非小细胞肺癌患者与健康人群血清及尿液特异性差异多肽及肺鳞癌患者血清多肽与化疗疗效相关性的探索性研究
发布时间：2018-09-01 13:33
【摘要】：背景:肺癌(lung cancer)是目前最常见的恶性肿瘤。在全世界范围内,肺癌的发病率和死亡率均居各种恶性肿瘤的榜首。在我国,肺癌为男性肿瘤发病率及死亡率的第一位,女性肿瘤发病率的第二位,死亡率的第一位。肺癌主要分为小细胞肺癌(small cell lung cancer,SCLC)和非小细胞肺癌(non-small cell lung cancer,NSCLC)两大类。NSCLC是肺癌最常见的病理类型,占新发肺癌病例的80%以上,其病理类型主要包括:肺鳞癌(squamous cell carcinoma of lung)和肺腺癌(adenocarcinoma of lung)。目前NSCLC患者的5年生存率只有10%-15%,然而早期诊断,早期手术的I期NSCLC患者,术后的10年生存率可达92%。但是由于起病隐匿,且缺乏灵敏度和特异性均能满足临床需求的早期诊断手段,致使初诊时已有70%以上的NSCLC患者失去了手术机会。因此早发现,早诊断,早治疗才是延长患者生存期,提高患者生活质量的关键所在。目前常用的诊断肺癌的方法,如病理学,影像学及肺癌标记物。穿刺活检是明确病理的主要手段。但其为有创检查,且部分患者不易通过穿刺取到活检;影像学诊断是目前初步诊断肺癌最常用的手段,其价格较高,有射线损害,并且早期肺癌患者体内癌变的细胞远远小于影像学技术可测量的最小阈值;肿瘤标志物作为目前最常用的肺癌初筛手段,其敏感度及特异性均不够临床需求,这三种常规诊断肺癌的方法均难以实现肺癌的早期诊断。因此亟需一种可以方便开展的用于NSCLC早期诊断的手段。尽管抗血管生成治疗、免疫治疗、靶向治疗等已被FDA批准为晚期肺鳞癌患者的二线或多线治疗手段。晚期肺鳞癌的临床治疗仍停留在以传统化疗为主的阶段,铂二联方案化疗依然是晚期肺鳞癌患者主要的一线治疗手段。而在临床观察中发现,由于缺乏有效手段能在化疗前预测患者化疗疗效,致使部分肺鳞癌患者承受化疗带来的毒性作用却未能从化疗中获益。单个基因标志物如ERCC1、RRM1、TUBB3及XRCC1,因细胞毒药物应答与单个基因标志物的相关性较弱,在进一步扩大样本验证的临床试验中均未能有效预测疗效。因此寻找有效的化疗疗效预测手段,对肺鳞癌患者进行化疗疗效预测,对于使肺鳞癌患者更好的从化疗中获益,减少化疗带来的毒性作用至关重要。如今蛋白质组学已广泛应用于血液、尿液等各种体液及组织的研究,成为应用于各种肿瘤研究的蛋白质组学工具。本研究从NSCLC患者的诊断及治疗两个方面进行蛋白质组学研究,以期对NSCLC患者进行更好地诊断及治疗,实现对于患者的全程管理。第一部分运用MALDI-TOF质谱仪检测非小细胞肺癌患者与健康人群血清及尿液特异性差异多肽目的:本研究应用基质辅助激光解析电离飞行时间质谱(matrix-assisted laser desorption/ionization-time of flight-mass spectrometry,MALDI-TOF-MS)检测NSCLC患者与健康者血清及尿液样本之间的多肽差异,建立NSCLC患者的诊断模型及病理分类模型,为进行NSCLC患者的早期诊断,明确病理奠定基础。内容:收集2014年10月-2016年4月期间于我院肺部肿瘤内科就诊的经组织病理学或细胞学诊断为NSCLC患者的血液及尿液样本。收集肺鳞癌患者血清样本82例,尿液样本41例,肺腺癌患者血清样本82例,尿液样本41例。健康者的血清及尿液样本均取自自愿者,其中血清样本115例,尿液样本74例。经血清及尿液样本预处理,MALDI-TOF-MS质谱检测,CPT软件处理分析,得到NSCLC患者的诊断模型及病理分类模型。所建立模型可用于NSCLC患者的早期筛查,明确诊断及明确病理。方法:将一般情况相匹配的NSCLC患者血清及尿液样本,健康人群血清及尿液样本,按照3:1的比例随机分为训练组和验证组:训练组用于建立NSCLC患者诊断及病理分类模型;验证组用以验证所建立的诊断及病理分类模型。采用铜离子鳌合纳米磁珠(MB-IMAC-Cu2+)提取血清及尿液样本中的多肽,运用MALDI-TOF-MS检测,并通过CPT软件分析处理得到训练组NSCLC患者与健康者的血清及尿液多肽指纹图谱。分别应用Clin Pro Tools(CPT)软件自带的3种不同的生物学算法:快速分类法(Quickclassifier,QC算法)、遗传算法(genetic algorithm,GA算法)和监督神经网络算法(supervised neural network,SNN算法)建立诊断模型。选取最优算法所建立的NSCLC患者血清和尿液诊断模型。应用验证组对所建立诊断模型进行盲样验证。对肺鳞癌及肺腺癌两种不同病理分型的血清及尿液样本所得到多肽指纹图谱进一步分类,应用CPT软件内置的3种不同的生物学算法建立病理分类模型,选取最优算法所建立的NSCLC患者的血清及尿液病理分类模型,应用验证组进行盲样验证。结果:将一般情况相匹配的164例NSCLC患者血清(鳞癌腺癌各82例),82例NSCLC患者尿液(鳞癌腺癌各41例)及115例健康人群血清,74例健康人群尿液按照3:1的比例随机分成训练组和验证组:训练组由124例NSCLC患者血清(鳞癌腺癌各62例),62例NSCLC患者尿液(鳞癌腺癌各31例)及85例健康人群血清,54例健康人群尿液组成,用于建立NSCLC患者诊断及病理分类模型;验证组由40例NSCLC患者血清(鳞癌腺癌各20例),20例NSCLC患者尿液(鳞癌腺癌各10例)及30例健康人群血清,20例健康人群尿液组成,用以验证所建立的诊断及病理分类模型。训练组在800-10000Da范围内寻找差异多肽,并找到具有统计学意义的差异多肽(p0.001)。NSCLC患者vs健康人,血清样本中发现有107个差异多肽,具有统计学意义的差异多肽有52个。其诊断模型最优算法为GA算法,所建立的NSCLC诊断模型由5个多肽(2105.93Da,867.25Da,4093.08Da,7651.25Da,5341.39Da)组成,模型的识别率为96.77%,交叉验证率为89.86%。应用验证组样本对所建立模型进行盲样验证,该模型准确率为92.9%(65/70),灵敏度为95.0%(38/40),特异性为90%(27/30);尿液样本中发现有131个差异多肽,具有统计学意义的差异多肽有19个,其诊断模型最优算法为SNN算法,所建立诊断模型由9个多肽(1718.26Da,2193.17Da,1378.24Da,812.56Da,2376.22Da,5954.38Da,5810.98Da,2812.4Da,2438.33Da)组成,模型的识别率为98.25%,交叉验证率为91.74%。应用验证组样本对所建立模型进行盲样验证,该模型准确率为90%(36/40),灵敏度为95%(19/20),特异性为85.0%(17/20);从NSCLC患者与健康人群血清及尿液样本差异多肽中,查见1个相同的具有统计学意义的差异多肽:3242Da,经鉴定为纤维蛋白原α。肺鳞癌vs肺腺癌,血液样本中发现有96个差异多肽,具有统计学意义的差异多肽有20个,其分类模型最优算法为GA算法,所建立病理分类模型由5个多肽(9312.15Da,3242.37Da,4213.52Da,5297.33Da,4645.83Da)组成,模型的识别率为90.1%,交叉验证率为75.65%,应用验证组样本对所建立模型进行盲样验证,该模型准确率为82.5%(33/40),灵敏度为85.0%(17/20),特异性为80.0%(16/20);尿液样本中发现有119个差异多肽,具有统计学意义的差异多肽有0个,未能建立病理分类模型。结论:本研究表明NSCLC患者与健康者,肺鳞癌与肺腺癌患者的血清及尿液多肽存在差异。运用MALDI-TOF-MS技术建立NSCLC患者诊断模型及病理分类模型,具有较高的敏感度和特异性。所建立模型可用于NSCLC患者早期诊断及明确病理的补充手段。