eMERGE Phase IV Clinical Center at Mass General Brigham

摘要：为了使PRS的应用得以开发和实施，我们的Emerge IV提案来自 合作伙伴医疗保健利用大型生物库(&GT；105,000人同意40,000人的基因数据)，临床 电子健康记录(EHR)中的数据来自最大的综合医疗保健系统中的400万名患者 新英格兰的提供商、先进的生物信息学专业知识、在PRS开发方面的领先地位和州- 最先进的基因分析，在返回基因组结果方面的成熟专业知识，以及使用 改变临床流程和评估结果的信息技术。我们建议在我们的基础上 实现具体目标的专门知识：目标1(发现)：假设：多基因风险得分将 允许我们根据常见复杂特征的遗传风险对出现的受试者进行分层。使用 最大的可用基因组数据资源，我们将计算和验证用于冠状动脉疾病的新PR， 不同祖先的房颤、2型糖尿病、结直肠癌和严重抑郁症。我们会1) 比较和基准现有的PRS建造方法在不同祖先群体中的性能， 2)开发新的统计方法，用于稳健的跨种族PRS预测，并将PRS与STESTED相结合 临床危险因素和家族史。我们将从我们的网络同事那里获得额外的15-e- 表型(共20个)，目标是确定高危个人，例如，前2%的PR风险目标2 (RiskInsight Report/ELSI)：我们将开发一份包含临床风险因素、家族的《风险洞察报告》 历史，以及为高风险参与者提供循证建议的PR(前2% 表型特定的PRS分布)，用于电子临床实施。我们将评估风险 ELSI子目标1中的通信格式：测试两个模拟的影响和可解释性 RiskInsight报告将PR概括为：(A)二分型(将患者定义为高危VS 中等/低风险)或(B)定量(为患者提供风险百分位数的数字估计)， 在两种情况下都有关联的临床建议。然后，我们将通过对不同的HCP和 患者，模拟报告被HCP和患者理解的程度。目标3(成果)： 假设：医生将根据ECDs改变他们对患者的监测和治疗 RiskInsight报道。在高危受试者的HCP中，我们将看到至少一种临床变化 在与受试者进行披露讨论后的关注。我们将招募2500名参与者，以实施 RiskInsight中的临床PR报告使用智能FHIR应用程序与EHR集成的ECDS。这个 主要结果将是在收到以下定义的e-CDS后12个月内是否有任何HCP采取任何行动 订购筛查测试，开预防性药物，或提供生活方式建议。我们将进行 向高风险参与者披露结果的效果分析，以确定结果如何个性化 改变了患者在实验室价值、降低风险行为或医疗保健利用方面的结果。

Abstract: To enable the application of PRS development and implementation, our eMERGE IV proposal from Partners HealthCare leverages a large biobank (>105,000 consented with genotype data on 40,000), clinical data in the electronic health records (EHR) for >4 million patients from the largest integrated health care provider in New England, advanced bioinformatics expertise, prior leadership in PRS development and state- of-the-art genetic analysis, established expertise in returning genomics results, and experience using information technology to transform clinical processes and assessing outcomes. We propose to build on our expertise to accomplish the specific aims: Aim 1 (Discovery): Hypothesis: Polygenic risks scores will allow us to stratify eMERGE subjects based on genetic risk for common complex traits. Using the largest available genomic data resources, we will calculate and validate new PRS for coronary artery disease, atrial fibrillation, type 2 diabetes, colorectal cancer and major depression across diverse ancestries. We will 1) compare and benchmark the performance of existing PRS construction methods in different ancestral groups, 2) develop novel statistical methods for robust trans-ethnic PRS prediction, and integrate PRS with established clinical risk factors and family history. We will obtain PRS from our network colleagues for an additional 15-e- phenotypes (total 20) with a goal of identifying high-risk individuals, e.g., top 2% of PRS risk Aim 2 (RiskInsight Report/ELSI): We will develop a “Risk Insight Report” with clinical risk factors, family history, and PRS with evidence-based recommendations for high risk participants (top 2% of phenotype specific PRS distribution) for electronic clinical implementation. We will assess risk communication formats in our ELSI Sub-Aim 1: To test the impact and interpretability of two mock RiskInsight Reports summarizing PRS as either (a) dichotomous (defining the patient as high-risk vs intermediate/low risk) or (b) quantitative (providing a numerical estimate of the percentile of risk for the patient), with linked clinical recommendations in both cases. We will then assess, through surveys of diverse HCPs and patients, the extent to which the mock reports are understood by both HCPs and patients. Aim 3 (Outcomes): Hypothesis: Physicians will alter their surveillance and treatment of patients based on eCDS of RiskInsight Reports. Among HCPs for high-risk subjects, we will see at least one change in clinical care after disclosure discussions with subjects. We will recruit 2500 participants for implementation of clinical PRS in RiskInsight Reports using a SMART on FHIR app for eCDS integrated with the EHR. The primary outcome will be whether any HCP took any action within 12 months after receipt of e-CDS defined by ordering screening tests, prescribing a preventive medication, or providing lifestyle advice. We will conduct analyses of the effect of disclosing results to high risk participants to determine how personalized results changed patient outcomes in laboratory values, risk reduction behaviors, or health care utilization.