Nonalcoholic fatty liver disease (NAFLD) represents a spectrum of hepatic abnormalities in individuals who consume alcohol in amounts that are not considered harmful to liver. The gamut of NAFLD varies in severity from simple steatosis, a presumably benign state, to progressively more serious disease that includes non-alcoholic steatohepatitis (NASH), hepatic fibrosis and ultimately liver cirrhosis.1 Although prospective natural history studies of NAFLD are not currently available, it is estimated that 1-2% of patients with liver steatosis (20-30% of western population) are at risk of developing cirrhosis over 15-20 years. In contrast, among patients with NASH (2-3% of western population) the risk of developing cirrhosis has been estimated to be about 12% over 8 years.2 NAFLD has become an epidemic in the developed world and represents the liver manifestation of metabolic syndrome. Better understanding of the pathobiology of liver steatosis and its progression to NASH will have major impact on the prognosis and therapy of NAFLD patients. This is particularly important for individuals affected by NASH given the greater association of NASH than simple steatosis with cirrhosis and complications of advanced liver disease including hepatic de-compensation and development of hepatocellular carcinoma.
In the events that lead to NAFLD both environmental and genetic factors are in play. In 2008, a genome-wide association study (GWAS) of coding single nucleotide polymorphisms (SNPs) identified a non-synonymous variant (rs738409 C→G) of adiponutrin [patatin-like phospholipase domain containing 3 (PNPLA3)] conferring predisposition to NAFLD.3 The cytosine (C) to guanine (G) substitution results in an isoleucine to methionine exchange at amino acid 148 (I148M) of the 481-residue protein. The G allele variation of PNPLA3 was significantly associated with liver fat content in European-Americans, African-Americans and Hispanics independent of body mass index (BMI), diabetes status, alcohol consumption and ancestry.3 Recently, the rs738409 variant of PNPLA3 was also associated with alcoholic liver disease and alcoholic cirrhosis.4 In a small but carefully executed study by Sookoian et al., the G allele of rs738409 was associated with liver steatosis and, for the first time, NAFLD severity independent of BMI, age, sex and insulin sensitivity.5 Three large and independent studies published in HEPATOLOGY (one in April 2010 and two in the current issue) confirm the association of adiponutrin with steatosis and severity of NAFLD.6, 7, 8
PNPLA3 is a member of the patatin-like phospholipase family and is expressed in several human tissues including adipose tissue, liver, muscle, bone, skin and macrophages with the highest expression in liver.9, 10 The function of PNPLA3 in humans in vivo is largely unknown but in vitro studies with recombinant human PNPLA3 in Sf9 cells have shown it can both synthesize and hydrolyze triglycerides (TG).11 Although the TG lipolytic activity predominates in Sf9 cells, the TG synthesis activity of PNPLA3 is unique and distinct from the traditional acyl-CoA-dependent TG synthesis enzymes in that it is acyl-CoA-independent (Figure 1). Specifically, PNPLA3 acts as a transacylase, which synthesizes intracellular TG by transferring acyl groups from monoglycerides (acyl group donors) to mono- and diglycerides (acyl group acceptors).11 Although the transacylase activity of PNPLA3 has not been confirmed by other investigators,12 the expression of PNPLA3 in humans is regulated, at least in adipose tissue, in a similar manner with that of genes involved in lipogenesis. In particular, adipose PNPLA3 mRNA has been shown to increase by glucose and insulin,13 decrease with caloric restriction and become normalized upon re-feeding.14
There is no information on the regulation of PNPLA3 in human liver because of the challenge to collect serial liver biopsies. In mice, hepatic PNPLA3 mRNA decreased dramatically with fasting and increased upon re-feeding.15, 10 Incubation of cultured hepatocytes (HuH7 cells) with the fatty acids palmitate, oleate or linoleate increased PNPLA3 protein mass whereas incubation with the long-chain polyunsaturated fatty acids arachidonic or eicosapentaenoic did not alter it.10
The effects of the rs738409 G variant of PNPLA3 (I148M) have been recently investigated in mouse liver and in cultured hepatoma cells (HuH-7).12 Over-expression of the wild type human PNPLA3 did not affect liver TG content in mice. In contrast, over-expression of the I148M isoform increased liver TG by ∼3.5-fold. The catalytic activity of PNPLA3-I148M in hydrolyzing TG was abolished suggesting that the rs738409 variant promotes accumulation of intrahepatic TG by inhibiting TG hydrolysis. The current structural model of PNPLA3 that has been proposed is that the longer side chain of methionine (compared to isoleucine) extends into the catalytic site of PNPLA3 restricting access of the substrate (i.e., TG) to the catalytic residues and, thereby, inactivating PNPLA3 resulting in intrahepatic TG accumulation.12