但需进一步扩大样本量完善及验证预测模型。第二部分应用MALDI-TOF-MS检测肺鳞癌患者血清多肽并分析与其化疗疗效相关性目的:本研究应用基质辅助激光解析电离飞行时间质谱(MALDI-TOF-MS)检测初治晚期肺鳞癌患者接受紫杉醇类联合铂类化疗前血清多肽,并分析其与化疗疗效的相关性。建立晚期肺鳞癌患者化疗疗效预测模型,为进行化疗疗效的个体化预测,指导肺鳞癌患者进行个体化化疗奠定基础。内容:本研究共入组2014年10月-2016年4月期间就诊于我院肺部肿瘤内科经组织病理学或细胞学诊断为肺鳞癌患者81例。一线接受紫杉醇类联合铂类方案化疗,并每2周期进行疗效评价,分为肺鳞癌化疗敏感组及化疗耐药组。经血清样本预处理,MALDI-TOF-MS质谱检测,CPT软件处理分析,得到晚期肺鳞癌患者化疗疗效预测模型。所建立模型可用于预测紫杉醇类联合铂类方案化疗疗效。方法:收集治疗前晚期肺鳞癌患者的血清样本,一线行紫杉醇类联合铂类方案化疗,并每两周期进行疗效评价。按照实体瘤疗效评价标准(Response Evaluation Criteria in Solid Tumors RECIST1.1)评价治疗疗效。评效为CR或PR的肺鳞癌患者归为化疗敏感组,评效为PD的肺鳞癌患者归为化疗耐药组。将入组标本按照3:1的比例随机分为训练组(敏感组I与耐药组I)和验证组(敏感组II与耐药组II)。采用MB-IMAC-Cu2+进行血清预处理,分离得到血清样本中的多肽,MALDI-TOF-MS检测训练组血清多肽并得到血清多肽指纹图谱。CPT软件系统分析处理,得到两组样本间差异多肽,并应用CPT软件内置的3种不同的生物学算法(SNN,GA,QC算法)建立疗效预测模型。选取最优算法所建立的肺鳞癌化疗疗效预测模型,运用验证组对所建立模型进行盲样验证,得到模型的准确率,灵敏度,特异性。统计入组患者治疗疗效,通过统计学方法分析,得到血清差异多肽与两组患者PFS之间的相关性。结果:入组的81例一线接受紫杉醇类联合铂类方案化疗的初治晚期肺鳞癌患者。其中CR为0例,PR为40例(49.4%,40/81),PD为41例(50.6%,41/81)。训练组共纳入30例敏感患者(敏感组I),31例耐药患者(耐药组I);验证组共纳入敏感(敏感组II)与耐药(耐药组II)患者各10例。敏感组I中位PFS为7.2个月(95%CI:4.4-14.5);耐药组I中位PFS为1.8个月(95%CI:0.7-3.5)。敏感组I与耐药组I患者血清样本中发现有96个差异多肽,其中具有统计学意义的差异多肽有16个(p0.001)。疗效预测模型最优算法为GA算法,所建立模型由5个多肽(1897.75Da,2023.93Da,3683.36Da,4269.56Da,5341.29Da)组成。该模型对敏感组患者的识别率为95.11%,交叉验证率为89.18%。应用验证组患者样本进行验证,该模型总的准确率85%(17/20),灵敏度90%(9/10),特异性80%(8/10)。敏感组I中位PFS为7.2个月(95%CI:4.4-14.5);耐药组I中位PFS为1.8个月(95%CI:0.7-3.5)。结合临床预后参数PFS运用双变量相关分析,得到各个多肽峰与PFS间的相关系数。结果发现:4232.04Da,4269.56Da的差异多肽与肺鳞癌患者PFS存在相关性(p0.01)。其中4269.56Da多肽同样用于建立化疗疗效预测模型,进一步说明了其多肽与化疗疗效的密切关系。进一步进行多肽峰的鉴定,得到2个在化疗耐药组患者血清中表达上调的多肽峰。其质荷比为:1897Da,2023Da,经鉴定为补体C4a,补体C3f。结论:本研究表明化疗敏感组及化疗耐药组患者的血清多肽存在差异,运用MALDI-TOF-MS技术建立疗效预测模型,可用于预测紫杉醇类联合铂类方案化疗疗效,并且具有较高的敏感度和特异性。但需进一步扩大样本量完善及验证预测模型。
[Abstract]:BACKGROUND: Lung cancer is the most common malignant tumor at present. In the world, the incidence and mortality of lung cancer rank first among all kinds of malignant tumors. Small cell lung cancer (SCLC) and non-small cell lung cancer (NSCLC) are two major types. NSCLC is the most common pathological type of lung cancer, accounting for more than 80% of new lung cancer cases. The pathological types include squamous cell carcinoma of lung and adenocarcinoma of lung. The annual survival rate is only 10%-15%. However, the 10-year survival rate of early stage I NSCLC patients is 92%. However, due to the concealed onset, lack of sensitivity and specificity can meet the clinical needs of early diagnosis, more than 70% of the initial diagnosis of NSCLC patients have lost the opportunity for surgery. Puncture biopsy is the main method of defining pathology, but it is invasive, and it is difficult for some patients to get biopsy through puncture. The most commonly used method of step-by-step diagnosis of lung cancer is high price, with radiation damage, and the cancer cells in early lung cancer patients are far less than the minimum threshold that can be measured by imaging technology; tumor markers as the most commonly used means of lung cancer screening, their sensitivity and specificity are not enough for clinical needs, these three routine diagnosis of lung cancer. Although anti-angiogenesis therapy, immunotherapy and targeted therapy have been approved by FDA as second-line or multi-line therapies for patients with advanced lung squamous cell carcinoma, the clinical treatment of advanced lung squamous cell carcinoma remains in the traditional chemotherapy. In the primary stage, platinum-based chemotherapy is still the main