PNPLA3 Is Associated With Hepatocellular Carcinoma, While a Five‐Variant Polygenic Risk Score Shows Limited Discrimination in Korean Patients With MASLD
Jaejun Lee, Jung Hoon Cha, Hee Sun Cho, Hyun Yang, Mi Young Byun, Seok Keun Cho, Seong Wook Yang, Si Hyun Bae, Seung Up Kim, Su Jong Yu, Pil Soo SungABSTRACT
Background/Aims
Genetic variants in PNPLA3, TM6SF2, MBOAT7, GCKR, and HSD17B13 have been associated with MASLD progression and hepatocellular carcinoma (HCC). We evaluated their associations with prevalent HCC and the performance of a five‐variant polygenic risk score (PRS‐5) in Korean patients with MASLD.
Methods
A total of 1082 patients with MASLD from four university‐affiliated hospitals were included, of whom 99 had HCC. Genotyping was performed for PNPLA3, TM6SF2, MBOAT7, GCKR, and HSD17B13. Associations with HCC were analyzed under additive, dominant, and recessive models. The PRS‐5 was calculated using previously reported weighted coefficients.
Results
The PNPLA3 rs738409 G allele showed a significant dose‐dependent association with HCC. HCC prevalence increased progressively across genotypes (CC 6.5%, GC 8.6%, GG 11.4%; P for trend = 0.029). Under a recessive model, GG homozygotes had higher HCC prevalence than GC/CC carriers (11.4% vs. 7.8%, p = 0.047). Notably, the association between PNPLA3 and HCC was evident only among patients with advanced fibrosis, with no significant association observed in those without advanced fibrosis, suggesting a potentially fibrosis‐dependent association. In contrast, no significant associations were observed for TM6SF2, MBOAT7, GCKR, or HSD17B13 genotypes under additive or dominant models. The PRS‐5 demonstrated limited discrimination for HCC presence (AUROC 0.566, 95% CI 0.507–0.626), which was substantially lower than that reported in Western populations.
Conclusions
PNPLA3 was the only variant associated with prevalent HCC in this Korean MASLD cohort, with effects confined to patients with advanced fibrosis. The PRS‐5 showed limited discriminatory performance, suggesting the need for population‐specific genetic risk models.