【作者】 陈益定；

【导师】 郑树；

【作者基本信息】 浙江大学 ， 肿瘤学， 2004， 博士

【摘要】 大肠癌是人类复杂疾病之一，是我国常见的九大恶性肿瘤之一。据全国肿瘤防治办公室最近发表的统计结果，大肠癌调整死亡率列为第5位主要恶性肿瘤，在大城市的发病率男性为第三位，女性为第二位。尽管近年来大肠癌的诊断、治疗有了长足的进展，但大肠癌的总体疗效并无明显改善，手术后5年生存率仍徘徊在50％左右。而早期大肠癌患者预后较好，5年生存率达90％～100％。若能提高早期大肠癌的诊断率，即可明显提高大肠癌的总体治疗效果。但由于多种原因，目前到医院就诊的患者中，2／3为中晚期癌，大肠癌术后的复发转移严重影响着患者的生存率，因此若能对大肠癌的预后进行准确的估计、对术后的辅助治疗进行合理选择，以及对复发和转移进行早期检测将是解决这一问题的关键，从而提高大肠癌患者的总体生存率。 就目前而言，在大肠癌的早期诊断、预后估计、复发或转移的早期检测中均存在着不同的困难。如大肠癌早期诊断和筛检的方法很多，但存在特异性与敏感性之间的矛盾；Dukes分期或TNM分期在预后的估计尚存在很多的缺陷，以无淋巴结转移的Dukes’B期的患者为例，也有近30％的患者在5年内死于肿瘤复发或远处转移；大肠癌70％的复发或转移病人在2年内出现，通过症状、常规体检发现肿瘤复发转移时已失去治疗时机，而影像学、相关癌基因或肿瘤标记物的检测仍存在假阳性率及假阴性率高的问题。 大肠癌是一种多基因，多步骤的复杂性状的疾病，包含了多个基因突变的分子事件过程，包括抑癌基因的功能丧失，癌基因的活化等。因此用单个或数个因子的检测对大肠癌的早期诊断，预后的估计，以及在随访中的检测必定不可避免地存在敏感性和特异性矛盾。蛋白质组学能够识别鉴定细胞、组织或机体的全部蛋白质，提供了一组蛋白质的功能及其模式的信息，能够同时反映细胞内部的遗传特性和外界因素的影响的结果。因此深入研究恶性肿瘤的发生发展的过程，客观上需要在蛋白质组的水平进行进一步的探索。SELDI-TOF-MS(Surface Enhanced 浙江大学博士学位论文Laser nes呷tio叭onization一Time of Flight一Mass speetrometry，表面加强激光解吸电离一飞行时间质谱)技术是最近几年才发展起来的一种新的蛋白质组学研究方法。由于SELDI蛋白质芯片技术的高灵敏度及高通量性，能反映被检样本中蛋白质的全貌，进而同时找到多个蛋白质标志物，为早期诊断、预后估计、复发或转移的早期检测提供了可能。方法: 本研究中以 182例血清标本(其中55例大肠癌、35例大肠腺瘤、92例健康人)为研究对象，应用SELDI一TOF一MS质谱仪测定了大肠癌患者、大肠腺瘤患者与健康人血清标本的蛋白质指纹图谱，结合人工神经网络及支持向量机对这些数据进行处理，建立相应的指纹图谱诊断模型;进一步对43例Dukes，A、B、C期的大肠癌患者的预后进行了2年的随访，分析了血清蛋白质指纹图谱在预后估计中的价值;并对10例术前和术后不同时期的血清的蛋白质指纹图谱的进行了分析，探讨其在肿瘤复发或转移中的动态变化。结果: 1、大肠癌血清蛋白质指纹图谱实验方法的建立及蛋白质芯片种类的筛选:本研究在血清样本的处理时使用了两性离子去污剂0.5%CH妙S，应用cibacron Blue3GA(sigmalnc.)特异性地除去血清中的白蛋白，通过对4种不同的蛋白芯片IMAC固定金属亲合捕捉蛋白芯片，WCXZ弱阳离子交换蛋白芯片，SAXZ强阴离子交换蛋白芯片，H4疏水表面芯片进行研究发现，不同的芯片所能结合的血清蛋白质的数量是不同的，H4芯片在我们的研究中证实能得到更多有意义的峰值，能更好的应用于大肠癌血清蛋白质指纹图谱的研究，而且经改良后的检测血清蛋白质指纹图谱的实验方法是稳定可靠的，具有很好的重复性。 2、血清蛋白质指纹图谱在大肠癌早期诊断中的应用:182例血清标本的蛋白质指纹图谱经质谱仪收集数据，在收集每次实验数据前用已知分子量的标准蛋白芯片校正仪器测定的分子量，其误差小于0.1%。以质控血清作重复性检测，其峰值大小及其强度的变异系数(Cv:)均控制在误差范围内(0.05%和15%以下)，应用Proteinchip software 3.1(ciphergen Inc.)软件同时将所有样本的质谱数据M/z在2 000到30 000的峰值进行两次信噪比过滤。其中大肠癌与健康对照组检测到61 浙江大学博士学位论文个树Z峰，大肠癌与大肠腺瘤检测到235个M/Z峰。应用5倍交叉验证的人工神经网络模型(Artificial Neural Networks)分析大肠癌与健康对照样本指纹图谱的数据:以M/Z位于5911，8922，8944，8817的四个峰值的组合，在测试集上分析诊断模型的特异性为93.3%，敏感度为90.9%，丫buden指数为0.84242。采用3倍交叉验证线性的支持向量机(support Vector Machine)模型分析大肠癌与大肠腺瘤样本指纹图谱的数据:以h灯Z位于17247、18420、5911、9294、4654、21694、21742的7个峰值组合建立的诊断模型的特异性为83.2%，敏感度为893%，Youden指数为0.72484。182例大肠癌、大肠腺瘤与健康对照血清样本同时进行了C更多还原

