Plant-based diets and risk of cancer

描述（由申请人提供）：在2002-2007年间，我们收集了一个大型队列（N= 96,000），命名为复临健康研究-2 （AHS-2）。该队列包括25,000名黑人受试者，也有来自美国所有50个州和加拿大所有省份和地区的所有社会阶层的广泛代表。大约45%的人是素食主义者，但另外25%的人平均每天消耗139克肉。因此一个非常重要的特征就是饮食习惯的多样性。这项工作的主要目标是首先评估植物性饮食与杂食性饮食在结直肠癌、乳腺癌和前列腺癌风险方面的关联，因为之前的研究表明这些存在。接下来关注营养物质、植物化学物质、维生素和不同饮食模式的食物组，包括红肉、西红柿、乳制品、十字花科蔬菜、异黄酮、长链脂肪酸等。这一人群对大多数水果和蔬菜的摄取量相对较高，对大豆的摄取量接近亚洲人的水平，对n-3脂肪酸的摄取量和脂肪含量都很高。在这一人群中，许多关键饮食暴露的差异似乎是其他已知研究人群的1.5- 3倍。这意味着人口的统计能力是AHS-2的名义规模的1.5- 3倍，表明这是一项非常具有成本效益的研究。每位研究成员都完成了一份50页的调查问卷，包括详细的饮食、病史、体育活动、女性病史和人口统计数据。为了验证膳食摄入量估算的有效性，同时也便于回归分析中测量误差的修正，我们对1011名有代表性的受试者进行了大规模、全面的校准研究。从校准研究对象收集的数据包括6次精心安排的24小时饮食回顾，第二次食物频率问卷调查，两次为期一周的身体活动回顾，以及一组异常丰富的生物样本，用于估计生物标志物：血清、血浆、灰褐色皮毛、红细胞、夜间尿液和皮下脂肪抽吸。对于测量误差校正，将使用传统的回归校准，以及最近发表的一种创新的生物标志物引导校准方法。根据需要，我们从校准研究数据中估计了大量的摄入生物标志物。许多与饮食回忆的相关性都具有较高的价值。问卷饮食指数的有效性（与召回相比）也与其他人的最佳经验保持一致。通过与国家癌症登记处相匹配，以及从两年一次的调查问卷作为次要来源，发现偶发癌症。初步估计表明，我们只会错过2.5%的这些癌症。今后5年计划开展的主要活动有：a)继续并完成对癌症的检测；B)继续制作年度通讯，也包括流行文章，以促进群体保留，并保持最新的联系信息；C)最后也是最重要的是，完成分析并写出针对所列具体目标的手稿。

DESCRIPTION (provided by applicant): Between 2002-2007 we assembled a large cohort (N= 96,000) named Adventist Health Study-2 (AHS-2). The cohort contains 25,000 Black subjects, and also has a broad representation from all social classes, from all 50 states of the U.S. and all provinces and territories of Canada. About 45% are vegetarian, but another 25% on average consume 139 gm of meat per day. Thus a very important feature is a wide range of dietary habits. The main goals of this work are to first evaluate associations between plant-based compared to omnivorous diets on risks of colorectal, breast and prostate cancers, as previous work suggests that these exist. Next to focus on nutrients, phytochemicals, vitamins and food groups that differ between these dietary patterns, including red meat, tomatoes, dairy foods, cruciferous vegetables, isoflavones, long-chain fatty acids, and others. This population has a relatively high intake of most fruits and vegetables, a consumption of soy that approaches Asian levels, and has both high intake and high adipose levels of n-3 fatty acids. Many key dietary exposures in this population appear to have a variance that is 1.5- 3 times that of other well-known study populations. This translates to the statistical power of a population 1.5--3 times the nominal size of AHS-2, indicating that this is very cost-effective research. Each study member has completed a 50 page questionnaire that covers diet in detail, also medical history, physical activity, female history, and demographics. A large and comprehensive calibration study of 1011 representative subjects has been formed to validate the estimation of dietary intake, and also to facilitate measurement error correction in regression analyses. Data collected from calibration study subjects includes 6 carefully timed 24-hour dietary recalls, a second food frequency questionnaire, two one week physical activity recalls, and an unusually rich group of biological samples from which to estimate biomarkers: serum, plasma, buffy coat, RBCs, an overnight urine, and a subcutaneous fat aspirate. For measurement error correction, traditional regression calibration will be used, as well as an innovative recently published biomarker-guided calibration method. As required for this, we have estimated a large number of biomarkers of intake from the calibration study data. Many correlations with dietary recalls are relatively high-valued. The validity of questionnaire dietary indices (compared to recalls) is also in keeping with the best experience of others. Incident cancers are found by matching with state cancer registries, and also from biennial questionnaires as a secondary source. Initial estimates suggest we will miss only 2.5% of these cancers. The main categories of activity planned during the next 5 years are to a) continue and complete the detection of incident cancers; b) continue the production of an annual newsletter, also popular articles to promote cohort retention, and maintain up-to-date contact information; c) finally and most importantly, complete analyses and produce manuscripts that address the listed specific aims. PUBLIC HEALTH RELEVANCE: Breast, colorectal and prostate cancers afflict more than 550,000 Americans each year and kill nearly 120,000. This research will provide more definitive answers about the potential roles of plant-based dietary patterns, also meat consumption, to either increase or decrease risks of these cancers, by using an unusually informative U.S. study population. As many in this population also eat greater quantities of many particular vegetable foods, we can also associate intake of these with cancer risk. Great attention is given to improving accuracy of estimates.