first-line treatment for advanced lung squamous cell carcinoma. In clinical observation, the lack of effective means to predict the efficacy of chemotherapy before chemotherapy, resulting in some lung squamous cell carcinoma patients withstanding the toxic effects of chemotherapy but not benefit from chemotherapy. ERCC1, RRM1, TUBB3 and XRCC1, because of the weak correlation between cytotoxic drug response and single gene markers, can not effectively predict the efficacy in further clinical trials to verify the expanded sample. Proteomics has been widely used in the study of various body fluids and tissues, such as blood, urine, and so on. It has become a proteomics tool for the study of various tumors. The first part is the detection of serum and urine specific polypeptides by MALDI-TOF mass spectrometry in patients with non-small cell lung cancer and healthy people. The purpose of this study was to apply matrix-assisted laser desorption/ionization mass spectrometry (MALDI-TOF mass spectrometry). - time of flight-mass spectrometry, MALDI-TOF-MS) was used to detect the polypeptide differences between the serum and urine samples of patients with NSCLC and healthy subjects, to establish the diagnosis model and pathological classification model of NSCLC patients, and to lay the foundation for early diagnosis and pathological diagnosis of NSCLC patients. Blood and urine samples from 82 patients with squamous cell carcinoma of the lung, 41 with urine, 82 with adenocarcinoma of the lung and 41 with urine were collected. Serum and urine samples from healthy volunteers were collected, including 115 serum samples and 74 urine samples. The diagnosis model and pathological classification model of NSCLC patients were obtained by pretreatment of serum and urine samples, detection of MALDI-TOF-MS mass spectrometry and analysis of CPT software. The established model can be used for early screening of NSCLC patients, definite diagnosis and definite pathology. Liquid samples were randomly divided into training group and validation group according to the ratio of 3:1: training group was used to establish the diagnosis and pathological classification model of NSCLC patients; validation group was used to validate the established diagnosis and pathological classification model. Serum and urine polypeptide fingerprints of NSCLC patients and healthy subjects in training group were obtained by CPT software. Three different biological algorithms, Quick Classifier (QC algorithm), Genetic Algorithm (GA algorithm) and Supervised Neural Network (supervised neural network) were used respectively. The serum and urine diagnostic models of NSCLC patients established by the optimal algorithm were selected. The diagnostic models were blindly validated by the validation group. The serum and urine pathological classification models of NSCLC patients were established by three different biological algorithms built-in. The validation group was used to verify the pathological classification models. Results: The serum samples of 164 NSCLC patients (82 squamous cell carcinomas, 82 adenocarcinomas, 41 adenocarcinomas) and 82 urine samples of NSCLC patients (41 squamous cell carcinomas, 41 adenocarcinomas, 41 adenocarcinomas, respectively) were matched. The training group consisted of 124 NSCLC patients'serum (62 squamous cell carcinoma patients each), 62 NSCLC patients' urine (31 squamous cell