Genetic Risk Prediction of Breast Cancer for African Americans

项目摘要 全基因组关联已确定了180多个乳腺癌易感基因 研究(GWAS)，主要是在高加索人口。然而，这些风险变量中的许多不能直接 在非洲血统的女性身上也出现了这种情况，这表明因果变异尚未确定。多基因风险 积分值(PR)聚集了GWAS确定的常见遗传变异，已被开发用于预测 高加索女性患乳腺癌的遗传风险，但没有针对非裔美国女性的有效PR。 非洲血统群体的连锁平衡远不如高加索人和亚洲人广泛 种群，这使得非洲裔人口成为寻找致病变异的理想人群 定位乳腺癌易感基因。因此，我们建议进行一项全面的分析研究， 利用我们和公共领域现有的几种类型的乳腺癌基因数据集来 解决三个具体目标。首先，我们的目标是进行跨种族精细映射分析，以缩小范围 在180+个乳腺癌基因座中随机变异候选名单。我们将汇编和协调遗传 来自非洲裔女性乳腺癌研究的数据，包括7525例病例和6207名对照， 并利用高加索人(122,000例病例和105,000例对照)、东亚人的关联结果 (14,000个病例和13,000个对照)和拉丁裔(4,400个病例和7,500个对照)。我们将使用贝叶斯 统计方法，直接合并多个功能注释的顶端变异在每个基因座。 其次，我们的目标是通过以下方式在非裔美国人中建立乳腺癌多基因风险评分模型 利用功能注释、连锁不平衡和基因表达数据。几个PRS将 针对总体乳腺癌风险和雌激素受体进行开发，内部交叉验证和验证 与外部研究相结合。第三，我们的目标是开发乳腺癌风险预测模型，将两者结合起来 遗传因素和非遗传因素。拟议的研究将有效地利用几种类型的现有数据 使用创新的综合方法，并有可能通过缩小 包含因果变异的遗传区域。更重要的是，风险预测模型具有很好的潜力 将GWAS的知识转化为乳腺癌筛查的实践。

PROJECT ABSTRACT More than 180 susceptibility loci for breast cancer have been identified by genome-wide association studies (GWAS), mainly in Caucasian populations. However, many of these risk variants cannot be directly replicated in women of African ancestry, suggesting that causal variants are yet to be identified. Polygenic risk scores (PRS), which aggregate common genetic variants identified by GWAS, have been developed to predict genetic risk of breast cancer for Caucasian women, but there is no validated PRS for African American women. The linkage equilibrium in African ancestry populations is much less extensive than in Caucasian and Asian populations, which makes African ancestry population the ideal population to find causative variants after localizing a breast cancer susceptibility locus. Therefore, we propose a comprehensive analytical study that leverages several types of existing genetic datasets for breast cancer available to us and in public domains to address three specific aims. First, we aim to conduct cross-ethnic fine-mapping analysis for narrowing down casual variant candidate lists in 180+ loci of breast cancer. We will compile and harmonize genetic data from studies of breast cancer in women of African ancestry, including 7,525 cases and 6,207 controls, and leverage the association results from Caucasians (>122,000 cases and >105,000 controls), East Asians (>14,000 cases and >13,000 controls), and Latinos (4,400 cases and 7,500 controls). We will use a Bayesian statistical method to directly incorporate multiple functional annotations for the top variants in each locus. Second, we aim to develop breast cancer polygenic risk score models in African Americans by leveraging functional annotations, linkage disequilibrium, and gene expression data. Several PRSs will be developed for overall breast cancer risk and by estrogen receptor, cross-validated internally, and validated with external studies. Third, we aim to develop breast cancer risk prediction model by combining both genetic and non-genetic factors. The proposed study will efficiently utilize several types of existing data using innovative integrative approaches and has the potential to advance the field by narrowing down the genetic regions containing causal variants. More importantly, the risk prediction model has a good potential to translate knowledge from GWAS to the practice of breast cancer screening.