【作者】 何海飞；

【导师】 邓甬川；

【作者基本信息】 浙江大学 ， 肿瘤学， 2011， 博士

【摘要】 新辅助化疗,又称术前化疗、诱导化疗、初始化疗等,是指对非转移性的肿瘤,在应用局部治疗前进行的全身性、系统性的细胞毒性药物治疗,是近年乳腺癌治疗中的重要内容,已成为中晚期乳腺癌治疗的标准方案之一,而且已经扩展应用于有保乳意愿的早期可手术乳腺癌患者。虽然乳腺癌新辅助化疗在全世界范围内得到了广泛的关注,并进行了大量的基础研究和临床试验,但是乳腺癌新辅助化疗领域还存在许多分歧和争议,主要是完全缓解率低；无术前疗效评价的确切方法；缺少疗效预测和监测手段、不能个体化选择新辅助化疗；新辅助化疗后影响淋巴结分期等等。由于化疗具有个体差异性,即对于不同的乳腺癌患者新辅助化疗的效果差异很大,但是根据目前的检测手段,尚不能在新辅助化疗实施前或是化疗早期就明确患者对化疗方案的敏感性,只能在临床应用2-3疗程化疗后再行评价,若无效再考虑换行其他方案,这在一定程度上会影响到对部分患者病情的控制。同时新辅助化疗可以使一些患者的腋窝淋巴结消失或者降期,因此,准确的乳腺癌术前／新辅助化疗前淋巴结分期受到人们越来越多的重视。表面加强激光解吸电离-飞行时间质谱技术是近年发展起来的一种新的蛋白质组学研究方法,具有操作简单、样品用量少、灵敏度高、高通量等优点,可以一次性获得某一样品中大量的蛋白质数据,并且能够直接对临床标本进行检测,实现了质谱技术用于临床检测的飞跃。国内外不少研究者已成功将这项技术用于多种恶性肿瘤的诊断,但用于检测化疗敏感性预测因子、监测标志物及术前／新辅助化疗前淋巴结分期的相关性研究仍然比较少见。本课题旨在利用SELDI-TOF-MS,通过对乳腺癌患者新辅助化疗前血清进行蛋白质指纹图谱的检测,结合生物信息学的方法,找到有价值的蛋白质指纹图谱化疗敏感性预测模型,提高乳腺癌新辅助化疗的有效率,实现对乳腺癌患者的个体化治疗,避免化疗药物的毒副作用及无效用药；进一步利用血清蛋白质指纹图谱对化疗过程进行监测,动态观测疗效相关因子的变化,筛选出可用于监测疗效的蛋白质标志物；并且运用SELDI-TOF-MS技术来分析鉴定乳腺癌患者不同淋巴结分期蛋白质表达谱的全貌,进行术前／新辅助化疗前分期,用于指导治疗,判断预后。第一部分血清蛋白质指纹图谱在乳腺癌新辅助化疗疗效预测方面的研究为寻找新辅助化疗疗效相关的预测因子,我们进行了本部分研究。采用SELDI-TOF-MS技术及其配套的CM10蛋白质芯片我们检测了参与新辅助化疗的乳腺癌患者(按RECIST标准化疗有效组19例,化疗无效组15例)的血清蛋白质指纹图谱,筛选出6个差异蛋白质峰(质荷比为5869、3291、5894、9296、5953及2220)并建立了预测模型,其中5869 m/z、3291 m/z、5894 m/z、9296 m/z和5953 m/z这5个蛋白质峰在新辅助化疗有效的乳腺癌患者血清中表达较高,而2220m/z则在新辅助化疗无效的患者血清中表达较高。经过交叉验证,该预测模型的灵敏度为89.5%(17/19),特异性为73.3%(11/15),总准确率为82.4%(28/34)。按化疗前后肿块大小实际变化(肿块缩小组有25例,肿块增大组有9例)建立模型,共筛选出了5个有价值的差异蛋白质峰(质荷比为3323、5869、5894、7775及6470),其中3323 m/z、5869 m/z、5894 m/z和7775 m/z这4个蛋白质峰在肿块缩小组的乳腺癌患者血清中表达较高,而6470 m/z则在肿块增大组的患者血清中表达较高。经过交叉验证,该预测模型的灵敏度为96.0%(24/25),特异性为77.8%(7/9),总准确率为91.2%(31/34)5869 m/z和5894 m/z这两个蛋白质峰同时进入了两个疗效预测模型,可能是比较合适的新辅助化疗疗效预测标志物,下一步我们将对这两个蛋白质进行分离、纯化和分析解码,以进一步明确与新辅助化疗敏感或耐药的关系。第二部分血清蛋白质指纹图谱在乳腺癌新辅助化疗过程中的动态监测为了确定乳腺癌患者在新辅助化疗前后血清蛋白质水平上的变化,我们对34例乳腺癌患者新辅助化疗前后的血清进行了蛋白质指纹图谱检测,找出前后表达差异有统计学意义的蛋白质峰,并将蛋白质峰的变化情况在不同疗效组间进行对比,筛选出11个蛋白质峰在新辅助化疗后,两组间的变化趋势不同,因此可能与化疗疗效相关。然后我们进一步考察这11个蛋白质峰在25例参与动态监测的乳腺癌患者中每一例患者每次化疗后的变化情况,形成目标蛋白质在每例患者随化疗疗程表达情况的动态变化趋势图,以后面3次蛋白质峰的平均值与第一次(化疗前)的峰值进行比较,来判断变化趋势是上升还是下降,以减小单次测量所造成的误差。比较每个蛋白质峰的变化趋势在疗效组内和组间的分布情况,找到组内变化情况较一致而组间变化趋势不同的蛋白质峰。最后发现4302 m/z和4645 m/z在有效组中的10/13例(76.9%)患者随化疗进行表达下降,而在无效组中的8/12例(66.7%)患者随化疗进行表达上升；3291 m/z在有效组中的8/13例(61.5%)患者随化疗进行表达下降,而在无效组中的8/12例(66.7%)患者随化疗进行表达上升；5894 m/z在有效组中的8/13例(61.5%)患者随化疗进行表达下降,而在无效组中的9/12例(75%)患者随化疗进行表达上升(见表4.3)。我们认为这4个蛋白质峰在新辅助化疗过程中有监测意义,并且能区分治疗效果。进一步搜索蛋白质数据库,我们发现4302m/z的蛋白质峰可能是肾上腺髓质素-2(adrenomedullin-2, ADM-2),这个蛋白质有促进细胞生长、增殖和存活及促进血管、淋巴管新生的作用；4645 m/z的蛋白质峰可能是MSSP (myc single strand binding protein),作为c-Myc基因的调节子,参与DNA的复制和细胞周期从G1期到S期的转化；3291 m/z的蛋白质峰可能是胰高血糖素(glucagon),有研究发现GLP-2通过对肠上皮完整性的支持,使化疗的有效性不致遭到损坏,从而使肿瘤患者获得治疗方面的优势；5894 m/z这个蛋白质峰可能是活性氧物质调节子1(reactive oxygen species modulator 1),它通过调节内源性活性氧物质(ROS)影响多种信号通路使细胞生长失控ROMO1表达的增加在肿瘤发生、增殖过程中起着重要作用,而且Hwang等人发现ROMO1的表达可能与5-FU(氟尿嘧啶)的耐药有关。