基于血清外泌体环状RNA的胃癌早期筛查分类器的构建及评估

There is no simple and effective diagnostic method for the general population of gastric cancer. Therefore, it is of great significance to search for tumor markers with high sensitivity and specificity for early diagnosis and screening of gastric cancer. Circular RNA (circRNA) is a new kind of non-coding RNA that can be detected in the serum. It is not sensitive to nuclease, is more stable than plasma miRNA and long non-coding RNA, and has higher abundance in the serum. More importantly, previous studies have shown that intracellular circRNA has been shown to be a candidate marker for early detection of tumors. However, the invasive detection of intracellular circRNA limits its application. In this study, the exosome circRNA was taken as our research object. The exosomes in serum were firstly separated by the new membrane separation technique and then the high-throughput sequencing technique was applied to systematically identify the circRNA profiles of gastric adenocarcinoma, gastric ulcer, polyp of stomach, atrophic gastritis and normal samples. The differentially expressed circRNAs combined with the machine learning algorithm were used to construct the gastric early screening model, which is used in the early screening of gastric cancer in order to reduce the incidence and mortality of gastric cancer in China. At the same time, we study the function of abnormal exosome circRNA in serum with gastric cancer, and it is expected to find new therapeutic targets for gastric cancer.

胃癌目前尚无简便、有效的诊断方法进行全体人群的普查，寻找新的标志物对胃癌的早期诊断或筛查有重要意义。circRNA是一种新兴的非编码RNA分子，对核酸酶不敏感，相对于血浆miRNA、长非编码RNA更为稳定，且circRNA在血清外泌体中丰度较高。前期研究表明细胞内的circRNA已初步证明能够作为肿瘤早期检测的候选标志物，缺点是需要有创的从组织中提取检测。目前对于胃癌外泌体中的circRNA仍缺乏系统地研究。本课题以外泌体的circRNA为研究对象，采用新型膜分离技术分离血清中的外泌体，利用高通量测序技术，系统地比较胃腺癌和非癌样本外泌体中circRNA的表达谱，筛选胃癌样本中差异表达的circRNA并联合机器学习算法构建胃癌早期筛查分类器模型，应用于胃癌早期筛查，从而有助于降低我国胃癌的发生率和死亡率。同时研究胃癌病人外泌体中水平异常circRNA的功能，有望发现新的胃癌治疗靶点。

项目基于胞外囊泡来源的circRNA测序以及全转录组测序的数据，建立一套能够通过患者血浆检测来诊断胃癌发生的分类器，并寻找可能存在生物学功能的circRNA，验证其发挥的生物学功能。. 在临床样本高通量测序分析后，对在肿瘤发生发展过程中存在表达差异的血浆来源 EVs 中的 circRNA 分子进行标注，从数据中一共识别出4268个源自不同染色体位置的circRNA。其中，在对肿瘤样本、非肿瘤胃部炎症样本、无胃部疾病健康样本三个分组的circRNA表达情况进行分析后，识别了148个存在显著差异的circRNA分子，并结合对对应样本的全转录组测序分析，筛选出32个肿瘤分类器候选circRNA分子。随后我们设计了一套稳定高效的提取血浆EVs来源RNA，逆转录-荧光定量PCR的检测体系，扩大检测样本量并检测这32个circRNA分子在临床样本中表达情况，利用4种不同的机器学习算法（非线性支持向量机 [nSVM]、线性支持向量机 [SVM]、逻辑回归 [LR]、线性判别分析 [LDA]），构建了4组对不同分组间区分效果较好的分类器模型，最终选择基于非线性SVM及8个circRNAs构建的模型作为最终分类器模型。分类器模型经交叉验证后，其准确性为95.3%、敏感性为93.3%，特异性为96.4%。. 在完成临床样本检测后，利用生信分析以及预实验筛选出了3个在测序结果中显著差异表达的circRNA中被预测到可能具有生物学功能的分子，并在胃癌细胞株中进行细胞水平的功能研究（3个分子分别标注为CR-12、CR-15、CR-21），结果发现在胃癌细胞中对细胞的增殖、迁移、凋亡等功能上发挥着重要的调控作用。. 因此，本项目较好地完成以下3个研究目标：①构建血浆来源EVs circRNA预测胃癌分类器； ②优化并评价胃癌分类器的临床检测价值；③验证 CR-12、CR-15、CR-21参与了胃癌细胞系的细胞表型相关调控。