Skip to content

Not yet independently audited

This page was written from verified sources but has not yet passed the independent citation audit. Treat its claims as provisional.

Genetics

TL;DR — MASLD is unusually tractable genetically for a metabolic disease. Twin studies put heritability of hepatic steatosis at 0.52 (95% CI 0.31–0.73) and of hepatic fibrosis at 0.50 (0.28–0.72), with a shared genetic effect between the two of 0.756 (0.716–1, p<0.0001) (Loomba 2015, PMID 26299412; Cui 2016, PMID 27315352). The dominant common variant is PNPLA3 rs738409 (I148M), identified in 2008 in a multi-ancestry scan: homozygotes had more than twice the hepatic fat of non-carriers, and the allele was most frequent in Hispanic Americans — the group with the highest NAFLD prevalence — and rarest in African Americans (Romeo 2008, PMID 18820647). The counterweight is HSD17B13 rs72613567:TA, a protein-truncating splice variant that reduces nonalcoholic cirrhosis risk by 26% in heterozygotes and 49% in homozygotes, protects against steatohepatitis but not steatosis, and mitigates PNPLA3-associated injury (Abul-Husn 2018, PMID 29562163). Both have become drug targets: an antisense oligonucleotide against PNPLA3 achieved 89% hepatic mRNA knockdown with placebo-corrected liver-fat reductions of −7.6% and −12.2% (Armisen 2025, PMID 39798707), and an siRNA against HSD17B13 achieved a median 78% mRNA reduction at six months (Sanyal 2025, PMID 40581300). The largest GWAS meta-analysis to date implicates 17 loci and finds the top 10% and top 1% of genetic risk carry 2.5-fold to 6-fold increased risk of NAFLD, cirrhosis and HCC, with PheWAS suggesting at least seven subtypes (Chen 2023, PMID 37709864). None of this is in any clinical risk-stratification pathway.

Heritability

Trait Heritability estimate (95% CI) Method Source
Hepatic steatosis (MRI-PDFF) 0.52 (0.31–0.73), p<1.1×10⁻¹¹ 60 twin pairs (42 MZ, 18 DZ), adjusted for age, sex, ethnicity Loomba 2015, PMID 26299412
Hepatic fibrosis (MRE stiffness) 0.50 (0.28–0.72), p<6.1×10⁻¹¹ same cohort PMID 26299412
Shared genetic effect, steatosis ↔ fibrosis 0.756 (0.716–1), p<0.0001 65 twin pairs, bivariate ACE model Cui 2016, PMID 27315352

In the twin cohort, steatosis correlated between monozygotic twins (r²=0.70, p<0.0001) but not dizygotic (r²=0.36, p=0.2); fibrosis correlated between MZ twins (r²=0.48, p<0.002) but not DZ (r²=0.12, p=0.7). Cui additionally found significant shared gene effects between hepatic steatosis and blood pressure, triglycerides, glucose, HOMA-IR, insulin, HbA1c and low HDL, and between fibrosis and the same metabolic traits minus blood pressure — evidence that the genetic architecture of MASLD is partly the genetic architecture of metabolic syndrome.

The core variants

Gene / variant Direction Effect Key evidence
PNPLA3 rs738409 (p.I148M, G allele) risk ↑ >2-fold higher hepatic fat in GG vs non-carriers (p=5.9×10⁻¹⁰); associated with hepatic inflammation (p=3.7×10⁻⁴) Romeo 2008, PMID 18820647
PNPLA3 rs6006460 (p.S453I, T allele) risk ↓ lower hepatic fat in African Americans PMID 18820647
TM6SF2 rs58542926 (p.E167K, T allele) risk ↑ (liver), risk ↓ (CVD) genome-wide significant for NAFLD and for advanced fibrosis Anstee 2020, PMID 32298765; Chen 2023, PMID 37709864
HSD17B13 rs72613567:TA risk ↓ nonalcoholic liver disease −17% (het) / −30% (hom); nonalcoholic cirrhosis −26% / −49%; alcoholic cirrhosis −42% / −73% Abul-Husn 2018, PMID 29562163
GCKR / C2ORF16 risk ↑ genome-wide significant in histologically characterised cohort PMID 32298765
MBOAT7 / TMC4 risk ↑ validated GWAS locus PMID 37709864
MARC1, MTTP, APOE, TRIB1, GPAM, PTPRD, ADH1B mixed validated GWAS loci PMID 37709864
TOR1B, FTO, COBLL1/GRB14, INSR, SREBF1, PNPLA2 risk ↑ newly identified loci PMID 37709864
PYGO1 rs62021874 steatosis modifier p=8.2×10⁻⁸ for steatosis; implicates Wnt signalling PMID 32298765
LEPR (chr1) risk ↑ genome-wide significant for NASH subgroup PMID 32298765

