BACKGROUND & AIMS There are no biomarkers of nonalcoholic steatohepatitis (NASH) that are ready for routine clinical use. We investigated whether an analysis of PNPLA3 and TM6SF2 genotypes (rs738409 and rs58542926) can be used to identify patients with nonalcoholic fatty liver disease (NAFLD), with and without diabetes, who also have NASH.

METHODS We collected data from the Boramae registry in Korea on 453 patients with biopsy-proven NAFLD with sufficient clinical data for calculating scores. Patients enrolled from February 2014 through March 2016 were assigned to cohort 1 (n = 302; discovery cohort) and patients enrolled thereafter were assigned to cohort 2 (n = 151; validation cohort). DNA samples were obtained from all participants and analyzed for the PNPLA3 rs738409 C>G, TM6SF2 rs58542926 C>T, SREBF2 rs133291 C>T, MBOAT7-TMC4 rs641738 C>T, and HSD17B13 rs72613567 adenine insertion (A-INS) polymorphisms. We used multivariable logistic regression analyses with stepwise backward selection to build a model to determine patients' risk for NASH (NASH PT) using the genotype and clinical data from cohort 1 and tested its accuracy in cohort 2. We used the receiver operating characteristic (ROC) curve to compare the diagnostic performances of the NASH PT and the NASH scoring systems.

RESULTS We developed a NASH PT scoring system based on PNPLA3 and TM6SF2 genotypes, diabetes status, insulin resistance, and levels of aspartate aminotransferase and high-sensitivity C-reactive protein. NASH PT scores identified patients with NASH with an area under the ROC (AUROC) of 0.859 (95% CI, 0.817-0.901) in cohort 1. In cohort 2, NASH PT scores identified patients with NASH with an AUROC of 0.787 (95% CI, 0.715-0.860), which was significantly higher than the AUROC of the NASH score (AUROC, 0.729; 95% CI, 0.647-0.812; P = .007). The AUROC of the NASH PT score for detecting NASH in patients with NAFLD with diabetes was 0.835 (95% CI, 0.776-0.895) and in patients without diabetes was 0.809 (95% CI, 0.757-0.861). The negative predictive value of the NASH PT score <-0.785 for NASH in patients with NAFLD with diabetes reached 0.905.

CONCLUSIONS We developed a scoring system, based on polymorphisms in PNPLA3 and TM6SF2 and clinical factors that identifies patients with NAFLD, with or without diabetes, who have NASH, with an AUROC value of 0.787. This system might help clinicians better identify NAFLD patients at risk for NASH.