. We also found that the combination of FSTL1, stage, age, and gender achieved an AUC value of 0.76 in predicting bladder cancer recurrence. Based on the RT-qPCR results of FSTL1 gene expression, there was an overall decrease in the mRNA expression of FSTL1 in cancer tissues compared to their adjacent normal tissues. Subsequent validation demonstrated that the FSTL1 expression was downregulated at the gene and protein level compared to that in SVH cells.
Taken together, our results indicate that ECM-related genes correlate with immune cells, overall survival, and recurrence of BLCA. This study provides a machine learning model for predicting the survival and recurrence of BLCA patients.
膀胱癌（BLCA）是最常见的癌症之一，在所有癌症中排名第九。细胞外基质（ECM）基因激活了许多促进肿瘤发展的途径。本研究旨在通过 ECM 基因为 BLCA 的生存和复发提供预测模型。
从 GSE32894、GSE13507、GSE31684、GSE32548 和 TCGA-BLCA 队列中下载并分析 BLCA 样本的表达数据。通过差异表达基因分析、阶段相关基因分析和随机森林变量选择获得 ECM 相关基因。在 GSE32894 中，通过核心 ECM 相关基因构建 ECM，并在 GSE13507、GSE31684、GSE32548 和 TCGA-BLCA 队列中进行验证。研究 ECM 评分与细胞（T 细胞、成纤维细胞等）的相关性以及对免疫治疗药物的反应。选择了四个机器学习模型来构建预测 BLCA 复发的模型。共选择了 15 对 BLCA 和正常组织标本、人永生化尿路上皮细胞系和膀胱癌细胞系，用于验证 FSTL1 在正常组织和 BLCA 之间表达的差异。
鉴定出 6 个 ECM 基因（CTHRC1、MMP11、COL10A1、FSTL1、SULF1 和 COL5A3）为核心 ECM 相关基因。使用这 6 个选择的 ECM 相关基因计算每个 BLCA 患者的 ECM 评分。高 ECM 评分组的 BLCA 患者总生存率明显低于低 ECM 评分组。我们发现 ECM 评分与免疫细胞和成纤维细胞呈正相关，与肿瘤纯度呈负相关。在接受免疫治疗时，高 ECM 评分的 BLCA 患者表现出较高的反应率和较好的预后。我们还发现，FSTL1、分期、年龄和性别组合在预测膀胱癌复发方面的 AUC 值为 0.76。基于 FSTL1 基因表达的 RT-qPCR 结果，与相邻正常组织相比，癌症组织中 FSTL1 的 mRNA 表达总体下降。随后的验证表明，与 SVH 细胞相比，FSTL1 的表达在基因和蛋白质水平均下调。
综上所述，我们的结果表明 ECM 相关基因与免疫细胞、总生存率和 BLCA 的复发相关。本研究为预测 BLCA 患者的生存和复发提供了机器学习模型。
