EHR-based Genomic Risk Assessment and Management for Diverse Populations

PROJECT SUMMARY/ABSTRACT Recently, large-scale genome-wide association studies (GWAS) provide evidence for a substantial polygenic contribution to the risk of many common complex diseases. However, most of these studies were performed in Europeans, and new data and methods are necessary to tailor polygenic risk prediction to non-Europeans, to ensure that genomic stratification does not further exacerbate health disparities. The overarching goal of the eMERGE-IV network is to leverage genetic and electronic health record (EHR) data for diverse populations to design, validate and test the clinical utility of ancestry-tailored polygenic risk scores for common diseases. As a current member of the eMERGE network, Columbia University has significantly advanced its goals, having recruited over 2,500 diverse patients for sequencing and return of actionable findings, leading the effort to transition the network to the OMOP Common Data Model to improve the efficiency, accuracy, reproducibility and portability of electronic phenotypes, and contributing a widely-adopted XML parser for structuring genetic test reports. Since our last application, the Columbia Precision Medicine Initiative has also grown and now includes participation in several national initiatives, such as the All-of-Us program, in which we have demonstrated our ability to rapidly recruit patients under-represented in biomedical research. Our scientific expertise combined with our strong tradition of patient-centered research and community engagement in a socioeconomically, racially, and ethnically diverse community of Northern Manhattan, positions us to successfully contribute as the Enhanced Diversity Clinical Site of the eEMERGE-IV network. We will leverage our prior experience with eMERGE, scientific expertise, and knowledge gained from participation in other national precision medicine initiatives to develop, optimize, validate and disseminate ancestry-tailored genomic risk assessment and clinical management tools. In Aim 1, we will continue to advance electronic phenotyping by contributing sharable natural language processing tools for converting clinical text into OMOP-based discrete data and facilitating phenotype interoperability. In Aim 2, we will develop and optimize accurate ancestry-tailored genome-wide polygenic predictors, integrate them with clinical risk predictions, and test their performance in diverse populations. In Aim 3, we will investigate ELSI issues related to the return of health risk predictions to diverse patients by ascertaining patients’, clinicians’, and IRB members’ views through focus groups. In Aim 4, we will develop portable EHR plug-ins to facilitate prospective risk communication and management using integrated genomic data, family history, and clinical data. In Aim 5, we will recruit 2,500 diverse patients and use a randomized controlled trial design to assess the impact of return of genomic prediction on the accuracy of risk perception, health surveillance, and risk reducing measures. This proposal will address major knowledge gaps in genetic risk assessment for diverse populations, and the solutions and knowledge gained will be broadly applicable to precision medicine for common complex traits across many clinical specialties.