基于尿液miRNAs差异表达谱的膀胱癌诊断和分期预测模型构建及临床应用研究

Bladder cancer is the most common malignancy of the urinary tract in China. Thirty percents of patients are highly suggestive of muscle invasion when diagnosis, and up to 30% non-muscle invasive patients could also occur with muscle invasion during follow up. Staging of the primary lesion, especially the differentiation of non-muscle invasive bladder cancer (Ta, Tis and T1) and muscle-invasive bladder cancer (pT2-pT4) is pivotal to guide therapeutic approach. Most of the current screening tests are expensive and usually invasive. Although urine markers have been studied extensively to help diagnose bladder cancer, there is still a great challenge owing to lack of specificity, especially markers for staging prediction. The more accurate and non-invasive tools that can effectively accomplish this task remain elusive..It has been reported that miRNAs were upregulated or downregulated in tumor and showed restricted tissue distribution, indicating they could be served as a novel biomarkers for tumor diagnosis. Recently, miRNAs have been found stable in urine and the levels of urine miRNAs correlate with the development of urinary tumor, which indicates a great potential to be useful for diagnosis. Changes of specific urine miRNAs as tumor biomarkers for bladder cancer have been reported. It will open a new realm of possibilities for non-invasive diagnosis and staging prediction of bladder cancer based on urine miRNAs profiles, which could help patients prevent delay of effective treatment..To define associations between miRNAs and bladder cancer, we sought to identify the specific urine miRNAs profiles to establish new models for diagnosis and staging predication. In our previous studies, the expression profiles of urine miRNAs have been analyzed by Solexa sequencing technique, and 37 miRNAs were found differently expressed. In the further studies, we will establish a high throughput Luminex technology to confirm the results by detecting each sample, screen the significant miRNAs, and further construct diagnostic model using SVM. Meanwhile, urine miRNAs profiles in different stages of bladder cancer will be analyzed based on the above methods, and the predictive model for muscle-invasive bladder cancer will be established using SVM combined with multivariate analysis by a logistic regression model. The diagnostic and predictive values of these models will be further assessed in clinic, which may show us medically values on diagnosis and staging prediction of bladder cancer and eventually guide therapeutic choices.

膀胱癌是泌尿系统最常见恶性肿瘤，如何能够早期发现并对肌层浸润性膀胱癌（pT2-pT4）进行有效预测是临床尚未解决的难题。近期文献报道，尿液中存在丰富而稳定的miRNAs，其表达水平与泌尿系统恶性肿瘤的发生、发展密切相关，miRNAs具备成为无创性生物标志物的潜能。本项目前期利用Solexa测序技术对膀胱癌尿液miRNAs表达谱进行系统性分析，初步获得了37个差异表达miRNAs。现拟建立针对上述miRNAs的Luminex高通量检测方法，并逐个进行验证分析，筛选差异有统计学意义的miRNAs，进而运用支持向量机构建膀胱癌诊断模型；同时运用上述方法并结合多元logistic回归方程，对不同分期膀胱癌尿液miRNAs进行研究分析，建立肌层浸润性膀胱癌预测模型。探讨尿液miRNAs在膀胱癌无创性诊断及肌层浸润性膀胱癌预测中的临床应用价值，为膀胱癌的早期发现及临床治疗方案的选择提供依据。

膀胱癌在我国的发病率和死亡率均占男性泌尿生殖系统肿瘤的首位，且呈逐年上升趋势。膀胱癌患者初诊时30％已发生肌层浸润，而在非肌层浸润性膀胱癌患者中，仍有30%会复发为浸润性膀胱癌。对于肌层浸润性膀胱癌（pT2-pT4）和非肌层浸润性膀胱癌（Ta、Tis、T1）的鉴别是临床尚未解决的难题，同时也是治疗方案选择的关键。近年来，体液miRNAs作为生物标志物的研究成为肿瘤研究领域的热点。课题前期运用高通量测序发现膀胱癌患者尿液上清中存在miRNAs特异表达谱，为膀胱癌的早期发现及分期预测提供了一条新途径。我们首先建立尿液游离miRNAs检测方法，并采用该方法检测尿液游离miRNAs的表达，发现尿液中存在丰富且稳定的miRNAs，且尿液游离miR-155、miR-214和miR-106b对膀胱癌的早期诊断具有一定的临床应用价值。采用建立的检测方法对膀胱癌尿液miRNAs的特异表达谱进行验证，发现let-7b-5p和miR-532-5p可作为膀胱癌尿液游离miRNAs检测的最佳内参组合，并分别构建基于七个尿液游离miRNAs分子的膀胱癌诊断模型和基于两个尿液游离miRNAs分子的膀胱癌分期预测模型，上述研究成果显著提高了膀胱癌的早期诊断和分期预测水平。此外，我们发现血清miR-210，miR-152，miR-486-3p，miR-422a-3p等可用于膀胱癌的诊断及预后判断。体外功能研究表明miR-203在膀胱癌组织中高表达，并可靶定Bcl-w和Survivin，增强顺铂的细胞毒性，降低膀胱癌对顺铂化疗的敏感性。miR-214作为抑癌基因可通过下调靶基因PDRG1抑制膀胱癌细胞增殖、迁移、侵袭，促进凋亡。本课题的研究成果将为膀胱癌的无创性诊断及分期预测提供依据，并为初步揭示miRNAs在膀胱癌发生发展中的作用机制奠定基础。在研究过程中，我们还发现血清及尿液上清中mRNA及其他非编码RNA（lncRNA和circRNA）也可以作为生物标志物用于膀胱癌的诊断和预后监测，为体液非编码RNA用于膀胱癌的诊断、分期预测及预后监测后续系统研究提供了重要依据和线索。