【作者】 张薇；

【导师】 陈坤；

【作者基本信息】 浙江大学 ， 流行病与卫生统计学， 2007， 硕士

【摘要】 研究背景与目的乳腺癌（breast cancer）是指发生于乳腺小叶和导管上皮的恶性肿瘤。据WHO的最新报道和我国卫生部相关统计资料显示，乳腺癌已成为威胁我国女性健康的重要疾病。并且乳腺癌的发生有低龄化趋势。乳腺癌是一种与多种因素有关的疾病，其发生是年龄、地域、生殖因素、环境因素、社会心理因素等多因素共同作用的结果。目前研究表明，饮酒、吸烟、乳腺疾病史，应激性生活事件及其伴随的烦恼、焦虑、疲倦和抑郁情绪是等均与乳腺癌的患病风险密切相关，而怀孕年龄早、怀孕次数多，以及长期高强度的体育锻锻练是乳腺癌的保护因素，另外激素受体水平不同对乳腺癌的易感性影响亦有差别。乳腺癌不仅仅是环境因素作用的结果，也受遗传因素的影响。它是环境因素与遗传因素共同作用，导致细胞周期紊乱、细胞失控性生长所致的一类疾病。细胞生长从G1-S-G2-M有着严格的周期顺序及时间。其中处于G1晚期控制细胞进入S期的检查点（check point）尤为重要，细胞一旦通过了此限制点，将会自主完成细胞周期而不再依赖外源性增殖分裂信号的刺激。p53通路相关蛋白和细胞周期调节蛋白在控制细胞周期从G₁-S期的转过程中换起到关键的作用。细胞周期G₁-S期中涉及的p53通路相关多态基因已有多种见于报道，其中主要的有p53 codon72 Arg／Pro、intron 3 16bp副本、intron 6 G／A多态和p21 70C／T等。而CCND1 870 G／A则是目前研究者较为关注的细胞周期调节蛋白基因。本研究旨在通过自然人群为基础的病例对照研究，结合应用现场流行病学和分子流行病学研究方法技术，从人群水平分析明确细胞G1-S期相关调控基因多态性与乳腺癌发病遗传易感性的的关系，阐明其特异的遗传易感性标志，为乳腺癌遗传易感性研究提供人群水平的有力证据，最终为乳腺癌人群预防和临床治疗服务。材料与方法采用自然人群为基础的病例对照设计，以浙江省嘉善县10个乡镇参加1989-1990结直肠癌普查的64 693人的随访队列为研究的源人群，以1990年5月至2005年5月发生的，曾经参加1989-1990年普查的原发性乳腺癌存活病例共118例为研究对象，排除失访和拒访后共余84例组成病例组。再在89年的肿瘤普查名册中，根据普查号按年龄分层随机抽取280个女性组成对照组，每一个对照同时安排2个同居住地、同性别、同年龄（±5岁）的人作为候补调查对象，校对过程中排除死亡、发病、失访、拒访和DNA抽提失败，最后有效对照269例。采用统一编制并经过预调查的问卷对研究对象进行调查，调查内容包括：一般状况、生活方式与习惯、饮食习惯与种类、既往疾病史等，由经过培训考核合格的调查员直接询问被调查对象本人后填写。食物摄入量的调查采用根据统一的食物模型回忆一年中摄入数量和频率的方法。经调查对象知情同意后，采集静脉血5ml，分装储存备用。实验室检测采用改良盐析法抽提外周血白细胞DNA；p53 codon 72 Arg／Pro、intron 6 G／A以及p21和CCND1 G870A基因多态性检测均采用以聚合酶链反应为基础的限制性内切酶片断长度多态性（PCR-RFLP）方法；p53 intron 3 16bp dumplication采用PCR后直接电泳分析法。实验室质量控制通过盲法检测和重复检测实现。调查表经统一编码，EPIDATA软件两遍输入，校对无误后建立数据库进行统计分析。分析前进行必要的变量代换及编码。分析过程主要包括：病例组和对照组的年龄、婚姻状况、职业和文化程度分别作均衡性x²分布检验；相关因素的相对危险度估计采用采用非条件Logistic回归计算比值比（OR）及其95％可信区间（CI）；采用EH Linkage Software 1.2分析p53基因多个位点间是否存在连锁不平衡现象；基因—基因和基因—环境交互作用采用叉生分析和相乘交互作用模型进行分析。全部统计分析均在SPSS13.0、EH Linkage Software 1.2和Excel 2000软件中进行。结果1．环境暴露因素与乳腺癌关系在本研究人群中，年龄分布和婚姻状况在对照组和病例组中没有显著性差异。职业和文化程度在对照组和病例组间分布不均衡，差异具有统计学显著意义（x²值分别为12.17和17.14，P值分别为0.11和0.00）。经年龄因素调整后，非条件Logistic回归分析发现，文化程度和职业与乳腺癌的关系有统计学意义，初中文化程度者患乳腺癌的风险较之其他文化程度者有统计学意义的升高（OR=3.47，95％CI：1.50～8.04，P＜0.01）；与从事其他工作的人相比，农民患乳腺癌的风险则呈统计学意义的下降（OR=0.33，95％CI：0.19～0.56，P＜0.05）。一般环境因素中，吸烟是可明显增加患乳腺癌的风险，OR值达3.70（95％CI：1.03～13.29，P＜0.05）。而饮酒、饮茶与乳腺癌的危险性无统计学意义上的相关。平时心情一般或较差也可显著增加患乳腺癌的危险性，OR值为3.03（95％CI：1.60～5.76，P＜0.01）。而有良性乳腺疾病的个体较之健康个体患乳腺癌的风险增高达8.93倍（95％CI：2.00～39.85，P＜0.05）。饮食因素中，经常食用炸鱼和腌鱼是乳腺癌的危险因素，两者的OR值分别为1.86（95％CI：1.06～3.25，P＜0.05）利2.08（95％CI：1.20～3.61，P＜0.01）。女性生理周期因素中，病例组和对照组在初潮年龄、总孕产次数和哺乳时间上均存在显著差异：病例组的初潮年龄要比对照组平均早0.7岁（P=0.005），对照组的总孕产次数要比病例组多0.5次，并且存在随着孕产次数增加乳腺癌患病风险降低的趋势（趋势P=0.006）。哺乳同样可能对乳腺癌具有保护效应，对照组平均哺乳时间比病例组长11.8个月（P=0.004）。2．基因多态性与乳腺癌关系根据目前研究，尚不能认为p53 codon 72 Arg／Pro、intron 6 G／A、intron 3 16bp副本插入、p21C870T和CCND1 A870G是乳腺癌风险的独立预测因子。另外，对照组p53基因单体型分布存在遗传连锁不平衡现象，而在病例组中没有观察到该现象。A-Arg-G和A-Pro-G是最常见的2类单体型，其在病例组和对照组的分布频率总和分别为93.2％利90.3％。3．基因-基因交互对乳腺癌风险的影响本研究关于p53 codon 72、intron 6和intron 3三位点之间，以及p53与p21 C70T、CCND1 A870G基因多态之间的两两交互均没有观察到对乳腺癌的患病风险产生统计学意义上的影响。4．环境暴露因素与基因多态性交互对乳腺癌风险的影响当初潮年龄＞16岁时，p53 codon 72和intron 6的野生纯合基因型Arg／Arg和GG均使乳腺癌的患病风险明显降低。但是，交互作用模型分析却发现初潮年龄与p53该三个位点的基因多态之间不存在有统计学意义的交互作用（P＞0.05）。其他女性生理周期相关的因素与基因多态之间的交互作用均未达统计学意义（P＞0.05）。因此，尚不能认为p53、p21C70T及CCND1 A870G多态基因与女性生理周期因素之间存在显著的交互作用。结论本次以自然人群为基础的病例对照研究揭示：嘉善县人群中乳腺癌的危险因素包括吸烟、平时心情较差、喜食腌鱼、炸鱼和良性乳腺疾病史。初潮年龄、总孕产次数和哺乳时间在病例组和对照组间分布具有统计学差异，其与乳腺癌关系有待进一步论证。根据本次研究，尚不能认为p53 codon 72 Arg／Pro、intron 6 G／A、intron 3 16bp副本插入、p21C870T和CCND1 A870G是乳腺癌风险的独立预测因子。另外，p53、p21C70T及CCND1 A870G基因与基因、基因与环境因素之间的交互作用也尚不明确，有待进一步研究。更多还原