The histology-anchored GWAS matters disproportionately. Most GWAS use radiological hepatic triglyceride and therefore cannot address steatohepatitis or fibrosis. Anstee's study of 1,483 European histologically characterised NAFLD cases against 17,781 matched controls (replication: 559 cases, 945 controls) found genome-wide significance at four loci — GCKR/C2ORF16, HSD17B13, TM6SF2, PNPLA3 — and in case-only quantitative-trait analysis PNPLA3 reached genome-wide significance for steatosis, fibrosis and NAFLD activity score. Restricting to advanced fibrosis (≥F3), the chr2, chr19 and chr22 signals held (PMID 32298765).

The multi-ancestry meta-analysis combined imaging-measured NAFLD (n=66,814) with diagnostic-code NAFLD (3,584 cases vs 621,081 controls) across ancestries. Implicated genes point to mitochondrial function, cholesterol handling and de novo lipogenesis as causal contributors — matching the mechanisms described in pathogenesis — and PheWAS suggested at least seven NAFLD subtypes (Chen 2023, PMID 37709864).

PNPLA3: from variant to prognosis

The variant does not just raise liver fat; it changes disease trajectory and does so unequally across groups.

  • Incident cirrhosis. In the Michigan Genomics Initiative (n=7,893) and UK Biobank (n=46,880), PNPLA3 GG genotype, diabetes, obesity and ALT ≥2× upper limit of normal each raised cirrhosis incidence, with additive effects. Among patients in the indeterminate FIB-4 band (1.3–2.67), those with both diabetes and GG genotype had cirrhosis incidence comparable to patients with high-risk FIB-4 >2.67, and 2.9–4.8 times that of patients with diabetes but CC/CG genotypes. FIB-4 <1.3 remained low-risk even with high-risk genotype (Chen 2023, PMID 36758837).
  • Fibrosis trajectory with age. Among 570 prospectively recruited adults with MRE and genotyping, a genetic risk score (PNPLA3 risk alleles minus HSD17B13 protective alleles) split trajectories: in the high-GRS group liver stiffness rose per decade (β=0.28 kPa, 95% CI 0.12–0.44, p=0.001) with no such rise in the low-GRS group. PNPLA3 alone predicted higher stiffness (C/G β=0.32 kPa, p=0.034; G/G β=0.87 kPa, 95% CI 0.52–1.22, p<0.0001), with G/G diverging significantly by age 44 and confirmed in a Latin American validation cohort (Díaz 2025, PMID 40334848).
  • Histological progression and regression. In 671 NASH CRN participants with serial biopsies, the G allele raised fibrosis progression (aHR 1.31, 1.05–1.64), the HSD17B13 A allele lowered it (aHR 0.69, 0.51–0.92) and raised both MASH resolution (1.58, 1.13–2.22) and fibrosis regression (1.42, 1.09–1.85); TM6SF2 had no effect on histological change (Vilar-Gomez 2026, PMID 40998180).
  • Sex interaction. There is a reported statistical interaction between female sex and p.I148M for steatosis and fibrosis (p<10⁻¹⁰) and for advanced fibrosis/HCC (p=0.034). Hepatic PNPLA3 expression was higher in women than men with obesity (p=0.007), correlated with oestrogen in mice, and was induced by ER-α agonists in human hepatocytes and organoids; an ER-α-binding site within a PNPLA3 enhancer was identified and CRISPR editing showed it drives p.I148M upregulation with lipid-droplet accumulation and fibrogenesis in multilineage spheroids (Cherubini 2023, PMID 37749332). This supplies a candidate mechanism for sex-dependent risk; it does not by itself establish why an individual woman progresses after menopause.
  • Ancestry. The 2008 discovery explicitly framed the variant as explaining ancestry-related differences in hepatic fat: most common in Hispanics, the highest-risk group, with a separate protective allele more common in African Americans, the lowest-risk group (PMID 18820647).

