切换至 "中华医学电子期刊资源库"
高级检索
中华腔镜泌尿外科杂志(电子版)

中华腔镜泌尿外科杂志(电子版) ›› 2024, Vol. 18 ›› Issue (02) : 131 -140. doi: 10.3877/cma.j.issn.1674-3253.2024.02.003

临床研究

机器学习模型评估RAS亚家族基因对膀胱癌免疫治疗的作用
黄艺承1, 梁海祺1, 何其焕1, 韦发烨1, 杨舒博1, 谭舒婷1, 翟高强1, 程继文1,†()   
  1. 1. 530021　南宁，广西医科大学第一附属医院泌尿外科
  • 收稿日期:2024-01-02 出版日期:2024-04-01
  • 通信作者: 程继文
  • 基金资助:
    国家自然科学基金(82160483); 广西科技基地和人才专项(Guike20238090); 广西泌尿系统疾病临床医学研究中心(Guike20297081)

Machine learning models assess the roles of RAS subfamily genes in immunotherapy based on bladder cancer

Yicheng Huang1, Haiqi Liang1, Qihuan He1, Faye Wei1, Shubo Yang1, Shuting Tan1, Gaoqiang Zhai1, Jiwen Cheng1,†()   

  1. 1. Department of Urology, the First Affiliated Hospital of Guangxi Medical University, Nanning 530021, China
  • Received:2024-01-02 Published:2024-04-01
  • Corresponding author: Jiwen Cheng
全文 ( ) 下载PDF ( ) 导出引用
引用本文：

黄艺承, 梁海祺, 何其焕, 韦发烨, 杨舒博, 谭舒婷, 翟高强, 程继文. 机器学习模型评估RAS亚家族基因对膀胱癌免疫治疗的作用[J]. 中华腔镜泌尿外科杂志(电子版), 2024, 18(02): 131-140.

Yicheng Huang, Haiqi Liang, Qihuan He, Faye Wei, Shubo Yang, Shuting Tan, Gaoqiang Zhai, Jiwen Cheng. Machine learning models assess the roles of RAS subfamily genes in immunotherapy based on bladder cancer[J]. Chinese Journal of Endourology(Electronic Edition), 2024, 18(02): 131-140.

目的

分析RAS亚家族基因在膀胱癌中的表达，探究RAS亚家族基因对免疫治疗应答的预测作用。

方法

本研究主要基于癌症基因组图谱（TCGA）数据库的407例膀胱癌和19例癌旁正常组织的mRNA测序数据。首先在转录水平对比35个RAS亚家族基因在肿瘤和正常组织的表达差异，并在人类蛋白质图谱数据库中使用膀胱癌和癌旁正常组织的免疫组织化学染色图像进行蛋白质水平验证。采用单因素和多因素Cox回归分析评估RNA亚家族基因的预后作用。GSE32894数据集308例膀胱癌作为预后分析的验证集。使用CIBERSORT算法计算TCGA肿瘤样本的肿瘤免疫浸润细胞丰度。基于IMvigor210试验的298例膀胱癌患者免疫治疗应答结果，使用随机森林（RF）模型筛选关键基因，并用逻辑回归（LR）、支持向量机（SVM）、人工神经网络（ANN）模型进行验证。GSE145140数据集的1例膀胱癌单细胞RNA测序数据，评估关键基因在膀胱癌单细胞中的表达情况。

结果

多数RAS亚家族基因在膀胱癌中表达发生改变，并可作为生存预后的预测因素。RAS亚家族基因与M1巨噬细胞浸润的相关性较强，并与一些免疫检查点分子共表达。RHEBL1，RRAD，GEM，RAP2B，HRAS，RERG，NRAS被RF模型认为是最能预测免疫治疗应答的关键基因。

结论

本研究初步确认RAS亚家族基因的表达可作为膀胱癌的预后因子，并在肿瘤免疫细胞浸润和免疫治疗应答中发挥作用。

关键词: 膀胱癌, 癌症基因组图谱, RAS亚家族基因, 人类蛋白质图谱, 机器学习, 免疫治疗, 抗程序性细胞死亡受体/配体, 免疫检查点
Objective

To analyze the expression of RAS subfamily genes in bladder cancer, and explore the predictive effect of RAS subfamily genes on immunotherapy response.

