Many liver disease-related gene variants have been identified from large-scale genome-wide association studies (GWAS) [59,60] or exome-wide association studies [61,62], which associate gene variants with diseases in patients. These studies have greatly contributed to the understanding of NAFLD genetics, pathogenesis, and variable prognosis [9]. Variants in the patatin-like phospholipase domain-containing protein 3 (PNPLA3) and transmembrane 6 superfamily member 2 (TM6SF2) genes are examples that have been suggested to contribute to the observed differences in NAFLD [63,64,65,66].

PNPLA3, or adiponutrin (ADPN), is a membrane-bound protein highly expressed in the liver and adipose tissue. It is responsible for non-specific hydrolase activity against various lipid substrates and is regulated by retinol levels [67,68]. In a 2008 genome-wide association study, Romeo and colleagues found that the change at codon 148 from isoleucine to methionine (p.I148M) (rs738409) was strongly associated with increased hepatic fat content [65]. Individuals carrying this variant had two-fold greater hepatic fat content compared to non-carriers. The PNPLA3 p.I148M variant has since been well characterised and has been proposed to be a strong modifier of NAFLD pathogenesis [69]. Studies have suggested that this could be due to the variant inducing a gain- or loss-of-function to the PNPLA3 protein [68,70,71]. However, more recently, it has been discovered that overexpressing this variant impaired the PPARγ, retinoid X receptor, and retinoic acid receptor signalling pathways, while the Jun N-terminal kinase (JNK) and activator protein-1 signalling pathways were upregulated [72]. The dysregulation of these pathways increases inflammatory activity, which could explain why patients possessing the PNPLA3 p.I148M variant are more susceptible to liver inflammation and steatohepatitis development [72]. In an even more recent study, it has been suggested that the PNPLA3 p.I148M variant could instead be a neomorphic variant, in which the protein gains an altered (novel) function that is different from normal [73]. In the study by Romeo et al., the frequency of the PNPLA3 p.I148M variant was determined to be highest in Hispanics (49%), followed by European-Americans (23%) and African-Americans (17%) [65]. Similar findings were also determined by Wagenknecht and colleagues who had used a different US cohort [66]. In a 2011 meta-analysis, the PNPLA3 p.I148M variant was also associated with more than a three-fold greater risk for fibrosis and NASH compared to non-carriers [74]. Serum ALT levels were also measured in this study and were found to be 28% higher in those with the homozygous risk variant.

TM6SF2 is another well-known regulator of NAFLD through its positive association with plasma triglyceride levels [64]. In a 2014 study, the TM6SF2 variant change at codon 167 from glutamic acid to lysine (p.E167K) (rs58542926) was found to be associated with approximately three-fold higher hepatic triglyceride in non-Hispanic European-Americans carrying the homozygous KK mutation compared to EE mutation carriers [63]. Further functional studies in this paper using Tm6sf2 short hairpin ribonucleic acid (shRNA) knockdown mice also supported this finding [63]. However, the exact function of TM6SF2 remains unknown. In a 2016 multiethnic study, homozygotes for the TM6SF2 p.E167K variant were found to be more prevalent in the Hispanic population (MAF = 0.089) than Caucasian (MAF = 0.061) or African-American (MAF = 0.033) populations [75]. This study also discovered that Caucasians and Hispanics homozygous for the variant had significantly elevated ALT and aspartate aminotransferase (AST) levels compared to heterozygous carriers. The African-American and Caucasian populations were also found to have significantly higher fat content in the liver, while Hispanics did not when comparing between homozygous and heterozygous carriers [75].