carcinoma patients each) and 85 healthy people's serum, 54 healthy people's urine was used to establish the diagnosis and diagnosis of NSCLC. The validation group consisted of 40 NSCLC patients'serum (20 squamous cell carcinoma, 20 adenocarcinoma), 20 NSCLC patients' urine (10 squamous cell carcinoma, 10 adenocarcinoma) and 30 healthy people's serum, and 20 healthy people's urine to validate the established diagnosis and pathological classification model. There were 107 different polypeptides in the serum samples of healthy people with NS CLC (p0.001). There were 52 different polypeptides with statistical significance. The optimal algorithm of diagnosis model was GA algorithm. The diagnosis model of NSCLC was composed of five polypeptides (2105.93 Da, 867.25 Da, 4093.08 Da, 7651.25 Da, 5341.39 Da). The accuracy of the model was 92.9% (65/70), the sensitivity was 95.0% (38/40) and the specificity was 90% (27/30). 131 differential peptides were found in urine samples, 19 of which were statistically significant. The optimal algorithm of the diagnosis model was SNN algorithm, the established diagnosis model is composed of nine peptides (1718.26Da, 2193.17Da, 1378.24Da, 812.56Da, 2376.22Da, 5954.38Da, 5810.98Da, 2812.4Da, 2438.33Da). The recognition rate of the model is 98.25%, the cross validation rate is 91.74%. The validation of the model by the validation group of blind samples, the accuracy rate of the model is 90% (36/40), the sensitivity is 95% (95%). 19/20, specificity 85.0% (17/20); from the serum and urine samples of patients with NSCLC and healthy people, we found one of the same statistically significant polypeptides: 3242Da, identified as fibrinogen alpha. lung squamous cell carcinoma vs lung adenocarcinoma, 96 differential polypeptides were found in blood samples, 20 of which were statistically significant. The optimal algorithm of the classification model is GA algorithm. The pathological classification model is composed of five polypeptides (9312.15Da, 3242.37Da, 4213.52Da, 5297.33Da, 4645.83Da). The recognition rate of the model is 90.1%, the cross validation rate is 75.65%. The accuracy and sensitivity of the model are 82.5% (33/40) and 85.0% (85.0%) respectively. 17/20, specificity 80.0% (16/20); urine samples found 119 different polypeptides, statistically significant difference polypeptides 0, can not establish a pathological classification model. Conclusion: This study shows that NSCLC patients and healthy people, lung squamous cell carcinoma and lung adenocarcinoma patients serum and urine polypeptides are different. The established model can be used for early diagnosis of NSCLC patients and as a supplementary means for defining pathology. However, further enlargement of the sample size is needed to improve and validate the prediction model. The second part uses MALDI-TOF-MS to detect serum polypeptides in patients with squamous cell carcinoma of the lung and to analyze their relationship with chemotherapy. Objective: To detect serum polypeptides in patients with advanced lung squamous cell carcinoma before chemotherapy with paclitaxel and platinum by matrix-assisted laser-resolved ionization time-of-flight mass spectrometry (MALDI-TOF-MS) and to analyze the correlation between serum polypeptides and chemotherapy efficacy. Content: A total of 81 patients with lung squamous cell carcinoma diagnosed by histopathology or cytology from October 