Methods

This study was mainly based on the mRNA sequencing data of 407 cases of bladder cancer and 19 cases of adjacent normal tissues from the cancer genome atlas (TCGA) database. First, the expression differences of 35 RAS subfamily genes in tumor and normal tissues were compared at the transcriptional level, and the immunohistochemical staining images of bladder cancer and adjacent normal tissues were used in the human protein map database for protein level verification. Use univariate and multivariate Cox regression analysis to evaluate the prognostic role of RNA subfamily genes. GSE32894 data set 308 cases of bladder cancer were used as the validation set for prognostic analysis. Calculate the abundance of tumor immune infiltrating cells in TCGA tumor samples using the CIBERSORT algorithm. Based on the immunotherapeutic response results of 298 bladder cancer patients in IMvigor210 trial, random forest (RF) model was used to screen key genes, and logical regression (LR), support vector machine (SVM), and artificial neural network (ANN) models were used to verify. The single cell RNA sequencing data of one case of bladder cancer in GSE145140 dataset were used to evaluate the expression of key genes in the single cell of bladder cancer.

Results

The expression of lots of RAS subfamily genes were altered in bladder cancer and could be used as predictors of prognosis. RAS subfamily genes were strongly associated with M1 macrophage infiltration and co-expressed with some immune checkpoint molecules. RHEBL1, RRAD, GEM, RAP2B, HRAS, RERG, and NRAS were considered by the RF model to be the key genes that can best predict the response to immunotherapy.

Conclusion

This study preliminarily confirms that the expression of RAS subfamily genes can be used as prognostic factors for bladder cancer, and play a role in tumor immune cells infiltration and immunotherapy response.

