1/4 Leveraging EHR-linked biobanks for deep phenotyping, polygenic risk score modeling, and outcomes analysis in psychiatric disorders

项目摘要 重度抑郁症（MDD）、焦虑症和物质使用障碍（SUD）是常见的、复杂的、 经常共同出现并与显著功能障碍相关的精神病学特征， 医疗保健利用率和成本，以及更高的死亡风险。这三种情况不仅在中国非常普遍， 并产生巨大的社会负担，但我们的团队和其他人最近的研究表明， 显示来自共同遗传变异的共享协方差显著有助于这些精神疾病， 合并症。需要大量的数据集来了解遗传学的多方面相互作用，包括 多基因风险评分（PRS）和健康因素的社会决定因素，如就业和教育 达到，可以增加这些精神疾病和临床结果的风险，如多种 精神病住院治疗PRS已显示出风险预测的潜力，但PRS用于 精神疾病的研究才刚刚开始电子健康记录（EHR）的使用提供了承诺 在临床实践中观察到的患者队列中检查这些关系。但 在精神疾病及其治疗的研究中，电子病历的使用还处于起步阶段。这项研究将解决 在“基因型-精神病表型”关系方面， 来自四个医疗机构EHR相关生物库的地理多样性人群样本 中心-哥伦比亚，康奈尔，马约诊所和西奈山。我们的目标是：（1）改进方法 用于MDD、焦虑和SUD的EHR表型分析，以及基于> 3000万数据集的相关结果 EHR，（2）评估PRS和这些条件之间的关联，以及（3）评估关联 PRS和结局之间的关系，包括MDD中的治疗抵抗和患者的医疗保健利用 抑郁症、焦虑症和SUD。PRS分析将使用来自生物库的数据，其中> 50，000人具有两种EHR GWAS数据这项研究的成功完成将产生新的数据，以提高我们对 PRS对常见精神疾病的临床效用。

PROJECT ABSTRACT Major depressive disorder (MDD), anxiety disorders, and substance use disorders (SUDs) are common, complex psychiatric traits that frequently co-occur and are associated with significant functional impairment, increased healthcare utilization and cost, and higher mortality risk. Not only are these three conditions highly prevalent in the general population and generate a huge societal burden, but recent studies by our team and others have shown that shared covariance from common genetic variation significantly contributes to these psychiatric comorbidities. Large data sets are needed to understand how the multifaceted interplay of genetics, including polygenic risk scores (PRSs), and social determinants of health factors, such as employment and educational attainment, can increase the risk of these psychiatric disorders and clinical outcomes, such as multiple psychiatric hospitalizations. PRSs have shown potential for risk prediction, but the clinical utility of PRSs for psychiatric conditions is just starting to be explored. Use of Electronic Health Records (EHRs) offers the promise of large data sets to examine these relationships in cohorts of patients seen in clinical practice. However, the use of EHRs is in its infancy in the study of psychiatric disorders and their treatment. This study will address critical knowledge gaps in “genotype-psychiatric phenotype” relationships in large, demographically and geographically diverse population-based samples derived from EHR-linked biobanks across four medical centers - Columbia, Cornell, Mayo Clinic and Mount Sinai. Our objectives are to (1) develop improved methods for EHR phenotyping of MDD, anxiety, and SUDs, and related outcomes based on a data-set of >30 million EHRs, (2) evaluate associations between PRSs and these conditions, as well as (3) assess the association between PRSs and outcomes including treatment resistance in MDD and healthcare utilization in patients with MDD, anxiety and SUD. The PRS analyses will utilize data from biobanks with >50,000 persons with both EHR and GWAS data. Successful completion of this study will generate new data in improving our understanding of the clinical utility of PRSs for commonly occurring psychiatric disorders.