【Abstract】 Background and PurposeBrest cancer is a kind of malignancy on mammary glands and duct epithelial cells. According to the relevant statistics form WHO and national Ministry of Health, breast cancer has been an important disease which threats the health of women in our country. And it also shows younger trend in the incidence of breast cancerBreast cancer is a disease induced by comprehensive factors, including age, region, reproductive factors, environmental factors, psycho-social factors and so on. The current study shows that alcohol consumption, smoking, history of breast disease, stressful life events, and troubles, anxiety, fatigue and depression are all risk factors closely related with breast cancer, while early age pregnancy, high number of pregnancies, and long-term high-intensity physical exercise are all protective factors of breast cancer. What’s more, the difference on the level of estrogen receptor also inference the susceptibility of breast cancer.Breast cancer is not only a result of environmental factors but also affected by genetic susceptibility. It is a kind of diseases induced by interaction of environmental and genetic factors, which leads to turbulence of cell cycle and uncontrolled cell proliferation. There is strict cell cyclesequencing from G1-S-G2-M, among which G1-S check point is especially important. If this check point is spanned, cell cycle will carry through without stimulation of exterior signal. p53 pathway and cell cycle related regulating protein play an important role at the G1-S check point. Among them, p53 codon72 Arg/Pro、 intron 3 16bp induplication、 intron 6 G/A, p21 70C/, CCND1 A871G polymorphisms attract more attention currently.The purpose of this population-based case-control study is to explore the association between genetic polymorphism in GI-S cell cycle regulation and the susceptibility to breast cancer on a population level by using field epidemiology and molecular epidemiology methods comprehensively, in order to clarifying specific genetic susceptibility marks of breast cancer, providing research evidence for genetic susceptibility to breast cancer on a population level, and serving crowd prevention and clinical treatment ultimately.Material and MethodsIn this population-based case-control study, a colorectal cancer screening cohort in Jiashan County, Zhejiang Province with 64693 persons enrolled in which covered 10 towns and was established during 1989 to 1990 was treated as the study population. 