Key words: Bladder cancer, TCGA, RAS subfamily genes, The Human Protein Atlas, Machine learning, Immunotherapy, PD-1/PD-L1, Immune checkpoint
图1 RAS亚家族基因在膀胱癌和癌旁正常组织中表达量对比
图2 单因素Cox回归分析TCGA-BLCA数据集中RAS亚家族基因对膀胱癌患者的总体生存预后影响
表1 TCGA-BLCA数据集中RAS亚家族基因及TNM病理分期对膀胱癌预后影响的单因素和多因素Cox回归分析
基因及病理分型分期 单因素Cox回归分析 多因素Cox回归分析
HR(95%CI) P值 HR(95%CI) P值
RAP2A 1.3(1.0～1.6) 0.046 0.79(0.46～1.36) 0.394
DIRAS1 1.1(1.0～1.1) 0.022 1.12(0.97～1.30) 0.122
RASL10B 1.1(1.0～1.2) 0.007 1.22(1.01～1.48) 0.041
RASL10A 0.9(0.8～1.0) 0.017 0.81(0.66～1.00) 0.051
NKIRAS1 1.3(1.0～1.7) 0.018 1.01(0.63～1.61) 0.981
NKIRAS2 1.6(1.1～2.3) 0.013 2.12(1.02～4.40) 0.044
RASL12 1.1(1.0～1.2) 0.020 0.95(0.71～1.28) 0.756
GEM 1.1(1.0～1.2) 0.013 1.11(0.88～1.41) 0.386
DIRAS3 1.1(1.0～1.2) 0.012 0.94(0.74～1.20) 0.643
病理分型 0.7(0.5～1.0) 0.023 1.09(0.56～2.12) 0.811
病理M分期 3.3(1.6～6.9) 0.002 0.73(0.22～2.46) 0.608
病理N分期 1.6(1.3～1.9) <0.001 1.53(0.90～2.58) 0.112
病理T分期 1.7(1.4～2.1) <0.001 1.37(0.77～2.44) 0.292
TNM总体分期 1.7(1.4～2.1) <0.001 0.94(0.45～1.95) 0.859
表2 GSE32894数据集中RAS亚家族基因对膀胱癌预后影响的单因素和多因素Cox回归分析
基因 单因素Cox回归分析 多因素Cox回归分析
HR(95%CI) P值 HR(95%CI) P值
RAP2A 1.8(0.9～3.6) 0.089 1.65(0.75～3.65) 0.214
RASL10A 1.5(1.0～2.3) 0.079 1.35(0.81～2.26) 0.243
NKIRAS1 1.3(0.6～2.8) 0.581 1.08(0.45～2.59) 0.871
NKIRAS2 3.5(1.3～10.0) 0.016 3.66(1.29～10.36) 0.015
RASL12 1.9(1.1～3.2) 0.021 1.31(0.62～2.77) 0.481
GEM 3.8(1.5～9.5) 0.005 2.15(0.64～7.24) 0.218
DIRAS3 1.4(0.1～19.0) 0.812 0.56(0.05～6.55) 0.645
图3 RAS亚家族基因表达量与免疫细胞、免疫检查点分子表达相关度热图注：a为RAS亚家族基因表达量与肿瘤浸润免疫细胞丰度的相关度；b为RAS亚家族基因与免疫检查点分子表达的相关度；方格中标注*代表P<0.05，蓝色代表正相关（R>0），红色代表负相关（R<0）
图4 RRAS、MRAS蛋白在膀胱肿瘤和正常尿路上皮中染色强度的对比注：a为RRAS蛋白在膀胱肿瘤中的染色强度为阴性；b为RRAS蛋白在膀胱正常尿路上皮中的染色强度为弱；c为MRAS蛋白在膀胱肿瘤中的染色强度为弱；d为MRAS蛋白在膀胱正常尿路上皮中的染色强度为中度
表3 膀胱肿瘤和癌旁正常组织RRAS和MRAS免疫组织化学染色强度对比
染色强度 RRAS(例) MRAS(例)
肿瘤(n=12) 癌旁(n=2) 肿瘤(n=10) 癌旁(n=2)
阴性 6 0 0 0
弱 5 2 3 0
中性 1 0 5 2
强 0 0 2 0
图5 机器学习模型基于RAS亚家族基因对膀胱癌免疫治疗应答进行预测和模型验证注：a为使用RF模型构建决策树；b为RF模型识别出预测免疫治疗应答的最佳基因；c～e为基于RF模型识别的7个关键基因（RHEBL1，RRAD，GEM，RAP2B，HRAS，RERG，NRAS），分别使用RF、LR、SVM和ANN模型进行预测准确度验证；f～g为使用决策曲线验证RF模型的可靠性
图6 随机森林（RF）模型中RAS亚家族关键基因在膀胱癌单细胞中的表达模式注：a示膀胱癌单细胞RNA测序数据集识别出9个细胞簇，包括膀胱肿瘤细胞、正常尿路上皮细胞、内皮细胞、肌细胞、成纤维细胞、T细胞、巨噬细胞和两个未知细胞簇；b～h示7个关键基因（RHEBL1，RRAD，GEM，RAP2B，HRAS，RERG，NRAS）分别在上述9个细胞簇中的表达量
[1]
Colicelli J. Human RAS superfamily proteins and related GTPases [J]. Sci STKE, 2004, 2004(250): Re13.
[2]
Perucho M, Goldfarb M, Shimizu K, et al. Human-tumor-derived cell lines contain common and different transforming genes [J]. Cell, 1981, 27(3 Pt 2): 467-476.
[3]
Oxford G, Theodorescu D. The role of RAS superfamily proteins in bladder cancer progression [J]. J Urol, 2003, 170(5): 1987-1993.
[4]
Goecks J, Jalili V, Heiser LM, et al. How machine learning will transform biomedicine [J]. Cell, 2020, 181(1): 92-101.
[5]
Ming C, Viassolo V, Probst-Hensch N, et al. Machine learning-based lifetime breast cancer risk reclassification compared with the BOADICEA model: impact on screening recommendations [J]. Br J Cancer, 2020, 123(5): 860-867.
[6]
Kent WJ, Sugnet CW, Furey TS, et al. The human genome browser at UCSC [J]. Genome Res, 2002, 12(6): 996-1006.
[7]
