Molecular Genetic Epidemiology of leading U.S. Cancers

在美国，几乎一半的成年人新诊断的癌症将是三种类型：肺癌、乳腺癌和前列腺癌。如果要在降低癌症总发病率方面取得重大进展，找到预防这些主要肿瘤的方法将是重要的。为此，基础科学家已经在实验系统中确定了各种基因，这些基因可能在这些癌症的病因学中起着重要作用。一致地，流行病学家已经确定了与癌症风险增加相关的环境因素(例如吸烟和肺癌)。此外，统计遗传学家已经表明，对于大量的恶性肿瘤，患上癌症并存活下来的风险并不均匀地分布在整个人群中。鉴于癌症表型的复杂性和癌细胞基因组结构的戏剧性破坏，对终末期疾病的新治疗方法的近期成功只持谨慎乐观态度是合理的。更有效的短期策略可能是早期诊断或预防。对于大多数癌症类型，在疾病的早期阶段进行治疗干预已经证明了很高的有效率。该实验室正试图识别基因构成标记，以区分患乳腺癌的高危人群，或预防干预措施中出现并发症的风险人群。为了实现这些目标，正在使用两个样本人群：从他莫昔芬(13,388名妇女)的大型第三阶段试验中收集的仅病例收集和病例对照研究(共750例)。在实验室中，使用不同的病例对照人群组(总共493人)进行的初步研究集中在候选乳腺癌易感基因上，包括那些参与雌激素生物合成和代谢的基因。芳香酶基因(Cyp19)的等位基因分布在病例组和对照组之间有显著差异。此外，该实验室还开发了MALDI-TOF MS分析，对广泛收集的有助于剖析雌激素生物合成和代谢途径的基因标记进行了分析，特别关注文献中描述的功能显著的等位基因变异。正在对38个变种进行评估，仅在病例中进行三苯氧胺研究，以解决与预防干预并发症的潜在关联，并在病例对照人群中确定指示乳腺癌发展高风险的标志物。该实验室还试图通过在一项病例对照研究中检查候选基因变异，来确定改变肺癌发病风险的基因。在这项研究中，评估了14个基因的15个变异体，这些基因涉及I期和II期代谢。当在756个吸烟者样本子集中评估这15个候选易感基因变异，并将结果引入Logistic回归模型(根据遗传背景进行调整)时，只有GSTA4与肺癌显著相关。此外，还确定了烟草烟雾、饮食摄入量和候选基因变异之间的交互作用。在对329名吸烟者进行的分层分析中，健康的饮食模式(高纤维和碳水化合物的摄入量，低蛋白质和动物脂肪的摄入量)与GSTM1缺失个体患肺癌的风险降低有关。这一结果表明，饮食因素可以影响致癌物代谢酶在肺癌风险中的作用。利用G1/S检查点途径，已经进行了解释肺癌基因表达的路径分析方法的试点应用。之所以选择这一途径，是因为它已被证明在肺癌的发展中发挥着重要作用。最初的研究检验了基因间相关性的拟合模型是否反映了目前关于肺癌这一途径中发生的变化的生物学知识。用17个正常肺细胞系和55个肺癌细胞系的基因表达数据进行通径分析。关于该通路中的许多关键基因，包括转化生长因子-b、Smad4、p53、gsk3b、cdc25A、p27、p21、cdk2、细胞周期蛋白E、细胞周期蛋白6、细胞周期蛋白4、细胞周期蛋白D和细胞周期蛋白Rb，人们已经知道了很多，并根据现有的数据对模型进行了测试。第一个也是最简单的模型考察了上游基因对Rb表达的直接影响。在正常细胞中，35%的Rb表达变化与CDKs和Cyclins的表达有关。这些蛋白质的增加与Rb表达的减少相对应，这些蛋白质是G1向S转变的信号。肿瘤标本中Rb表达较低，CDK4和CDK6对Rb表达影响不大。在肿瘤中，细胞周期蛋白E对Rb表达的影响似乎有所增加，但细胞周期蛋白E与其调控因子CDK2有明显的解偶联，表明这些作用可能调节不佳。这项初步研究与有关正常细胞和癌细胞中这些基因的文献一致，并表明通径分析可以用来可靠地检验关于正常和疾病状态下发生的相关性模式变化的假设。

Almost half of new cancer diagnoses in adults in the U.S. will be of three types: cancers of the lung, breast, and prostate. If significant progress is to be made in reducing total cancer incidence, it will be important to finds means of preventing these major tumors. Toward this end, basic scientists have identified a variety of genes in experimental systems that may be important in the etiology of these cancers. Concordantly, epidemiologists have identified environmental factors that are associated with increased cancer risk (e.g., smoking and lung cancer). In addition, statistical geneticists have shown that for a large number of malignancies, the risk of developing and surviving cancer is not uniformly distributed throughout the population. Given the complexity of cancer phenotypes and the dramatic disruption of cancer cells genomic constitution, it is rational to only be cautiously optimistic about the near term success of novel therapeutic approaches to end-stage disease. A more efficacious short-term strategy may be early diagnosis or prevention. For most cancer types, therapeutic intervention in early stages of disease has demonstrated high rates of efficacy. The laboratory is attempting to identify genetic constitutional markers that distinguish individuals at high risk for developing breast cancer, or at risk of developing complications from prevention interventions. Two sample populations are being used to achieve these goals: a case only collection from a large phase 3 trial of Tamoxifen (13,388 women) and a case-control study (750 in all). Preliminary studies in the laboratory using a distinct case-control population set (493 in all) have focused on candidate breast cancer susceptibility genes, including