84 of 118 prevalent patients of primary breast cancer diagnosed from May 1990 to May 2005 in the cohort, eliminating the missing follow-up cases and refusing investigation cases, made up of the case group. According to age and sex distributions of the cohort, 280 persons were selected out as controls by stratified sampling, and actually 269 persons both accepted the investigation and provided effective blood samples. All of the controls alive never had neoplasm.A constructed questionnaire elicited information on the demographic condition, life styles and habits, dietary habits and categories, and past disease histories. All subjects were interviewed face-to-face by trained interviewers. Intake quantities of all common kinds of foods were obtained by asking subjects per time intake and intake frequencies of foods, and uniform-made food models were used to assist the food investigation.Also with subjects’ permission, a sample of venous blood (5ml) was taken at the same time of interview, which was separated into two sections of 2ml blood with sodium citrate anticoagulation and 3ml blood without anticoagulation to gain blood serum. All blood samples were stored at -60 centigrade degrees(°C) refrigerator for long-term conservation. Genome DNA was extracted from leucocytes by improved salt out method. Genotypes of p53 codon 72 Arg/Arg, intron 6 G/A polymorphisms, and p21 70C/T, CCND1 870A/G were detected by PCR-restriction lengthen polymorphism (PCR-RFLP). The p53 intron 3 16bp induplication was detected by directly electrophoresing PCR products on agarose gels.All questionnaires were uniformly coded, and then input into computer for double times. After revising and correcting the data, statistical analysis was carried out. Distributional characteristics of qualitative variables between cases and controls were analyzed by x~2 tests, and a non-conditional Logistic model was used for estimations of OR values for breast cancer of related study factors, chi-square trend test for judging dose-reaction relationship. The Hardy-Weinberg equilibrium test in control group was finidhed by x~2 goodness-of-fit test. EH Linkage Software 1.2 was used to identify the linkage disequilibrium. And the interaction model was applied for analysis of gene-gene and gene-environment interaction to breast cancer. All analyses were performed with SPSS for Windows 13.0, EH Linkage Software 1.2and Excel 2003.Results1. Environmental exposure and breast cancerIn our study, the distribution of age and marital status had no difference between cases and controls, while occupation and education showed uneven distribution between cases and controls, with x~2 =12.17 and 17.14, P=0.11 and 0.00 respectively, which indicated statistical significance. After being adjusted by age, non-conditional Logistic regression analysis suggested that subjects with education degrees of junior middle school had a statistically significant increased risk of breast cancer with an OR value of 3.47(95%CI: 1.50-8.04, P<0.01), however, farmers were at a statistically significant decreased risk of breast cancer with an OR value of 0.33(95%CI:0.19-0.56, P<0.05).Cigarette smoking was associated with increased risk of breast cancer distinctly, with an OR value of 3.70 (95%CI: 1.03-13.29, P<0.05). But there was no statistical significance between drinking, tea and breast cancer. Also, bad mood was a risk factor for breast cancer (OR=3.03, 95%CI: 1.60-5.76, P<0.01). Compared with people without