Sjödahl G, Lauss M, Lövgren K, et al. A molecular taxonomy for urothelial carcinoma [J]. Clin Cancer Res, 2012, 18(12): 3377-3386.
[8]
Lee HW, Chung W, Lee HO, et al. Single-cell RNA sequencing reveals the tumor microenvironment and facilitates strategic choices to circumvent treatment failure in a chemorefractory bladder cancer patient [J]. Genome Med, 2020, 12(1): 47.
[9]
Mariathasan S, Turley SJ, Nickles D, et al. TGF β attenuates tumour response to PD-L1 blockade by contributing to exclusion of T cells [J]. Nature, 2018, 554(7693): 544-548.
[10]
Chen B, Khodadoust MS, Liu CL, et al. Profiling tumor infiltrating immune cells with CIBERSORT [J]. Methods Mol Biol, 2018, 1711: 243-259.
[11]
Ru B, Wong CN, Tong Y, et al. TISIDB: an integrated repository portal for tumor-immune system interactions [J]. Bioinformatics, 2019, 35(20): 4200-4202.
[12]
Kaufman DS, Shipley WU, Feldman AS. Bladder cancer [J]. Lancet, 2009, 374(9685): 239-249.
[13]
Bellmunt J, de Wit R, Vaughn DJ, et al. Pembrolizumab as second-line therapy for advanced urothelial carcinoma [J]. N Engl J Med, 2017, 376(11): 1015-1026.
[14]
Powles T, Park SH, Voog E, et al. Avelumab maintenance therapy for advanced or metastatic urothelial carcinoma [J]. N Engl J Med, 2020, 383(13): 1218-1230.
[15]
Bernal Astrain G, Nikolova M, Smith MJ. Functional diversity in the RAS subfamily of small GTPases [J]. Biochem Soc Trans, 2022, 50(2): 921-933.
[16]
Zhang J, Lodish HF. Constitutive activation of the MEK/ERK pathway mediates all effects of oncogenic H-ras expression in primary erythroid progenitors [J]. Blood, 2004, 104(6): 1679-1687.
[17]
Thomas J, Sonpavde G. Molecularly targeted therapy towards genetic alterations in advanced bladder cancer [J]. Cancers (Basel), 2022, 14(7): 1795.
[18]
Biswas PK, Kwak Y, Kim A, et al. TTYH3 modulates bladder cancer proliferation and metastasis via FGFR1/H-Ras/A-Raf/MEK/ERK pathway [J]. Int J Mol Sci, 2022, 23(18): 10496.
[19]
Zhou X, Qiu G, Yang Y, et al. Circ_0001955 promotes non-small cell lung cancer cell proliferation and invasion by regulating MiR-29a-3p/NKIRAS2 axis to activate the nuclear factor-κB pathway [J]. Pathol Int, 2023, 73(9): 434-443.
[20]
Coelho MA, de Carné Trécesson S, Rana S, et al. Oncogenic RAS signaling promotes tumor immunoresistance by stabilizing PD-L1 mRNA [J]. Immunity, 2017, 47(6): 1083-99.e6.
[21]
Wang HY, Wu XY, Zhang X, et al. Prevalence of NRAS mutation, PD-L1 expression and amplification, and overall survival analysis in 36 primary vaginal melanomas [J]. Oncologist, 2020, 25(2): e291-e301.
[22]
Bodhale N, Nair A, Saha B. Isoform-specific functions of Ras in T-cell development and differentiation [J]. Eur J Immunol, 2023, 53(7): e2350430.
[23]
Cao Y, Liang W, Fang L, et al. PD-L1/PD-L1 signalling promotes colorectal cancer cell migration ability through RAS/MEK/ERK [J]. Clin Exp Pharmacol Physiol, 2022, 49(12): 1281-1293.
[24]
Gao C, Chen J, Bai J, et al. High glucose-upregulated PD-L1 expression through RAS signaling-driven downregulation of PTRH1 leads to suppression of T cell cytotoxic function in tumor environment [J]. J Transl Med, 2023, 21(1): 461.
[25]