2014 to April 2016 were enrolled in this study. They received paclitaxel combined with platinum chemotherapy in the first line and were evaluated every 2 cycles. After pretreatment of serum samples, detection of MALDI-TOF-MS and analysis of CPT software, a model for predicting the efficacy of chemotherapy in advanced lung squamous cell carcinoma was established. The model can be used to predict the efficacy of paclitaxel combined with platinum regimen chemotherapy. Methods: Serum samples from patients with advanced lung squamous cell carcinoma before and after treatment were collected. Samples were treated with paclitaxel plus platinum regimen chemotherapy on the first line, and the efficacy was evaluated every two cycles. Treatment efficacy was evaluated according to the Response Evaluation Criteria in Solid Tumors RECIST 1.1. Patients with CR or PR were classified as chemosensitivity group and those with PD were classified as chemosensitivity group. Group B. The samples were randomly divided into two groups according to the ratio of 3:1: training group (sensitive group I and drug resistance group I) and validation group (sensitive group II and drug resistance group II). In this paper, three different biological algorithms (SNN, GA, QC algorithm) built in CPT software were used to establish a therapeutic effect prediction model. Results: 81 patients with advanced lung squamous cell carcinoma received paclitaxel plus platinum chemotherapy in the first line, including 0 CR, 40 PR (49.4%, 40/81), 41 PD (50.6%, 41/81). The training group was included in the study. 30 sensitive patients (sensitive group I), 31 resistant patients (resistant group I); 10 sensitive patients (sensitive group II) and 10 resistant patients (resistant group II) were included in the validation group; the median PFS in sensitive group I was 7.2 months (95% CI: 4.4-14.5); and the median PFS in resistant group I was 1.8 months (95% CI: 0.7-3.5). 96 different polypeptides were found in the serum samples of sensitive group I and resistant group I. Among them, 16 were statistically significant (p0.001). The optimal algorithm for predicting the therapeutic effect was GA algorithm, and the established model consisted of five.
【学位授予单位】：中国人民解放军军事医学科学院
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R734.2
本文编号：2217361
[Abstract]:BACKGROUND: Lung cancer is the most common malignant tumor at present. In the world, the incidence and mortality of lung cancer rank first among all kinds of malignant tumors. Small cell lung cancer (SCLC) and non-small cell lung cancer (NSCLC) are two major types. NSCLC is the most common pathological type of lung cancer, accounting for more than 80% of new lung cancer cases. The pathological types include squamous cell carcinoma of lung and adenocarcinoma of lung. The annual survival rate is only 10%-15%. However, the 10-year survival rate of early stage I NSCLC patients is 92%. However, due to the concealed onset, lack of sensitivity and specificity can meet the clinical needs of early diagnosis, more than 70% of the initial diagnosis of NSCLC patients have lost the opportunity for surgery. Puncture biopsy is the main method of defining pathology, but it is invasive, and it is difficult for some patients to get biopsy through puncture. The most commonly used method of step-by-step diagnosis of lung cancer is high price, with radiation damage, and the cancer