Validating Cancer Risk Models: a Pilot Study to Evaluate Cost-efficient Methods

摘要：为了促进有针对性的癌症预防策略，帮助患者权衡基因定制干预的成本和收益，我们需要统计模型，提供准确和精确的预测一个人发展特定不良后果的风险。这样的模型必须使用有风险个体的队列数据进行评估。理想的评估是在一个大队列中对所有受试者进行基因分型，使用基因分型对所有队列成员分配风险，然后将分配的风险与随后的结果联系起来。然而，这种完整的基因分型成本太高，不可行。我们的目标是测试两种新的具有成本效益的方法来评估遗传协变量的附加价值。第一种是病例队列设计，仅对队列成员的一个子集进行基因分型。第二种方法使用病例对照数据和贝叶斯规则来估计基因分型增加的精度。我们的目标是使用来自加州教师研究（CTS）的队列数据和基于人群的病例对照数据来检验这两种方法。我们将利用以下数据评估现有卵巢癌和乳腺癌风险模型：a)所有CTS受试者；b)病例队列设计中的CTS受试者子集；c)病例对照研究的受试者。我们将使用BRCAPRO评分作为BRCA1和BRCA2致病性突变的替代品。这将允许我们将使用(b)和(c)的推论与(a)的推论进行比较，这将形成黄金标准。

DESCRIPTION (provided by applicant): Validating Cancer Risk Models: a Pilot Study to Evaluate Cost-efficient Methods Abstract To facilitate targeted cancer prevention strategies and help patients weigh the costs and benefits of genetically tailored interventions, we need statistical models that provide accurate and precise projections of a person's risk of developing a given adverse outcome. Such models must be evaluated using cohort data on individuals at risk of the outcome. The ideal evaluation would genotype all subjects in a large cohort, use the genotypes to assign risks to all cohort members, and then relate assigned risks to subsequent outcomes. However such complete genotyping is too costly to be feasible. Our goal here is to test two new cost-effective methods for evaluating the added value of genetic covariates. The first is a case-cohort design that genotypes only a subset of the cohort members. The second uses case-control data and Bayes Rule to estimate the amount of increased precision gained with genotyping. Our goal is to use cohort data from the California Teachers Study (CTS) and population-based case-control data to test the two methods. We will do so by evaluating existing risk models for cancers of the ovary and breast using data from: a) all CTS subjects; b) a subset of CTS subjects in a case- cohort design; and c) subjects from a case-control study. We will use BRCAPRO scores as surrogates for pathogenic mutations of BRCA1 and BRCA2. This will allow us to compare inferences using (b) and (c) with those from (a), which will form the gold standard. PUBLIC HEALTH RELEVANCE: Project Narrative To facilitate targeted cancer prevention strategies and help patients weigh the costs and benefits of genetically tailored interventions, we need statistical models that provide accurate and precise projections of a person's risk of developing a given adverse outcome. The goal of this project is to test two new cost-effective methods for evaluating the value of adding genetic covariates to these models using data from studies of cancers of the breast and ovary.