Membrane-bound O-acyltransferase domain containing 7 (MBOAT7) is involved in phospholipid remodelling via the incorporation of unsaturated fatty acids into lyso-phospholipids [76]. Carriers of the common rs641738 MBOAT7 variant (missense change about 500 bp upstream of the gene) had increased hepatic fat content, risk for fibrosis and severe liver damage compared to individuals without the variant [77]. In fact, this variant may be associated with the full spectrum of NAFLD including HCC [78]. This is attributed to the variant reducing MBOAT7 gene expression in the liver [76,77], which therefore reduces the amount of arachidonic acid being bound to lyso-phosphatidylinositol [76]. The rs641738 MBOAT7 variant is most common in the European-American population (MAF = 0.42), followed by the African-American (MAF = 0.34) and Hispanic (MAF = 0.33) populations [76]. A recent 2021 study showed that Mboat7 knockout mice fed with a high fat, methionine-low, and choline-deficient diet had significantly higher hepatic triglyceride levels than those fed with a normal diet [79]. There was also increased fibrotic development in the knockout mice with minimal impact on inflammation. These changes were also observed in biopsy-proven NAFLD individuals with the rs641738 MBOAT7 variant, which suggested that hepatic fibrosis development occurred independently from inflammation. Additional findings highlighted similarities between Mboat7 knockout mice and NAFLD individuals with the homozygous MBOAT7 variant; similar changes in lipid composition and species between these models were observed [79].

Glucokinase regulator (GCKR) is a liver glucokinase inhibitor involved in glucose metabolism [80,81]. Individuals who carried either the rs780094 (intronic variant) or the rs1260326 (change at codon 446 from proline to leucine (p.P446L)) GCKR variants displayed similar metabolic trait changes that were both associated with NAFLD and NASH [81]. Both variants are in strong linkage disequilibrium with each other [81,82]. The GCKR p.P446L variant in particular is more well known and is responsible for inducing loss-of-function of the gene [76], which interferes with the fructose-6-phosphate negative feedback inhibition mechanism on glucokinase [78]. This increases glucose uptake and metabolism, the availability of malonyl-CoA for increased DNL, and inhibition of CPT1A for reduced β-oxidation [80]. The GCKR p.P446L variant is more prevalent in the African-American population (MAF = 0.86) compared to the Hispanic (MAF = 0.67) and European-American (MAF = 0.60) populations [76].

Hydroxysteroid 17-beta dehydrogenase 13 (HSD17B13) is a hepatocyte-specific and lipid droplet-associated protein [83,84]. HSD17B13 gene expression is about six-fold higher in patients with NASH than healthy controls (p = 0.003) [84]. Similar to TM6SF2, not much is known about the exact function of the HSD17B13 gene [85]. However, members of the HSD17B family are responsible for promoting NAD(P)H/NAD(P)+-dependent oxidoreductase activity, which in turn modulates the balance between the less potent 17-ketosteroids and more potent 17β-hydroxysteroids [83]. The rs72613567 variant is one of the most common HSD17B13 variants identified in patients with NAFLD (MAF = 0.194) and has high linkage disequilibrium with the HSD17B13 rs6834314 variant [84]. The rs72613567 variant introduces an additional (duplicate) adenine nucleotide, which results in loss-of-function due to poor enzymatic activity [83,84]. Studies have shown that the rs72613567 variant is associated with decreased risk for ALD, NASH, and chronic liver disease [84,85,86]. This was due to the variant alleviating the severity of liver damage from inflammation and fibrosis, which was evident by reduced serum ALT and AST levels. These results highlight a protective role of the rs72613567 variant, which is to reduce the risk of further disease progression from steatosis [84,85,86]. In conjunction with PNPLA3, TM6SF2, or MBOAT7 variants, the HSD17B13 rs72613567 variant has been used to improve the accuracy for predicting NASH severity and advanced fibrosis [83]. There are several other HSD17B13 variants located in non-coding genomic regions that have been associated with liver fat content and liver enzyme concentrations [84].

Recent GWAS studies have highlighted numerous new gene variants that may be associated with NAFLD pathogenesis in their respective populations [60,87,88]. The genes associated with these variants include apolipoprotein E (APOE) (rs429358), glycerol-3-phosphate acyltransferase mitochondrial (GPAM or GPAT) (rs2792751), interleukin 17 receptor A (IL17RA) (rs5748926), mitochondrial amidoxime reducing component 1 (MARC1 or MTARC1) (rs2642438 and rs2642442), tribbles pseudokinase 1 (TRIB1) (rs2954021), and one locus near zinc finger protein 90-cadherin 1 (ZFP90-CDH1) (rs698718). Most of these genes are associated with the mitochondria, lipoprotein signalling or inflammation. However, only a small number of variants have been characterised through functional studies.