cells in early lung cancer patients are far less than the minimum threshold that can be measured by imaging technology; tumor markers as the most commonly used means of lung cancer screening, their sensitivity and specificity are not enough for clinical needs, these three routine diagnosis of lung cancer. Although anti-angiogenesis therapy, immunotherapy and targeted therapy have been approved by FDA as second-line or multi-line therapies for patients with advanced lung squamous cell carcinoma, the clinical treatment of advanced lung squamous cell carcinoma remains in the traditional chemotherapy. In the primary stage, platinum-based chemotherapy is still the main first-line treatment for advanced lung squamous cell carcinoma. In clinical observation, the lack of effective means to predict the efficacy of chemotherapy before chemotherapy, resulting in some lung squamous cell carcinoma patients withstanding the toxic effects of chemotherapy but not benefit from chemotherapy. ERCC1, RRM1, TUBB3 and XRCC1, because of the weak correlation between cytotoxic drug response and single gene markers, can not effectively predict the efficacy in further clinical trials to verify the expanded sample. Proteomics has been widely used in the study of various body fluids and tissues, such as blood, urine, and so on. It has become a proteomics tool for the study of various tumors. The first part is the detection of serum and urine specific polypeptides by MALDI-TOF mass spectrometry in patients with non-small cell lung cancer and healthy people. The purpose of this study was to apply matrix-assisted laser desorption/ionization mass spectrometry (MALDI-TOF mass spectrometry). - time of flight-mass spectrometry, MALDI-TOF-MS) was used to detect the polypeptide differences between the serum and urine samples of patients with NSCLC and healthy subjects, to establish the diagnosis model and pathological classification model of NSCLC patients, and to lay the foundation for early diagnosis and pathological diagnosis of NSCLC patients. Blood and urine samples from 82 patients with squamous cell carcinoma of the lung, 41 with urine, 82 with adenocarcinoma of the lung and 41 with urine were collected. Serum and urine samples from healthy volunteers were collected, including 115 serum samples and 74 urine samples. The diagnosis model and pathological classification model of NSCLC patients were obtained by pretreatment of serum and urine samples, detection of MALDI-TOF-MS mass spectrometry and analysis of CPT software. The established model can be used for early screening of NSCLC patients, definite diagnosis and definite pathology. Liquid samples were randomly divided into training group and validation group according to the ratio of 3:1: training group was used to establish the diagnosis and pathological classification model of NSCLC patients; validation group was used to validate the established diagnosis and pathological classification model. Serum and urine polypeptide fingerprints of NSCLC patients and healthy subjects in training group were obtained by CPT software. Three different biological algorithms, Quick Classifier (QC algorithm), Genetic Algorithm (GA algorithm) and Supervised Neural Network (supervised neural network) were used respectively. The serum and urine diagnostic models of NSCLC patients established by the optimal algorithm were selected. The diagnostic models