Genome-wide association and polygenic risk score analyses of MASLD-related hepatic steatosis severity in a Taiwanese Han population.
TL;DR
This study delineates the genetic landscape of ultrasound-defined hepatic steatosis severity in a large Taiwanese Han population and demonstrates the robust ability of the polygenic risk score under continuous shrinkage to predict disease stage.
Abstract
Background
Hepatic steatosis is a highly prevalent metabolic condition and a major contributor to chronic liver disease, particularly in Asian populations. This study was conducted to identify genetic variants associated with the severity of ultrasound-defined hepatic steatosis in a Taiwanese Han population and evaluate the variants' biological relevance through cross-ancestry meta-analysis and functional validation.
Methods
We conducted a genome-wide association study (GWAS) of 133,895 individuals with data in the Genetic Biobank of China Medical University Hospital. Hepatic steatosis severity was graded on an ordinal scale (0-3) using standardised ultrasonographic criteria. The polygenic risk score (PRS) under continuous shrinkage (CS) approach was used to construct a PRS for genetic risk stratification across severity grades.
Findings
In the GWAS, we identified 1,229 single-nucleotide polymorphisms associated with hepatic steatosis severity (P < 1 × 10⁻5), with the strongest signals discovered at the PNPLA3 locus (rs738408 and rs738409). The PRS-CS model exhibited strong discriminative performance (area under the receiver operating characteristic curve = 0.875) and a clear dose-response relationship with ultrasound-defined steatosis severity, substantially outperforming the risk model based on body mass index only. In vitro assays confirmed significantly increased lipid accumulation and intracellular triglyceride content in I148M-expressing hepatocytes under lipid-loading conditions.
Interpretation
This study delineates the genetic landscape of ultrasound-defined hepatic steatosis severity in a large Taiwanese Han population and demonstrates the robust ability of the PRS to predict disease stage.