A combined polygenic risk score provides a data-driven biomarker for metabolic syndrome

Leiby, Jacob S.; Choe, Eun Kyung; Kim, Dokyoon, University of Pennsylvania School of Medicine

Metabolic syndrome (MetS) is defined by a set of metabolic conditions: increased blood pressure, elevated triglycerides, a large waistline, low HDL cholesterol, and increased blood glucose levels. MetS confers an increased risk of cardiovascular disease and type 2 diabetes. Diagnosis occurs if three or more of these conditions are met; however, the definition thresholds vary between different health organizations. In this study, we aim to develop a more objective marker of MetS through genetic risk scores. By combining the polygenic risk scores (PRSs) for each condition, we propose a novel genetic risk score for MetS. The target dataset for genetic and phenotypic analysis is a Korean comprehensive health-checkup cohort, GENIE. The summary statistics used in creating the PRSs are from the Korean Genome and Epidemiology Study. For each MetS component, we calculated a PRS and used it to predict MetS. Next, we developed a composite PRS (cPRS) by combining each of the component scores into a new feature and used this to predict MetS. We analyzed the strength of association of cPRS with MetS by comparing the PRS odds ratios (OR) and by splitting the individuals into the top 10%, middle 20%, and bottom 10% of cPRS and comparing the OR between groups. The AUC for predicting MetS from the independent condition PRSs ranged from 0.68-0.69 and increased to 0.70 for cPRS. The OR for the condition PRSs ranged from 1.07-1.26 and increased to 1.30 for cPRS. The cPRS quantile analysis shows an OR of 1.39 comparing the top to middle groups, 0.74 comparing the bottom to middle, and 3.03 comparing the top to bottom groups. The results show that the cPRS can better predict MetS than its component parts. A further analysis will compare cPRS to MetS PRS to explore how they differ in association with MetS.