A genomics-based strategy to precision phenotyping and drug repositioning in cardiometabolic diseases

项目摘要/摘要 目前，美国共有1820万人患有2型糖尿病和非酒精性脂肪肝 (NAFLD)。心脏代谢性疾病的易感性是高度可变的，目前还没有FDA批准的药物 存在治疗非酒精性脂肪肝的方法。最近的研究表明，心脏代谢性疾病有几个共同的遗传因素， 我们的长期目标是揭示基因组和非基因组风险因素之间的复杂相互作用。 疾病的发展，以改进风险预测和确定重新用于治疗NAFLD的药物靶点。 在我们的第一个目标中，我们将应用英国生物库中的因果单一和多重因果中介分析来确定 可作为2型糖尿病潜在干预靶点的中间或适度的内表型 和非酒精性脂肪肝。我们提供了一个精确的表型鉴定框架，并量化了多少个体水平的基因 如果对中间内表型进行干预，疾病的负担可以减轻。它最终可能会 使临床医生能够检测到早期偏离特定于患者的基线风险，而他们自己仍然 无症状，预示着随后的疾病症状的出现。 我们的第二个目标是识别和验证潜在的用于NAFLD的药物靶点，使用基因组和 真实世界的数据。我们将使用两种方法确定候选药物靶点：1)工具变量分析 使用“可用药”基因组的遗传工具(例如孟德尔随机化分析)和2)a 基于药物活性知识和基因表达特征的计算基因表达签名方法 疾病病理生理学。对候选药物靶点的药物疗效的预测有效性将使用 退伍军人健康管理局医疗系统中910万退伍军人的真实数据，360万 宾夕法尼亚大学医学临床数据仓库中的患者和Vanderbilt合成数据库中的350万患者 衍生品。NAFLD患者的长期治疗效果将通过模拟靶向试验进行评估 肝硬变、肝失代偿、肝移植和肝癌作为主要治疗终点 五年的跟踪调查。使用自身对照病例对健康患者进行短期药物疗效评估 以丙氨酸转移酶变化为主要结果的系列分析。预计我们的基因组学- 药物再利用的知情和药物流行病学方法将加快药物发现工作 并导致使用现有的药物治疗NAFLD，缩短了药物开发时间。

Project Summary / Abstract A total of 18.2 million people in the U.S. currently live with type 2 diabetes and non-alcoholic fatty liver disease (NAFLD). Susceptibility to cardiometabolic diseases is highly variable, and currently no FDA-approved drugs exist to treat NAFLD. Recent work suggests that cardiometabolic diseases share several genetic factors, and our long-term goal is to reveal the complex interplay between genomic and non-genomic risk factors in the development of disease to improve risk prediction and identify drug targets for repurposing to treat NAFLD. In our first aim we will apply causal single and multiple causal mediation analysis in the UK Biobank to identify intermediate or moderating endophenotypes that can serve as potential intervention targets for type 2 diabetes and NAFLD. We provide a framework for precision phenotyping and quantify how much individual-level genetic burden for disease can be reduced if one would intervene on intermediary endophenotypes. It may ultimately enable clinicians to detect early departures from patient-specific baseline risk that, while themselves are still asymptomatic, are predictive of the subsequent onset of disease symptoms. Our second aim is to identify and validate drug targets for potential repurposing in NAFLD using genomic and real-world data. We will identify candidate drug targets using two approaches: 1) instrumental variable analysis using genetic instruments of the `druggable' genome (e.g., Mendelian Randomization analysis) and 2) a computational gene expression signature-based approach based on the knowledge of drug activity and disease pathophysiology. Predictive validity of drug efficacy for candidate drug targets will be assessed using real-world data of 9.1 million Veterans in the Veterans Health Administration healthcare system, 3.6 million patients in the Penn Medicine clinical data warehouse, and 3.5 million patients in the Vanderbilt Synthetic Derivative. Long-term therapeutic efficacy will be evaluated using emulated target trials in NAFLD patients with cirrhosis, hepatic decompensation, liver transplant, and liver cancer as the primary treatment endpoint during five years of follow-up. Short-term drug efficacy will be evaluated in healthy patients using self-controlled case series analysis with change in alanine transferase as the primary outcome. It is anticipated that our genomics- informed and pharmaco-epidemiological approach to drug repurposing will accelerate drug-discovery efforts and lead to the use of existing agents to treat NAFLD with shortened drug development times.