were blindly validated by the validation group. The serum and urine pathological classification models of NSCLC patients were established by three different biological algorithms built-in. The validation group was used to verify the pathological classification models. Results: The serum samples of 164 NSCLC patients (82 squamous cell carcinomas, 82 adenocarcinomas, 41 adenocarcinomas) and 82 urine samples of NSCLC patients (41 squamous cell carcinomas, 41 adenocarcinomas, 41 adenocarcinomas, respectively) were matched. The training group consisted of 124 NSCLC patients'serum (62 squamous cell carcinoma patients each), 62 NSCLC patients' urine (31 squamous cell carcinoma patients each) and 85 healthy people's serum, 54 healthy people's urine was used to establish the diagnosis and diagnosis of NSCLC. The validation group consisted of 40 NSCLC patients'serum (20 squamous cell carcinoma, 20 adenocarcinoma), 20 NSCLC patients' urine (10 squamous cell carcinoma, 10 adenocarcinoma) and 30 healthy people's serum, and 20 healthy people's urine to validate the established diagnosis and pathological classification model. There were 107 different polypeptides in the serum samples of healthy people with NS CLC (p0.001). There were 52 different polypeptides with statistical significance. The optimal algorithm of diagnosis model was GA algorithm. The diagnosis model of NSCLC was composed of five polypeptides (2105.93 Da, 867.25 Da, 4093.08 Da, 7651.25 Da, 5341.39 Da). The accuracy of the model was 92.9% (65/70), the sensitivity was 95.0% (38/40) and the specificity was 90% (27/30). 131 differential peptides were found in urine samples, 19 of which were statistically significant. The optimal algorithm of the diagnosis model was SNN algorithm, the established diagnosis model is composed of nine peptides (1718.26Da, 2193.17Da, 1378.24Da, 812.56Da, 2376.22Da, 5954.38Da, 5810.98Da, 2812.4Da, 2438.33Da). The recognition rate of the model is 98.25%, the cross validation rate is 91.74%. The validation of the model by the validation group of blind samples, the accuracy rate of the model is 90% (36/40), the sensitivity is 95% (95%). 19/20, specificity 85.0% (17/20); from the serum and urine samples of patients with NSCLC and healthy people, we found one of the same statistically significant polypeptides: 3242Da, identified as fibrinogen alpha. lung squamous cell carcinoma vs lung adenocarcinoma, 96 differential polypeptides were found in blood samples, 20 of which were statistically significant. The optimal algorithm of the classification model is GA algorithm. The pathological classification model is composed of five polypeptides (9312.15Da, 3242.37Da, 4213.52Da, 5297.33Da, 4645.83Da). The recognition rate of the model is 90.1%, the cross validation rate is 75.65%. The accuracy and sensitivity of the model are 82.5% (33/40) and 85.0% (85.0%) respectively. 17/20, specificity 80.0% (16/20); urine samples found 119 different polypeptides, statistically significant difference polypeptides 0, can not establish a pathological classification model. Conclusion: This study shows that NSCLC patients and healthy people, lung squamous cell carcinoma and lung adenocarcinoma patients serum and urine polypeptides are different. The established model can be used for early diagnosis of NSCLC patients and as a supplementary means for defining pathology. However, further enlargement of the