Guidelines from the PRIMED Consortium on the use of population descriptors in polygenic risk model development and application

Abstract Whether and how to use descriptors such as race, ethnicity, and ancestry in genomic research and clinical risk models is an ongoing and evolving topic of debate, because how we refer to populations and individuals can have known and unforeseen consequences, including detrimental scientific, clinical, social, and ethical impacts. Clinical risk models incorporating polygenic risk scores (PRS) may perform differently for individuals from populations different from those used in training and validation; in particular, the correlation between polygenic predictions and predicted traits often decrease as a function of genetic distance between training and target samples. Thus, describing the populations used to develop risk models and the populations where they will be applied is essential to evaluate clinical validity and utility. Previous work has often used race, ethnicity or continental ancestry (inferred from self-report and/or genetic data) to describe training samples and target populations, but such labels run the risk of reinforcing biological determinist views of race and may not capture important aspects of the population that can affect polygenic model performance, such as region and socioeconomic status. Population descriptors may also be directly included in clinical risk models. The Polygenic RIsk MEthods in Diverse Populations (PRIMED) Consortium covened a multidisciplinary Population Descriptors Working Group to propose guidelines on conceptualizing, applying and communicating population descriptor information in the context of PRS development and application. These guidelines emphasize the importance of matching the chosen descriptors to the scientific and clinical goal; transparency; and engaging affected communities and subject matter experts to ensure that adequate considerations are given to ethical, legal, and social implications of their research and its communication. We present these guidelines and an accompanying data model designed to support transparency and flexibility. Citation Format: Peter Kraft, Iain Konigsberg, Stephanie Gogarten. Guidelines from the PRIMED Consortium on the use of population descriptors in polygenic risk model development and application [abstract]. In: Proceedings of the 16th AACR Conference on the Science of Cancer Health Disparities in Racial/Ethnic Minorities and the Medically Underserved; 2023 Sep 29-Oct 2;Orlando, FL. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2023;32(12 Suppl):Abstract nr B109.