下一步我们将对这4个蛋白质进行分离、纯化和分析解码,并进一步研究其与化疗疗效的关系.第三部分血清蛋白质指纹图谱在乳腺癌患者腋窝淋巴结分期方面的研究为了找到准确、微创的新辅助化疗前腋窝淋巴结分期方法,避免化疗后失去区域淋巴结转移情况这一乳腺癌最重要的预后信息,我们进行了此部分研究。比较18例有淋巴结转移乳腺癌患者和16例无淋巴结转移乳腺癌患者的血清蛋白质质谱数据,筛选出5个蛋白质峰的组合(3164 m/z、3979 m/z、9196 m/z、2734m/z和3963 m/z)建立淋巴结转移的预测模型。其中3164 m/z、3979m/z和3963 m/z这3个蛋白质峰在淋巴结有转移的乳腺癌患者血清中表达较高,而9196 m/z和2734 m/z这2个蛋白质峰则在淋巴结无转移的乳腺癌患者血清中表达较高。交叉验证结果显示该模型的灵敏度为88.9%(16/18),特异度为87.5%(14/16),总准确率为88.2%(30/34)进一步比较发现,3979 m/z蛋白质峰的表达水平与淋巴结转移数目相关(R2=0.898)。通过搜索蛋白质数据库我们了解到3979 m/z蛋白质峰可能是一种叫P物质的蛋白质,有研究显示P物质参与乳腺癌细胞的增殖、抵抗放射线损伤、防止凋亡、诱导生长因子和促血管生成因子以及促进乳腺癌骨转移等多种病理进程,因此,P物质与乳腺癌腋窝淋巴结转移的关系值得我们进一步深入研究。结论1、由5869 m/z、3291 m/z、5894 m/z、9296 m/z、5953 m/z和2220 m/z六个蛋白质峰组合所构建的模型能达到预测乳腺癌患者新辅助化疗后临床缓解情况的最佳预测效果,总准确率82.4%。2、由3323 m/z、5869 m/z、5894 m/z、7775 m/z和6470 m/z五个蛋白质峰组合所构建的模型能达到预测乳腺癌患者新辅助化疗后乳腺肿块大小变化的最佳预测效果,总准确率91.2%。3、5869 m/z和5894 m/z这两个蛋白质峰同时进入了两个疗效预测模型,可能是比较合适的新辅助化疗疗效预测标志物,值得进一步研究。4、4302 m/z、4645m/z、3291m/z和5894 m/z这4个蛋白质峰在新辅助化疗有效组和无效组间变化趋势相反,可作为用于新辅助化疗疗效监测的标志物。5、4302 m/z的蛋白质峰可能是肾上腺髓质素-2(adrenomedullin-2,ADM-2);4645 m/z的蛋白质峰可能是MSSP(myc single strand binding protein); 3291m/z的蛋白质峰可能是胰高血糖素(glucagon); 5894 m/z这个蛋白质峰可能是活性氧物质调节子1(Reactive oxygen species modulator 1, ROMO1)。下一步我们将对这4个蛋白质进行分离、纯化和分析解码,并进一步研究其与化疗疗效的关系。6、5894 m/z这个蛋白质峰既可以可以在新辅助化疗前预测疗效,又可以在化疗进行中监测疗效,值得我们在接下来的研究中加以重点关注。7、由3164 m/z、3979 m/z、9196 m/z、2734 m/z和3963 m/z五个蛋白质峰组合所构建的模型能达到预测乳腺癌腋窝淋巴结转移的最佳预测效果,总准确率达88.2%。8、3979 m/z蛋白质峰的表达水平与淋巴结转移数目相关：而且可能是一种叫P物质的蛋白质,它与乳腺癌腋窝淋巴结转移的关系值得我们进一步研究。更多还原

【Abstract】 Neoadjuvant chemotherapy, namely preoperative chemotherapy, induced chemotherapy, initial chemotherapy or primary chemotherapy, has gained wide acceptance for treating patients with locally advanced breast cancer. In these patients the most acknowledged benefit is the reduction in tumor size allowing either a complete resection of an otherwise unresectable tumor or breast conservation surgery in some patients with large tumors. Although many patients benefit from chemotherapy, some fail to respond. However, the success of neoadjuvant chemotherapy in any given individual cannot be predicted. The uncertain benefit for a particular individual, as well as significant toxicity of chemotherapy in all patients, calls for development of methods to select the right patients for treatment and spare those who will not benefit. Meanwhile the application of neoadjuvant chemotherapy can downstage some patients’ axillary lymph node (ALN) status, so people want to find more accurate, and less invasive means of predicting ALN metastasis. A novel technology, surface enhanced laser desorption/ionization-time of flight-mass