A Polygenic Risk Score to Refine Risk Stratification and Prediction for Severe Liver Disease by Clinical Fibrosis Scores

Background & Aims: A polygenic risk score based on well-known genetic variants in PNPLA3, TM6SF2, MBOAT7, and GCKR predicts hepatic fat content (polygenic risk score-hepatic fat content [PRS-HFC]). Here, we hypothesized that the addition of PRS-HFC to clinical fibrosis scores may improve risk stratification and prediction of severe liver disease (SLD). Methods: We used data from 266,687 individuals in the UK Biobank, evaluating the incidence of cirrhosis, decompensated liver disease, hepatocellular carcinoma, and/or liver transplantation during a median follow-up period of 9 years. Nonalcoholic fatty liver disease fibrosis score, Fibrosis-4, aspartate aminotransferase–to–platelet ratio, BARD, and Forns scores, and PRS-HFC, were computed. All analyses were stratified according to the presence of diabetes, obesity, and a positive fatty liver index (≥60). Results: Unfavorable genetics (PRS-HFC, ≥0.396) further stratified the risk of SLD in subjects in intermediate-/high-risk classes of fibrosis scores, with a higher effect in those with metabolic risk factors, and the prediction was improved by integrating PRS-HFC (areas under the receiver operating characteristic increased for all scores with a P value of approximately 10-2 to 10-4, except for the aspartate aminotransferase–to–platelet ratio in the overall population and in subjects with obesity). PRS-HFC improved diagnostic accuracies and positive predictive values for SLD in intermediate–high clinical score risk classes. Risk stratification and prediction were not affected or were poorly affected by unfavorable genetics in subjects without metabolic risk factors. Conclusions: Integration of genetics with clinical fibrosis scores refines individual risk and prediction for SLD, mainly in individuals at risk for nonalcoholic fatty liver disease. These data provide evidence from a prospective cohort that common genetic variants capture additional prognostic insights not conveyed by validated clinical/biochemical parameters.