HSD17B13: the protective allele, quantified

Outcome Heterozygote risk reduction (95% CI) Homozygote risk reduction (95% CI)
Alcoholic liver disease 42% (20–58) 53% (3–77)
Nonalcoholic liver disease 17% (8–25) 30% (13–43)
Alcoholic cirrhosis 42% (14–61) 73% (15–91)
Nonalcoholic cirrhosis 26% (7–40) 49% (15–69)

Source: Abul-Husn 2018, PMID 29562163 — discovered in 46,544 DiscovEHR exomes via ALT/AST association (ALT p=4.2×10⁻¹², AST p=6.2×10⁻¹⁰), replicated in 12,527, validated for clinical diagnoses in 37,173 and against histopathology in 2,391 liver samples. Two features make it a strong drug target: the variant protects against steatohepatitis but not steatosis — that is, it blocks the transition that matters — and it produces an unstable, truncated protein with reduced enzymatic activity, so loss of function is the protective direction, which is straightforward to mimic pharmacologically. It also mitigated injury associated with the PNPLA3 risk allele.

Rare protein-coding variants: a second, independent map

Common-variant GWAS and rare-variant exome burden testing find largely different genes, and the rare-variant map is where most of the recent drug targets have come from. Exome sequencing across 542,904 people with aminotransferase data, 24,944 patients with liver disease of any type and 490,636 controls found rare coding variants in APOB, ABCB4, SLC30A10 and TM6SF2 raising aminotransferases and liver-disease risk, and — the headline — a protective burden in CIDEB, which encodes a hepatic lipid-droplet structural protein (Verweij 2022, PMID 35939579):

CIDEB rare pLoF + missense burden (combined carrier frequency 0.7%) Effect per allele
ALT −1.24 U/L (95% CI −1.66 to −0.83), p=4.8×10⁻⁹
Liver disease of any cause OR 0.67 (0.57–0.79), p=9.9×10⁻⁷
Cirrhosis of any cause OR 0.50 (0.36–0.70)
NAFLD activity score (3,599 bariatric-surgery patients) −0.98 score units (−1.54 to −0.41)

CIDEB siRNA knockdown in oleate-challenged human hepatoma lines prevented large lipid-droplet formation, giving the variant a direct mechanistic reading. The pattern is the same one that made HSD17B13 attractive: protection is conferred by loss of function, across aetiologies and across severity, which is the easy direction to mimic pharmacologically.

Two further loss-of-function protective genes came from the Icelandic multiomic study, which combined 9,491 clinical NAFL cases with proton-density fat fraction from 36,116 liver MRIs, identifying 18 variants for NAFL and 4 for cirrhosis and rare protective predicted-loss-of-function variants in MTARC1 and GPAM (Sveinbjornsson 2022, PMID 36280732). It also integrated 4,907 plasma proteins in 35,559 Icelanders and 1,459 in 47,151 UK Biobank participants, and showed that proteomics can discriminate NAFL from cirrhosis — evidence that a protein-based non-invasive test may be achievable (noninvasive assessment). GPAM was independently corroborated: in a large multi-ancestry cirrhosis GWAS (12 cohorts; 18,265 cirrhosis cases, 1,782,047 controls, ~1M individuals with liver function tests, validated in 21,689 cases and 617,729 controls), rare coding variants in GPAM associated with lower ALT, supporting GPAM inhibition as a target (Ghouse 2024, PMID 38632349). That study also reported that PNPLA3 p.I148M interacts with alcohol intake, obesity and diabetes on cirrhosis and HCC risk; the statistical interaction does not by itself establish a simple multiplicative biological relationship. The study also derived a polygenic score associated with progression from cirrhosis to HCC.