APOE is involved in synthesis of high-density and very-low-density lipoproteins, as it encodes for one of the protein components [89]. APOE is suspected to be involved in lipid metabolism because plasma expression is elevated in individuals with NAFLD [90]. However, the underlying mechanisms linking APOE expression and NAFLD are not well understood [87]. The 5’ adenosine monophosphate-activated protein kinase (AMPK)/mammalian target of rapamycin (mTOR) autophagy axis could explain this knowledge gap [89]. In a 2020 study, ApoE knockout mice fed a high-fat diet exhibited more severe inflammation, fibrosis, and steatosis than wild-type mice [89]. It was suspected that ApoE deficiency caused mitochondrial dysregulation, which reduced AMPK/mTOR. This, in turn, increased oxidative stress (elevation of ROS) and inflammation but reduced autophagy, which therefore prolonged liver damage.

The GPAM gene encodes a mitochondrial enzyme that is implicated in the first step of G3P-mediated triglyceride synthesis [88]. It promotes lipid accumulation by shifting lipids away from oxidation [91]. GPAM gene expression was found to be increased by about five-fold in mice with steatosis or NASH with fibrosis, developed through being fed a fast-food diet [91]. On the contrary, Gpam knockout mice were found to promote lipid oxidation as GPAM no longer competes with CPT1A for acyl-CoAs, which reduced the susceptibility of mice to oxidative stress [91]. The rs2792751 GPAM p.I43V variant is associated with increased cholesterol levels and risk for NAFLD [88].

The IL17RA gene encodes a cytokine that is involved in immunity [60,92]. Liver expression of IL17RA is higher with obesity and is associated with increased inflammation and liver damage [60,92]. In contrast, reduced protein synthesis has been shown to protect against high-fat-induced stress leading up to NASH [92].

Similar to GPAM, MARC1 also encodes a mitochondrial enzyme, although the mechanisms underlying the association between MARC1 expression with NAFLD are not well understood [88]. Missense MARC1 variants are associated with reduced cholesterol levels and are protective against NAFLD [88] possibly due to the loss-of-function variants in MARC1.

TRIB1 is suggested to be involved in regulating hepatic glycogenesis and lipogenesis. shRNA knockdown or liver-specific deletion of Trib1 in mice induced steatosis through increased glucose, cholesterol, FA, and triglyceride content [93,94].

Not much is understood about the functional relationship between the ZF90 gene and NAFLD [60]. More is known about CDH1, which encodes E-cadherin, a tumour suppressor protein involved in cell–cell adhesion [60]. Liver-specific Cdh1 knockout mice fed a high-fat diet exhibit steatosis and spontaneously develop severe periportal and periductal inflammation and fibrosis compared to wild-type mice [95].

Whole-exome sequencing studies have also identified rare variants that could potentially be functionally associated with NAFLD [96,97]. The exome is the protein-coding region of the genome, and it is estimated that up to 60% of all known disease-causing variants are accounted for by exonic amino acid substitutions [98]. Genes associated with these exome-located variants include apolipoprotein B (APOB) (change at codon 2240 from lysine to a stop (p.K2240X)) and autophagy related 7 (ATG7) (rs143545741).

Similar to APOE, APOB also encodes for a component required in very-low-density and low-density lipoproteins for cholesterol transport [99]. APOB and its rare p.K2240X variant (stop gain) have both been associated with hepatic steatosis, and potentially with cirrhosis and liver cancer, according to longitudinal and large kindred-based studies [96,99]. However, the pathogenic impact of this particular APOB variant, and what pathways and mechanisms this variant affects remains unknown.

The rs143545741 ATG7 (change at codon 426 from proline to leucine (p.P426L)) variant is hypothesised to be loss-of-function [97]. Individuals of European descent carrying this variant are at risk for severe NAFLD that is greater than seven-fold compared to the general population. The ATG7 gene is involved in promoting autophagy via suppression of the autophagosome cargo protein p62 [97]. p62 is a known positive regulator of the NF-κB transcription factor that is central in the inflammatory pathway [100]. As such, the ATG7 p.P426L variant could potentially cause unrestrained p62 expression, which may facilitate inflammation and hepatocellular ballooning [97].