spectrometry (SELDI-TOF-MS) carved out such a new path to proteomics research on breast cancer. It can directly obtain high-throughput protein profilings from clinical samples with high sensitivity. The application of SELDI-TOF-MS has been reported to be capable of being used to diagnose malignancies of various organs, including the ovary, prostate, and pancreas. But the studys of proteomics to predict or monitor tumor response to chemotherapy are still very rare.Our current study was specifically designed to compare SELDI mass spectra of serum from chemotherapy responsive and non-responsive breast patients receiving the neoadjuvant treatment for identification the protein biomarkers which could predict or monitor tumor response to chemotherapy. We also hypothesized that proteomic alteration in the serum from breast cancer patients could predict ALN metastasis which could be caught by SELDI.Part 1The application of SELDI-TOF-MS in the prediction of tumor response to neoadjuvant chemotherapy in breast cancer patientsSELDI-TOF-MS protein chips were used to detect the serum proteomic patterns of 19 cases of chemotherapy responsive patients and 15 cases of chemotherapy non-responsive patients. Six potential biomarkers were found (5869 m/z,3291 m/z, 5894 m/z,9296 m/z.5953 m/z and 2220 m/z). The protein peaks of 5869 m/z.3291 m/z, 5894 m/z,9296 m/z and 5953 m/z were higher in the chemotherapy responsive patients while the protein peak of 2220 m/z was higher in the chemotherapy non-responsive patients. The prediction model was quite accurate with an accuracy of 82.4%.After neoadjuvant chemotherapy,25 patients got tumor shrinked while 9 patients’ tumor enlarged. With the method of SELDI-TOF-MS-5 protein peak were identified (3323 m/z,5869 m/z,5894 m/z,7775 m/z and 6470 m/z). The protein peaks of 3323 m/z,5869 m/z,5894 m/z and 7775 m/z were higher in the tumor shrinked patients while the protein peak of 6470 m/z was higher in the tumor enlarged patients. This model has an accuracy of 91.2%.As the two models both included the protein peaks of 5869 m/z and 5894 m/z, we think they may be potential biomarkers that can predict tumor response to neoadjuvant chemotherapy.Part 2The application of SELDI-TOF-MS in the monitoring of tumor response to neoadjuvant chemotherapy in breast cancer patientsWe first compared the protein fingerprints of the total 34 cases patients pre-chemotherapy and post-chemotherapy. Eleven discrepant proteins were screened which changed differently between the two groups. Then we checked these proteins in every 25 patients after every period of chemotherapy. The average value of proteins of last 3 periods was