A somatic genetic contribution nobody was looking for

Genetic risk in this disease is not entirely germline. Across 214,563 individuals with whole-exome sequencing in four cohorts (Framingham, ARIC, UK Biobank, Mass General Brigham Biobank), clonal haematopoiesis of indeterminate potential (CHIP) was associated with prevalent and incident chronic liver disease (OR 2.01, 95% CI 1.46–2.79, p<0.001) and with MRI-detectable liver inflammation and fibrosis (OR 1.74, 1.16–2.60, p=0.007); Mendelian randomisation on genetic predisposition to CHIP gave OR 2.37 (1.57–3.60), p<0.001 (Wong 2023, PMID 37046084). In a dietary NASH model, mice transplanted with Tet2-deficient haematopoietic cells developed more severe inflammation and fibrosis, mediated by the NLRP3 inflammasome and downstream cytokines in Tet2-deficient macrophages. This makes an age-acquired bone-marrow mutation a liver-fibrosis risk factor operating through the innate immune compartment described in pathogenesis, and it is not captured by any germline risk score.

Genotype as drug target

Agent Modality Target Result Source
AZD2693 antisense oligonucleotide PNPLA3 mRNA 89% least-square-mean hepatic PNPLA3 mRNA knockdown; placebo-corrected liver fat change at week 12 −7.6% (25 mg) and −12.2% (50 mg); half-life 14–33 days; no discontinuations or treatment-related SAEs; dose-dependent rise in serum PUFA-containing triglycerides; falls in hsCRP and IL-6 Armisen 2025, PMID 39798707 (NCT04142424, NCT04483947)
PNPLA3 siRNA siRNA PNPLA3 phase 1 in I148M homozygotes with MAFLD Fabbrini 2024, PMID 39083780
Rapirosiran (ALN-HSD) GalNAc-siRNA HSD17B13 mRNA paper reports median 78% liver mRNA reduction at 6 months (400 mg), Part A n=58 healthy and Part B n=46 with MASH, with biopsy-confirmed target engagement and no drug-induced liver injury. The linked live registry record instead says terminated; actual enrollment 6. The source discrepancy is unresolved and neither number should be substituted for the other Sanyal 2025, PMID 40581300; NCT04565717

These are the first genetically validated targets in this disease to reach patients, and the strategic logic is set out explicitly in a review of genetically validated targets in fatty liver disease (Lindén 2023, PMID 37207913). All are phase 1 with liver-fat or mRNA endpoints; none has yet reported histological or clinical outcomes. AZD2693 has advanced to the phase 2b FORTUNA study (PMID 39798707). See clinical trials landscape.

Polygenic risk scores

Score Composition Finding Source
PRS-HFC PNPLA3, TM6SF2, GCKR, MBOAT7 predicted HCC more robustly than any single variant (p<10⁻¹³) in a NAFLD cohort (n=2,566, 226 HCC) Bianco 2021, PMID 33248170
PRS-5 PRS-HFC adjusted for HSD17B13 high cut-offs (≥0.532 / ≥0.495) detected HCC with ~90% specificity but limited sensitivity in UK Biobank (n=364,048, 202 HCC) PMID 33248170
16-variant steatosis PRS GWAS-derived dietary effects on liver fat 1.4–3.0-fold larger in the top vs bottom quartile Chen 2024, PMID 38582304
3-SNP unweighted score PNPLA3, HSD17B13, TM6SF2 histological fibrosis progression aHR 1.37 (1.18–1.60); non-risk-allele sum associated with MASH resolution aHR 1.21 (1.02–1.43) Vilar-Gomez 2026, PMID 40998180
17-locus genetic risk GWAS meta-analysis top 10% and top 1% carry 2.5–6-fold increased risk of NAFLD, cirrhosis and HCC Chen 2023, PMID 37709864

The Bianco analysis is the closest thing to a causal argument: the impact of genetic risk variants on HCC was proportional to their predisposition to fatty liver (p=0.002), the association was mainly mediated through severe fibrosis but remained independent of fibrosis in clinically relevant subgroups, and it held in individuals without cirrhosis (p<0.05) in the NAFLD cohort and independently of classical risk factors and cirrhosis in UK Biobank (p<10⁻⁷). That last point connects directly to the pre-cirrhotic HCC problem — see MASLD-related hepatocellular carcinoma.

