Evaluation of the genome-informed risk assessment (GIRA) approach from eMERGE in an independent health system

Abstract

The Genome Informed Risk Assessment (GIRA) from eMERGE is an ongoing pragmatic prospective study designed to implement and evaluate genomic precision medicine across diverse clinical settings. Here, we examine the utility of the high-risk criteria from GIRA in assessing health risk through a retrospective evaluation of 9 adult conditions in a health system independent of eMERGE using the Penn Medicine Biobank (PMBB, n=48,279). We find a large proportion of patients - 30.4% (n=14,676) - labeled as high-risk based on the genomic components (monogenic and polygenic) of GIRA. Stratifying by ancestry revealed significant differences in high-risk classification, with higher rates in African/African American (56.6% vs. 50.1%, p=7.43x10-36) and lower rates in East (42.0%) and South Asian (40.0%) ancestries when considering all three high-risk components of GIRA. Enrichments of high-risk individuals were observed in the highest quartile of social deprivation index highlighting contextual influences on high-risk measures. The polygenic component of GIRA was a good predictor of prevalent cases. GIRA predicted incident disease albeit with lower accuracies for some conditions. We find demographic compositions of high-risk patients differed from the incident cases for certain conditions; for example, high-risk AFIB individuals where enriched for European ancestries in contrast with incident AFIB cases that were enriched for AFR ancestries. Overall, our results show the accuracy of GIRAs high-risk criteria as a biomarker to stratify high-risk patients for precision medicine and highlight its potential impact on the health system if implemented at scale.