Elucidating hereditary transthyretin-mediated heart failure risk using machine learning, polygenic risk and recall by genotype approaches in African ancestry individuals

项目总结/摘要 甲状腺素运载蛋白（TTR）基因的突变可导致异常淀粉样纤维沉积在 心肌，导致遗传性甲状腺素运载蛋白淀粉样心肌病（hATTR-CM），并导致心脏 失败最近已经开发出针对hATTR-CM的靶向疗法，并且已经显示出改善死亡率 和住院治疗。 最近，我们领导了一项研究（美国医学会杂志，2019年12月），该研究表明TTR V122 I突变，常见于种族/少数民族（非洲裔美国人（AA）为4%， 西班牙裔美国人（HAs）），使心力衰竭的风险增加两倍。尽管有这种强烈的影响，只有11%的人 的V122 I携带者心力衰竭被hATTR-CM正确诊断，这表明 疾病的诊断不足和误诊。我们进一步显示了亚临床证据， 年轻无症状V122 I携带者的超声心动图紊乱，提示可能出现早期体征 在疾病发作之前。 我们建议通过解决知识差距来扩展我们以前的工作，这是有针对性的 治疗，以充分发挥其潜力。包括：了解V122 I的不完全转化; 在基因分型不常见的大型医疗保健系统中识别V122 I携带者; 亚临床疾病负担。在目标1中，我们将研究多基因风险评分之间的相互作用， 由数百万个影响很小的单核苷酸变异组成，V122 I是一种单基因突变， 在6，609例AA和9，006例AA中，结合心力衰竭的临床风险因素分析了单一强效应 BioMe生物库中的HA和Penn Medicine生物库（PMBB）中的5，833个AA。在目标2中，我们将应用 机器学习工具，以多模态电子健康记录（EHR）数据识别V122 I携带者， 从西奈山的电子健康记录（EHR）数据存储库中提取了100万名患者。在目标3中，我们 评估淀粉样蛋白沉积对年轻、无症状、 V122 I载体通过召回V122 I载体进行成像评价，包括研究级超声心动图， 心脏磁共振和锝核扫描。 该提议是创新的，因为我们正在利用两个大型的不同祖先的EHR相关生物库， 学术卫生系统（西奈山的BioMe和宾夕法尼亚大学的PMBB），沿着 采用先进的方法，包括多种族多基因风险评分和机器学习方法 多模态EHR数据。我们进一步提出了基于基因型的患者回忆，并进行了深入的 使用综合心脏成像扫描进行表型分析。 该提案有可能实现种族/民族心力衰竭精准医学的潜力 通过向少数群体提供临床护理、人口管理、风险分层和临床试验方面的信息。

PROJECT SUMMARY / ABSTRACT Mutations in the Transthyretin (TTR) gene can lead to deposition of abnormal amyloid fibrils in the myocardium, resulting in hereditary transthyretin amyloid cardiomyopathy (hATTR-CM) and leading to heart failure. Targeted therapies for hATTR-CM have recently been developed and have shown to improve mortality and hospitalization. Recently, we led a study (Journal of American Medical Association, Dec 2019) that showed that the TTR V122I mutation, commonly observed in racial/ethnic minorities (4% in African Americans (AAs) and 1% in Hispanic Americans (HAs)), confers two-fold increased risk of heart failure. Despite this strong effect, only 11% of V122I carriers with heart failure were appropriately diagnosed with hATTR-CM, suggesting marked underdiagnosis and mis-diagnosis of the disease. We further showed subclinical evidence of echocardiographic derangements in young, asymptomatic V122I carriers, suggesting early signs can occur well before onset of disease. We propose to extend our prior work by addressing knowledge gaps which are necessary for targeted therapies to attain their full potential. These include: understanding the incomplete penetrance of V122I; identifying V122I carriers in large health care systems where genotyping is not common; and understanding subclinical disease burden. In Aim 1, we will examine the interplay between a polygenic risk score, which are comprised of millions of single nucleotide variants with small effects, and V122I, a monogenic mutation with a single strong effect, analyzed in conjunction with clinical risk factors on heart failure in in 6,609 AAs and 9,006 HAs in the BioMe biobank and 5,833 AAs in the Penn Medicine Biobank (PMBB). In Aim 2, we will apply machine learning tools to multi-modal electronic health record (EHR) data to identify V122I carriers in ~8 million patients from an electronic health record (EHR) data repository at Mount Sinai. In Aim 3, we will evaluate subclinical effects of amyloid deposition on cardiac structural/functional traits in young, asymptomatic V122I carriers by recalling V122I carriers for imaging evaluation including research-grade echocardiograms, cardiac magnetic resonance and technetium nuclear scanning. The proposal is innovative because we are utilizing two large diverse ancestry EHR-linked biobanks from academic health systems (BioMe at Mount Sinai, and PMBB at University of Pennsylvania), along with adopting cutting-edge methods including multi-ethnic polygenic risk scores, and machine learning approaches on multi-modal EHR data. We further propose patient recall based on genotypes and perform deep phenotyping using comprehensive heart imaging scans. This proposal has the potential to realize the potential of precision medicine for heart failure in racial/ethnic minorities by informing clinical care, population management, risk stratification and clinical trials.