Gene–environment interaction

Genetic risk is not fixed in its consequences. In 21,619 UK Biobank participants with 24-hour dietary recall and genotyping, Mediterranean diet and intake of fruit/vegetables/legumes and fish associated with lower liver fat content, red and processed meat and all genetic predictors with higher. Critically, all genetic predictors interacted with Mediterranean diet and fruit/vegetable/legume intake: dietary effects on liver fat were up to 3.8-fold larger in PNPLA3 GG than CC individuals and 1.4–3.0-fold larger in the top versus bottom PRS quartile, with stronger interactions in participants with overweight. The steatosis PRS also interacted with diet to affect cT1 (inflammation/fibrosis), and most dietary and genetic predictors associated with liver-related events or mortality by age 70 (Chen 2024, PMID 38582304). The practical inference — that dietary intervention may be more impactful in the genetically high-risk, not less — is the opposite of therapeutic fatalism, and has not been tested prospectively.

Genetic effects are also modified by non-genetic factors in the other direction: in the NASH CRN serial-biopsy analysis, age, sex, BMI and T2D significantly modified the effects of single variants and of the PRS (PMID 40998180).

Open questions

  • Should genotype enter risk stratification now? The evidence that it would help is substantial: genotype reclassifies indeterminate-FIB-4 patients into a high-risk band (PMID 36758837), predicts age-related fibrosis trajectory (PMID 40334848), improves HCC prediction over single variants and beyond cirrhosis status (PMID 33248170), and predicts histological progression and regression (PMID 40998180). No prospective study has tested whether a genotype-augmented pathway improves discrimination, reclassification or outcomes over FIB-4 plus elastography. Query run 2026-09-02: (NAFLD OR NASH OR MASLD OR MASH OR "fatty liver") AND (TM6SF2 OR HSD17B13 OR MBOAT7 OR GCKR OR "GWAS" OR "polygenic risk score") — 1,023 records; the evidence remains observational.
  • Do the genetically validated drugs change histology? Both PNPLA3 and HSD17B13 programmes have shown target engagement and, for AZD2693, liver-fat reduction (PMIDs: 39798707, 40581300), but liver fat is not the outcome-relevant variable (natural history). Phase 2b results are pending.
  • Do the loss-of-function protective genes translate? CIDEB (OR 0.50 for cirrhosis; PMID 35939579), MTARC1 and GPAM (PMIDs: 36280732, 38632349) all show the HSD17B13 pattern — protection by loss of function, across aetiologies — but none has a published clinical-stage programme with human liver endpoints retrieved as of 2026-09-02.
  • Is CHIP a modifiable liver risk factor? CHIP roughly doubles chronic liver disease risk with supporting Mendelian randomisation and a Tet2/NLRP3 mouse mechanism (PMID 37046084). Whether NLRP3 inhibition in CHIP carriers alters hepatic fibrosis has not been tested, and CHIP status is not measured in any MASLD cohort with paired histology.
  • How does the PNPLA3 effect vary with metabolic exposures? A large cirrhosis GWAS reports statistical interactions with alcohol intake, obesity and diabetes (PMID 38632349), while the incident-cirrhosis analysis models genotype, diabetes, obesity and ALT as additive predictors (PMID 36758837). Those results use different designs and scales and are not necessarily contradictory; clinically useful effect modification still needs prospective quantification.
  • What are the seven NAFLD subtypes? PheWAS on 17 loci suggested at least seven subtypes (PMID 37709864). They have not been characterised clinically, and no study has tested whether they respond differently to any therapy.
  • Why is TM6SF2 hepatoprotective for the heart and hepatotoxic for the liver? The variant raises liver fat while lowering circulating lipids and cardiovascular risk, and it does not influence histological progression in serial biopsies (PMID 40998180) despite genome-wide significance for advanced fibrosis cross-sectionally (PMID 32298765). The discordance is unexplained and matters for any drug that mimics the variant.
  • Does the PNPLA3–oestrogen interaction predict menopausal risk prospectively? The mechanism is well worked out in cells and organoids (PMID 37749332); no cohort has tested whether genotype plus menopausal status identifies the women who progress rapidly.
  • Are the ancestry-linked allele frequencies driving the observed prevalence differences, or is the association confounded? Romeo's 2008 framing has been repeated for 18 years (PMID 18820647) but the fraction of between-population prevalence variance explained by allele frequency versus environment has not been formally decomposed.