compared to the value of the first time in every patient in order to judge the protein peak was rise or drop after receiving chemotherapy. At last we found that the protein peak of 4302 m/z was drop in 76.9% patients of responsive group and rise in 66.7% patients of non-responsive group. So was the protein peak of 4645 m/z. The protein peak of 3291 m/z was drop in 61.5% patients of responsive group and rise in 66.7% patients of non-responsive group, and the protein peak of 5894 m/z was drop in 61.5% patients of responsive group and rise in 75% patients of non-responsive group. We think these 4 protein peaks may serve as biomarkers for monitoring tumor response to chemotherapy.Part 3The application of SELDI-TOF-MS in the ALN staging in breast cancer patientsSELDI-TOF-MS protein chips were used to detect the serum proteomic patterns of 18 cases of patients with ALN metastasis and 16 cases of patients without ALN metastasis. Five potential biomarkers were found (3164 m/z,3979m/z,9196 m/z,2734 m/z and 3963 m/z). The protein peaks of 3164 m/z,3979 m/z and 3963 m/z were higher in the patients with ALN metastasis while the protein peak of 9196 m/z and 2734 m/z were higher in the patients without ALN metastasis. This model has an accuracy of 88.2%.We also find the expression level of the protein peak of 3979 m/z was relevant to the number of metastatic lymph nodes. After searching for the protein database, we know that the protein peak of 3979 m/z maybe the substance P which is deserved for further studies. Conclusions1、The model composed by 5869 m/z,3291 m/z,5894 m/z,9296 m/z,5953 m/z and 2220 m/z could do the best in the division of chemotherapy responsive patients and chemotherapy non-responsive patients. The accuracy was 82.4%.2、The model composed by 3323 m/z,5869 m/z,5894 m/z,7775 m/z and 6470 m/z could do the best in the division of the tumor shrinked patients and the tumor enlarged patients. The accuracy was 91.2%.3、The protein peaks of 5869 m/z and 5894 m/z may be potential biomarkers that can predict tumor response to neoadjuvant chemotherapy.4、The 4 protein peaks (4302 m/z,4645 m/z,3291 m/z and 5894 m/z) may serve as biomarkers for monitoring tumor response to chemotherapy.5、The protein peak of 5894 m/z serves both as predictor and monitor for tumor response to chemotherapy, and it is deserved for further studies.6、The model composed by 3164 m/z,3979m/z,9196 m/z,2734 m/z and 3963 m/z could do the best in the division of patients with ALN metastasis and patients without ALN metastasis. The accuracy was 88.2%.7、The expression level of the protein peak of 3979 m/z was relevant to the number of metastatic lymph nodes.更多还原