Third Generation Polygenic Score for Chronic Kidney Disease - Abstract: Polygenic scores combine small effects of GWAS risk variants across the genome to improve personalized risk prediction. The first-generation polygenic scores consisted of simple sums of risk alleles at significant GWAS loci and showed only modest associations with disease risk. The second-generation scores used genome-wide approaches to improve risk prediction by harnessing extra information from non-significant GWAS loci. Genome-wide polygenic scores (GPS) for several complex traits have now been shown to exhibit clinically actionable effects motivating similar approaches for CKD. Recently, we developed and validated a GPS for CKD optimized for cross-ancestry performance. The top 2% of the GPS was associated with >3-fold increased risk of CKD across ancestries, the degree of risk equivalent to a family history of kidney disease. Although these results are promising, several critical gaps remain. First, there is a need to further improve the predictive properties of the GPS, and this can be achieved with more powerful GWAS and more sophisticated statistical methods. Second, the clinical utility of the GPS continues to be limited by partial cross-ancestry transferability. This problem can be addressed by incorporating larger and more diverse GWAS, enhanced models of ancestry-specific loci (e.g. accounting for pN264K in APOL1), and novel statistical methods aimed at improving portability. Third, polygenic effects are context-dependent, thus there is a recognized need to evaluate GPS performance across diverse clinical contexts. Forth, there are no validated integrated models for CKD that combine polygenic, monogenic, family history, and clinical risk factors into a unified CKD risk stratification framework. In this proposal, we aim to formulate a third-generation GPS for CKD and address the above gaps in order to accelerate clinical implementation of polygenic prediction in nephrology. We propose to model polygenic risk with the latest multi-ancestry GWAS for renal function involving 3.4 million individuals, and externally validate the new GPS in large ancestrally diverse testing cohorts. We will test the GPS across heterogenous clinical contexts, including in the setting of monogenic kidney disease, glomerular disease, and kidney transplant. Lastly, we will develop and validate an integrated genomic risk score for CKD using a population-based cohort of 20,000 eMERGE-IV participants and the Columbia CKD Biobank cohort of 10,000 CKD patients. Our team benefits from the existing infrastructure for efficient genetic analyses, access to all relevant datasets, and first-rate local genotyping, bioinformatic, statistical and computing resources, thus we are ideally positioned to make this project successful.
