Statistical and ethical considerations for the use of ancestry in polygenic risk scores

Abstract

The polygenic risk score (PRS) is an emerging tool in precision medicine that promises to provide individualized genetic risk estimates for developing a given trait or disease of interest. PRS have already shown improvement in risk assessments for certain conditions (e.g., cardiovascular disease) when used in conjunction with existing clinical risk assessment. However, many PRS have worse predictive accuracy for individuals of mixed or non-European genetic ancestry due to the overrepresentation of individuals of European descent in genomic research, leading to concerns of contributing to healthcare inequity. In this study, I explore the analytical and ethical challenges of addressing equity concerns in the context of using concepts of ancestry in PRS. First, I assessed the concordance of two PRS models for red blood cell traits in individuals with African ancestry to investigate whether both models identify the same individuals at high risk. The study revealed analytical challenges involved in working with underrepresented populations and small sample sizes with respect to PRS accuracy and reliability. Second, I conducted an ethical analysis focused on understanding the conceptualization of equity with respect to PRS and illustrated the practical constraints in achieving goals of equitable PRS development given the infrastructural constraints in genomic research practice. Finally, I interviewed experts in the field who are actively developing PRS for diverse populations to capture their experiences, challenges, and outlook on the use of race, ethnicity, and ancestry in PRS. I conclude by summarizing results of this study and offering suggestions for moving forward in addressing the urgent issue of equity in PRS.