sample size is needed to improve and validate the prediction model. The second part uses MALDI-TOF-MS to detect serum polypeptides in patients with squamous cell carcinoma of the lung and to analyze their relationship with chemotherapy. Objective: To detect serum polypeptides in patients with advanced lung squamous cell carcinoma before chemotherapy with paclitaxel and platinum by matrix-assisted laser-resolved ionization time-of-flight mass spectrometry (MALDI-TOF-MS) and to analyze the correlation between serum polypeptides and chemotherapy efficacy. Content: A total of 81 patients with lung squamous cell carcinoma diagnosed by histopathology or cytology from October 2014 to April 2016 were enrolled in this study. They received paclitaxel combined with platinum chemotherapy in the first line and were evaluated every 2 cycles. After pretreatment of serum samples, detection of MALDI-TOF-MS and analysis of CPT software, a model for predicting the efficacy of chemotherapy in advanced lung squamous cell carcinoma was established. The model can be used to predict the efficacy of paclitaxel combined with platinum regimen chemotherapy. Methods: Serum samples from patients with advanced lung squamous cell carcinoma before and after treatment were collected. Samples were treated with paclitaxel plus platinum regimen chemotherapy on the first line, and the efficacy was evaluated every two cycles. Treatment efficacy was evaluated according to the Response Evaluation Criteria in Solid Tumors RECIST 1.1. Patients with CR or PR were classified as chemosensitivity group and those with PD were classified as chemosensitivity group. Group B. The samples were randomly divided into two groups according to the ratio of 3:1: training group (sensitive group I and drug resistance group I) and validation group (sensitive group II and drug resistance group II). In this paper, three different biological algorithms (SNN, GA, QC algorithm) built in CPT software were used to establish a therapeutic effect prediction model. Results: 81 patients with advanced lung squamous cell carcinoma received paclitaxel plus platinum chemotherapy in the first line, including 0 CR, 40 PR (49.4%, 40/81), 41 PD (50.6%, 41/81). The training group was included in the study. 30 sensitive patients (sensitive group I), 31 resistant patients (resistant group I); 10 sensitive patients (sensitive group II) and 10 resistant patients (resistant group II) were included in the validation group; the median PFS in sensitive group I was 7.2 months (95% CI: 4.4-14.5); and the median PFS in resistant group I was 1.8 months (95% CI: 0.7-3.5). 96 different polypeptides were found in the serum samples of sensitive group I and resistant group I. Among them, 16 were statistically significant (p0.001). The optimal algorithm for predicting the therapeutic effect was GA algorithm, and the established model consisted of five.
【学位授予单位】：中国人民解放军军事医学科学院
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R734.2
【参考文献】
相关期刊论文 前4条
1 Ian Paul;J Mark Jones;;Apoptosis block as a barrier to effective therapy in non small cell lung cancer[J];World Journal of Clinical Oncology;2014年04期
2 姚晓军;刘伦旭;;肺癌的流行病学及治疗现状[J];现代肿瘤医学;2014年08期
3 安娟;汤传昊;王娜;刘毅;郭万峰;李晓燕;王子赫;何昆;刘晓晴;;MALDI-TOF质谱筛查NSCLC患者血清特异性多肽的探索性研究[J];中国肺癌杂志;2013年05期
4 时广利,胡秀玲,岳思东,宋长兴;血清肿瘤标志物在肺癌辅助诊断中的应用[J];中华肿瘤杂志;2005年05期
，本文编号：2217361
本文链接：https://www.wllwen.com/yixuelunwen/zlx/2217361.html