References

  1. Romeo S, Kozlitina J, Xing C, et al. Genetic variation in PNPLA3 confers susceptibility to nonalcoholic fatty liver disease. Nat Genet. 2008;40(12):1461-5. PMID 18820647
  2. Loomba R, Schork N, Chen CH, et al. Heritability of Hepatic Fibrosis and Steatosis Based on a Prospective Twin Study. Gastroenterology. 2015;149(7):1784-93. PMID 26299412
  3. Cui J, Chen CH, Lo MT, et al. Shared genetic effects between hepatic steatosis and fibrosis: A prospective twin study. Hepatology. 2016;64(5):1547-1558. PMID 27315352
  4. Abul-Husn NS, Cheng X, Li AH, et al. A Protein-Truncating HSD17B13 Variant and Protection from Chronic Liver Disease. N Engl J Med. 2018;378(12):1096-1106. PMID 29562163
  5. Anstee QM, Darlay R, Cockell S, et al. Genome-wide association study of non-alcoholic fatty liver and steatohepatitis in a histologically characterised cohort. J Hepatol. 2020;73(3):505-515. PMID 32298765
  6. Chen Y, Du X, Kuppa A, et al. Genome-wide association meta-analysis identifies 17 loci associated with nonalcoholic fatty liver disease. Nat Genet. 2023;55(10):1640-1650. PMID 37709864
  7. Chen VL, Oliveri A, Miller MJ, et al. PNPLA3 Genotype and Diabetes Identify Patients With Nonalcoholic Fatty Liver Disease at High Risk of Incident Cirrhosis. Gastroenterology. 2023;164(6):966-977.e17. PMID 36758837
  8. Díaz LA, Alazawi W, Agrawal S, et al. High inherited risk predicts age-associated increases in fibrosis in patients with MASLD. J Hepatol. 2025;83(4):849-859. PMID 40334848
  9. Vilar-Gomez E, Yates KP, Kleiner DE, et al. Genetic and non-genetic drivers of histological progression and regression in MASLD. J Hepatol. 2026;84(3):502-516. PMID 40998180
  10. Cherubini A, Ostadreza M, Jamialahmadi O, et al. Interaction between estrogen receptor-α and PNPLA3 p.I148M variant drives fatty liver disease susceptibility in women. Nat Med. 2023;29(10):2643-2655. PMID 37749332
  11. Bianco C, Jamialahmadi O, Pelusi S, et al. Non-invasive stratification of hepatocellular carcinoma risk in non-alcoholic fatty liver using polygenic risk scores. J Hepatol. 2021;74(4):775-782. PMID 33248170
  12. Chen VL, Du X, Oliveri A, et al. Genetic risk accentuates dietary effects on hepatic steatosis, inflammation and fibrosis in a population-based cohort. J Hepatol. 2024;81(3):379-388. PMID 38582304
  13. Armisen J, Rauschecker M, Sarv J, et al. AZD2693, a PNPLA3 antisense oligonucleotide, for the treatment of MASH in 148M homozygous participants: Two randomized phase I trials. J Hepatol. 2025;83(1):31-42. PMID 39798707
  14. Sanyal AJ, Taubel J, Badri P, et al. Phase I randomized double-blind study of an RNA interference therapeutic targeting HSD17B13 for metabolic dysfunction-associated steatohepatitis. J Hepatol. 2025;83(4):838-848. PMID 40581300
  15. Fabbrini E, Rady B, Koshkina A, et al. Phase 1 Trials of PNPLA3 siRNA in I148M Homozygous Patients with MAFLD. N Engl J Med. 2024;391(5):475-476. PMID 39083780
  16. Lindén D, Romeo S. Therapeutic opportunities for the treatment of NASH with genetically validated targets. J Hepatol. 2023;79(4):1056-1064. PMID 37207913
  17. Verweij N, Haas ME, Nielsen JB, et al. Germline Mutations in CIDEB and Protection against Liver Disease. N Engl J Med. 2022;387(4):332-344. PMID 35939579
  18. Sveinbjornsson G, Ulfarsson MO, Thorolfsdottir RB, et al. Multiomics study of nonalcoholic fatty liver disease. Nat Genet. 2022;54(11):1652-1663. PMID 36280732
  19. Ghouse J, Sveinbjörnsson G, Vujkovic M, et al. Integrative common and rare variant analyses provide insights into the genetic architecture of liver cirrhosis. Nat Genet. 2024;56(5):827-837. PMID 38632349
  20. Wong WJ, Emdin C, Bick AG, et al. Clonal haematopoiesis and risk of chronic liver disease. Nature. 2023;616(7958):747-754. PMID 37046084