【Abstract】 Colorectal cancer (CRC) is ranked the fifth in cancer-related deaths in China and third in frequency in men and second in women. The high mortality rate caused by this disease is to a significant extent associated with the poor diagnosis at its presymptomic stages. When CRC is detected early, more than 90% of persons with the disease live at least 5 years beyond diagnosis. Unfortunately, only 37% of colorectal cancers are diagnosed before they have spread. Furthermore, it’s important to identify those prognostic factors, which is a primary basis for therapeutic protocol selection and accurate prognosis of patients with colorectal cancer.Currently, there are some difficulty in early diagnosis, accurate prognosis estimate and early recurrence or/and metastasis detection. All the non-invasive detection techniques, which were used for clinical and screening, lack adequate sensitivity and specificity. The most reliable prognostic factor identified to date in colorectal cancer is the staging of disease at the time that treatment is initiated. Although the modified Dukes staging is commonly used worldwide, the number of applied variations makes correlation of different studies less than ideal. On the other hand, follow-up is carried out to find potentially curable recurrences. Detection of these recurrences can be worthwhile in terms of survival. Therefore, it is essential that more accurate and sophisticated methods be developed for this purpose.Because of the multifactorial nature of the colorectal cancer, it is very likely that a combination of several markers will be necessary to effectively detect and predict prognosis of the colorectal cancer. Proteomic methods detect the functioning units of expressed genes, through biochemical analysis of cellular proteins, to provide a protein fingerprint. The proteomic reflects both the intrinsic genetic programme of the cell and the impact of its immediate environment and is therefore valuable in biomarkerdiscovery. Distinct changes that occur at the protein level during the transformation of a normal cell into a neoplastic cell include altered expression, differential protein modification, changes in specific activity, and inappropriate localization, all of which may affect cellular function. SELDI-TOF (surface-enhanced laser desorption/ionization time of flight) mass spectrometry is one of the recently developed sophisticated technologies, which, based on capturing proteins/peptides by chemically modified surface, is specifically powerful for analyzing the complex biological samples. This technology, combined with bioinformatics, has been successfully used to analyze the complex serum proteins to explore the cancer-specific ’fingerprints’ or ’patterns’.Methods:182 serum samples including 55 colorectal cancer (CRC) patients, 35 colorectal adenoma (CRA) patients and 92 healthy people were detected by SELDI-TOF-MS. The data of spectra were analysed by the bioinformatics tools like artificial neural network (ANN) and support vector machine (SVM). The median follow-up time of the patients of Dukes’A, B, C was 20 months (12-24months). Using SELDI-TOF-MS, we expect to find some undocumented prognostic factors and create an individual risk assessment model for predicting the prognosis of the patients with colorectal cancer. We also used SELDI-TOF-MS to detect the serum of post-operation patients at intervals of 3 months, 6 months and 12 months to investigate the dynamic change of the serum protein fingerprint.Results:1. Phosphate buffer containing 0.5% CHAPS (pH 7.4) was added to serumsample and Cibacron Blue 3GA (Sigma) was added into the mixtures to delete serum albumin. The modified protocol of serum protein profiling was stable and repeatable, and applicable to other malignant tumor. Hydrophobic surface arrays (H4 chips) used for serumbiomarker discovery of the colorectal cancer have the advantage over other chips.2. Through noise filtration conducted by Ciphergen ProteinChip Software 3.1, there were 61 peaks detected for discriminate CRC from H更多还原