Chen JX, Chen DM, Wang D, et al. METTL3/YTHDF2 m6A axis promotes the malignant progression of bladder cancer by epigenetically suppressing RRAS [J]. Oncol Rep, 2023, 49(5): 94.
[26]
Zhao A, Li D, Mao X, et al. GNG2 acts as a tumor suppressor in breast cancer through stimulating MRAS signaling [J]. Cell Death Dis, 2022, 13(3): 260.
[27]
Handelman GS, Kok HK, Chandra RV, et al. eDoctor: machine learning and the future of medicine [J]. J Intern Med, 2018, 284(6): 603-619.
[28]
Zhang M, Liu Y, Yao J, et al. Value of machine learning-based transrectal multimodal ultrasound combined with PSA-related indicators in the diagnosis of clinically significant prostate cancer [J]. Front Endocrinol (Lausanne), 2023, 14: 1137322.
[29]
Feld E, Harton J, Meropol NJ, et al. Effectiveness of first-line immune checkpoint blockade versus carboplatin-based chemotherapy for metastatic urothelial cancer [J]. Eur Urol, 2019, 76(4): 524-532.
[30]
Sonpavde G, Dranitsaris G, Necchi A. Improving the cost efficiency of PD-1/PD-L1 inhibitors for advanced urothelial carcinoma: a major role for precision medicine? [J]. Eur Urol, 2018, 74(1): 63-65.
[31]
Deng X, Li T, Mo L, et al. Machine learning model for the prediction of prostate cancer in patients with low prostate-specific antigen levels: a multicenter retrospective analysis [J]. Front Oncol, 2022, 12:985940.
[32]
Fu Y, Liu K, Jiang X, et al. Hsa_circ_0008035 knockdown inhibits bladder cancer progression through miR-1184/RAP2B axis [J]. Urol Int, 2023, 107(6): 632-645.
[33]
Xiong Y, Huang H, Chen S, et al. ERK5-regulated RERG expression promotes cancer progression in prostatic carcinoma [J]. Oncol Rep, 2019, 41(2): 1160-1168.
[1] 池梦婷, 赵萍, 李静波, 朱嘉宁, 宋禄达, 罗渝昆, 李秋洋. 高帧频超声造影在膀胱癌分级及分期中的应用价值[J]. 中华医学超声杂志（电子版）, 2023, 20(11): 1107-1113.
[2] 尚峰进, 陈陆尧, 刘亚星, 张浩然, 连长红. 肿瘤相关中性粒细胞在胃癌发生发展和治疗中的研究进展[J]. 中华普通外科学文献(电子版), 2024, 18(01): 58-61.
[3] 曹飞, 庞俊. 前列腺癌免疫微环境中免疫抑制性细胞分类及其作用机制[J]. 中华腔镜泌尿外科杂志(电子版), 2024, 18(02): 121-125.
[4] 邓新军, 李正明, 李文彬. 广东省医学会泌尿外科疑难病例多学科会诊（第14期）——左肾原发恶性肿瘤并发于肺癌并脑转移[J]. 中华腔镜泌尿外科杂志(电子版), 2024, 18(01): 114-117.
[5] 张逸, 张继, 栾成明, 张传猛. 肿瘤定量参数对VI-RADS评分系统评估膀胱癌的辅助价值[J]. 中华腔镜泌尿外科杂志(电子版), 2024, 18(01): 41-45.
[6] 王浩, 王卓, 王琦, 高金莉, 田新涛, 张文元, 蒋文惠, 陆佳荪, 杨国胜, 温机灵. 经尿道激光操作架直视推拨法铥激光整块切除术治疗非肌层浸润性膀胱癌的初步经验[J]. 中华腔镜泌尿外科杂志(电子版), 2024, 18(01): 31-35.
[7] 朱迎, 赵征, 许达, 陆录, 殷保兵. 免疫检查点抑制剂治疗肝细胞癌的进展与展望[J]. 中华肝脏外科手术学电子杂志, 2024, 13(01): 5-10.
[8] 张占国. 靶向免疫治疗时代的肝癌肝切除术再思考[J]. 中华肝脏外科手术学电子杂志, 2024, 13(01): 11-15.
[9] 张宇, 余灵祥, 杨永平, 赵德希, 刁广浩, 杨木易, 赵亮, 刘佳, 李鹏, 张宁, 任辉. 原发性肝癌Ⅲa期降期后肝切除临床疗效分析[J]. 中华肝脏外科手术学电子杂志, 2024, 13(01): 78-82.
[10] 严帅, 岳志强, 赵江华, 陈琳, 吴金柱. 初始不可切除肝癌患者靶向免疫联合治疗后手术切除临床疗效[J]. 中华肝脏外科手术学电子杂志, 2024, 13(01): 83-87.
[11] 卓长华, 叶韵斌, 陈昌江, 简锦亮, 王志纬. 林奇综合征相关性异时性结直肠癌的治疗[J]. 中华结直肠疾病电子杂志, 2024, 13(01): 32-37.
[12] 赵海清, 张威, 李琴. 肌苷联合免疫检查点抑制剂在转移性结直肠癌患者中的临床疗效观察[J]. 中华结直肠疾病电子杂志, 2024, 13(01): 54-62.
[13] 段福孝, 王鑫宇, 孙爽, 于知宇, 张成. 结直肠癌患者周围神经侵犯预测模型的建立与评价[J]. 中华临床医师杂志(电子版), 2023, 17(11): 1154-1162.
[14] 张可, 闫琳琳, 王鹏飞, 章秀林, 赵帆, 胡守奎. 外泌体环状RNA在肿瘤免疫和癌症免疫治疗中的作用[J]. 中华临床医师杂志(电子版), 2023, 17(10): 1102-1108.
[15] 吕泉龙, 史文杰, 孙文国. 免疫检查点抑制剂在治疗转移性去势抵抗性前列腺癌中的研究进展[J]. 中华诊断学电子杂志, 2024, 12(01): 69-72.
阅读次数
全文


摘要

版权所有 中华医学会 中华医学电子音像出版社有限责任公司  京ICP备14006079号-1  网络出版服务许可证：(署)网出证(京)字第075号