benign breast disease, elevation of breast cancer risk was most pronounced among people with a history of benign disease (OR=8.93, 95%CI: 2.00-39.85, P<.05).For dietary factors, subjects prefer to fried or pickled fish had a statistically significant increased risk of breast cancer with OR value of 1.86 (95%CI: 1.06-3.25, P<0.05) and 2.08 (95%CI: 1.20-3.61, P<0.01) respectively.For factors about reproduction period of woman, menarche age, total number of pregnancies and time of lactation all showed uneven distribution between cases and controls. The mean menarche age of cases was 0.7 year earlier than controls (P=0.005). While the mean total number of pregnancies of cases was 0.5 times more than controls (P=0.001), and showed a decreasing trend of breast cancer risk as the total number of pregnancies increased. What’s more, lactation also had a protective effect to breast cancer with mean lactation time of controls 11.8 months more than cases (P=0.004).2. Genetic polymorphisms and breast cancerAnalysis about p53 codon 72, intron 6, intron 3, p21 C70T and CCND1 A870G all showed no statistically significant association between these genetic polymorphisms and breast caner.Linkage disequilibrium of p53 haplotypes was observed in control group, but not in case group. A-Arg-G and A-Pro-G were two haplotypes mostly observed, and the total percentage of which were 93.2% and 90.3% in cases and controls respectively.3. Gene-gene interaction and breast cancerNo statistically significant interaction to breast caner between p53 codon 72, intron 6, intron 3, p21 C70T and CCND1 A870G were observed in this study.4. Gene-environment interaction and breast cancerAmong subjects whose menarche age was lower than 16, the gene types of p53 codon 72 Arg /Arg and intron 6 GG were linked with decreased breast cancer risk, with P value higher than 0.05. However, analysis by interaction model showed that no statistically significant interaction to breast caner between menarche age and p53 genetic polymorphism were observed (P>0.05). In a word, gene-environment interaction to breast caner could not be confirmed in this study and further research was in need.ConclusionAccording to this population-based case-control study, cigarette smoking, bad mood, prefer to fried or pickled fish, and with benign breast disease are all risk factors for breast cancer, while menarche age, total number of pregnancies and time of lactation are factors distributing unevenly between cases and controls groups, that indicates certain association between such factors and susceptibility of breast cancer but should be confirmed further.However, analysis about p53 codon 72, intron 6, intron 3, p21 C70T and CCND1 A870G are all showed no statistically significant association between these genetic polymorphisms and breast caner. So they could not be considered as independent predictors to breast cancer yet. As to analysis of gene-gene and gene-environment interaction, the results also have no statistical significance, which indicates further research with larger sample size should be conducted in the near future.更多还原