those involved in estrogen biosynthesis and metabolism. A significant difference in allelic distribution of the aromatase gene (Cyp19) was observed between cases and controls. In addition, the laboratory has developed MALDI-TOF MS assays on an extended collection of gene markers useful in dissecting estrogen biosynthetic and metabolic pathways, focusing specifically on functionally significant allelic variants described in the literature. Thirty-eight variants are being assessed in the case only Tamoxifen studies to address potential associations with complications from prevention intervention, and in the case-control population to identify markers indicative of high risk for breast cancer development. The laboratory is also attempting to identify genes that modify the risk of developing lung cancer by examining candidate gene variation in a case-control study. In this study, fifteen variants in fourteen genes involved in Phase I and Phase II metabolism were evaluated. When these 15 candidate susceptibility gene variants were assessed in a 756-sample subset of smokers, and the results introduced into logistic regression models (following adjustments for genetic background), only GSTA4 was significantly associated with lung cancer. The interaction among tobacco smoke, dietary intake and candidate gene variation was also determined. On stratified analysis of 329 smokers, a healthy dietary pattern (high intakes of fibers and carbohydrates and low intakes of protein and animal fat) was associated with decreased lung cancer risk among GSTM1 null individuals. This result suggests that dietary factors can influence carcinogen metabolizing enzymes effects in lung cancer risk. Pilot application of path analysis methodologies for interpreting lung cancer gene expression has been performed utilizing the G1/S checkpoint pathway. This pathway was chosen because it has been shown to play an important role in the development of lung cancer. The initial study examined whether the fitted models of correlation between genes reflect current biological knowledge about the changes that occur in this pathway in lung tumors. Gene expression data from 17 normal lung and 55 lung carcinoma cell lines was used for path analysis. Much is known about many of the key genes in the pathway, which includes TGF-b, SMAD4, p53, gsk3b, cdc25A, p27, p21, cdk2, cyclin E, cdk6 cdk4, cyclin D and Rb, and models were tested based on the available data. The first and simplest model examined the direct influence of upstream genes on Rb expression. In normal cells, 35% of the variation in Rb expression appeared related to the expression of cdks and cyclins. Increases in these proteins, which signal G1 to S transition, corresponded to decreases in Rb expression. Tumor samples showed lower Rb expression, and Rb expression was not strongly influenced by cdk4 and cdk6. The influence of cyclin E over Rb expression appeared to increase in tumors, but there was an apparent decoupling of cyclin E and its regulator cdk2 suggesting that these effects may be poorly regulated. This initial study is consistent with the literature concerning these genes in normal and cancer cells, and suggests that path analysis can be used to reliably test hypotheses concerning changes in the patterns of correlations occurring in the normal and disease state