Statistical Methods for Epigenetic Mediation of Exposure-response Relations

描述(申请人提供)：现已证实，DNA甲基化和组蛋白修饰等表观遗传现象在基因表达中起着重要作用，表观遗传状态的变化可以影响许多疾病的风险。然而，环境因素、构成基因及其相互作用在决定表观遗传状态方面的作用研究较少，特别是表观遗传学作为环境暴露和疾病之间的中介的可能性。一些证据还表明，某些环境暴露会对后代人的健康结果产生影响。这一建议有两个广泛的方法学目标，即发展统计方法(1)评估表观遗传学在调节个体内暴露-反应关系中的作用，以及(2)评估其在调节暴露的跨代影响方面的作用。对于其中的第一个，我们将(目标1a)开发一个灵活的测量误差框架，该框架可以适应暴露和甲基化随时间的变化，并将其与各种类型的结果(二元、连续、纵向、审查的事件发生时间)联系起来，(目标1b)开发方法来总结复杂的甲基化情况，并将这些情况与表观遗传状态(环境和遗传)的决定因素和疾病风险联系起来。对于第二个目标，我们将开发一种传递不平衡检验的变体，将给定基因座的基因类型扩展到包括表观遗传标记，从而允许这些标记以非孟德尔方式从任何一个亲本传递。父母和子女的甲基化状态将被视为潜在变量，是他们自己和/或他们各自父母暴露的函数。这些方法的统计性能将通过计算机模拟来研究。我们将把这些方法应用于儿童健康研究(CHS)参与者的不同子组。CHS是对空气污染和其他暴露的慢性影响的长期队列研究，例如在12,000名学龄儿童中，宫内和环境中的烟草烟雾-DNA可用于全球和特定区域的甲基化分析；其中包括用于目标1纵向分析的新生儿血斑和后来的口腔细胞，以及用于目标2的外祖母吸烟数据的完整三组。我们的目标是开发广泛适用于环境表观遗传学研究的通用方法，仅使用来自CHS的数据作为动机和说明。例如，我们预计这些方法可以用于分析来自国家儿童研究的三代子样本的表观遗传学数据。

DESCRIPTION (provided by applicant): It is now well established that epigenetic phenomena such as DNA methylation and histone modifications play an important role in gene expression, and alterations in epigenetic states can affect the risk of many diseases. The role of environmental factors, constitutional genotype, and their interactions in determining epigenetic states has been less well studied, however, particularly the potential for epigenetics to serve as a mediator between environmental exposures and disease. Several lines of evidence also suggest that certain environmental exposures can have effects on health outcomes in subsequent generations. This proposal has two broad methodological aims, to develop statistical methods (1) to assess the role of epigenetics in mediating exposure-response relationships within an individual, and (2) to assess its role in mediating transgenerational effects of exposure. For the first of these we will (aim 1a) develop a flexible measurement error framework that can accommodate variation in exposure and methylation over time and relate it to a variety of types of outcomes (binary, continuous, longitudinal, censored time-to-event) and (aim 1b) develop ways of summarizing complex methylation profiles and relating these profiles to both determinants of epigenetic status (environmental and genetic) and to the risk of disease. For the second aim, we will develop a variant of the transmission-disequilibrium test, extending genotypes at a given locus to include epigenetic mark, allowing for transmission of these marks from either parent in a non-Mendelian fashion. The methylation status of parents and offspring will be treated as latent variables that are functions of their own and/or their respective parents' exposures. The statistical performance of these methods will be studied by computer simulation. We will apply these methods to various subsets of participants in the Children's Health Study (CHS) - a long-term cohort study of the chronic effects of air pollution and other exposures like in utero and environmental tobacco smoke in 12,000 school children - for whom DNA are available for global and locus- specific methylation assays; these include neonatal blood spot and later buccal cells for the longitudinal analysis in aim 1 and complete trios with grandmaternal smoking data for aim 2. Our goal is to develop generalizable methodology that will be broadly applicable to studies of environmental epigenetics, using data from the CHS simply as motivation and illustration. For example, we anticipate that these methods could be useful for the analysis of epigenetic data from three-generation subsamples from the National Children's Study.