The rs738409 variant is a common polymorphism with reported G allele frequency of 0.18-0.23, in Caucasian-American and Italian healthy controls.5-8 The study by Valenti et al., published in the April 2010 issue of HEPATOLOGY, examined the rs738409 polymorphism in 253 Italian NAFLD patients, 321 NAFLD patients from the United Kingdom (UK), 176 controls and 71 family trios (i.e., father, mother and affected child with NAFLD).6 The rs738409 GG genotype was more frequent in NAFLD patients (12%–14%) compared to control participants (2.8%) with an OR of 3.29 (95%, CI 1.8-6.9) adjusted for age, sex, and BMI. Moreover, in trios the G allele was over transmitted (68% compared to 32% of the wild-type C allele; P = 0.001) to the affected children. As shown in previous studies, patients carrying the GG genotype had significantly higher ALT levels (compared to CC and CG genotypes) in both the Italian and UK cohorts. In the combined cohort the G allele was associated with steatosis grade 2-3 (OR = 1.35, 95% CI = 1.04-1.76; P = 0.02), NASH (OR = 1.5, 95% CI = 1.12-2.04; P = 0.007) and fibrosis stage 2-4 (OR = 1.5, 95% CI = 1.09-2.12; P = 0.01) independent of age, sex, BMI, and type 2 diabetes. Of interest, the G allele was associated with NASH/fibrosis independently of steatosis.
In this issue of HEPATOLOGY, Rotman et al.7 report a confirmatory study in which six SNPs known to be associated with hepatic fat or elevated liver tests in previous GWAS3, 16 were genotyped in 894 adults and 223 children with histologically confirmed NAFLD. The G allele frequency of rs738409 in the NAFLD adult Caucasian cohort was 0.5. The G allele was associated with steatosis ≥33% (OR = 1.46, 95% CI = 1.07-2.01; p = 0.012), portal inflammation (OR = 1.57, 95% CI = 1.24-1.99; p = 2.5x10−4), lobular inflammation (OR = 1.84, 95% CI = 1.33-2.55; p = 0.002), Mallory-Denk bodies (OR = 1.60, 95% CI = 1.46-3.07; p = 0.015), and greater NAFLD activity score (p = 0.004) independent of age, sex, BMI, alcohol consumption, and type 2 diabetes. The association of the G allele with inflammation and Mallory-Denk bodies remained significant after adjustment for the degree of steatosis. The rs738409 G variant was significantly associated with higher fibrosis scores with an average score that was 0.47 stage higher in GG patients compared to CC patients independent of steatosis, inflammation and the presence of Mallory-Denk bodies. However, despite the association of the rs738409 G allele with features of advanced disease independent of steatosis, the allelic frequency of G was similar between the NASH group (49.2%) and the steatosis only group (51.8%). In these two patient groups, the G allele frequency was significantly greater than that in the control group (22.8%). Therefore, the rs738409 G allele did not differentiate NASH from steatosis alone. None of the six SNPs was associated with histological severity in the small pediatric cohort, although the rs738409 G allele was associated with a younger age at the time of biopsy independent of sex and BMI.
Finally, Speliotes et al.,8 performed another confirmatory study of 12 SNPs from 7 loci known to be associated with steatosis and increased liver tests,3, 16 using 592 cases of European ancestry from the Nonalcoholic Steatohepatitis Clinical Research Network. These genotypes were compared with data from 1405 ancestry-matched controls from the Myocardial Infarction Genetics Consortium. In the case-control analysis, among the 12 SNPs tested, only the rs738409 G allele was associated with histologic NAFLD (OR = 3.26, 95% CI = 2.11-7.21; p = 3.6 × 10−43). The same allele was associated with steatosis >5% (OR = 3.12 95% CI = 2.67-3.64; p = 1 × 10−46), lobular inflammation (OR = 3.08 95% CI = 2.64-3.57; p = 1.8 × 10−47), hepatocellular ballooning (OR = 3.21 95% CI = 2.68-3.82; p = 4.2 × 10−38), NASH (OR = 3.26 95% CI = 2.76-3.85; p = 2.1 × 10−44) and fibrosis (OR = 3.37 95% CI = 2.76-3.85; p = 2.1 × 10−44). These associations remained significant after adjusting for BMI, plasma lipids, and type 2 diabetes. In the case-only analysis, carriage of the G allele of rs738409 (frequency: 0.50) was significantly associated with having lower zone 3 centered steatosis and higher azonal and paracinal steatosis. The latter patterns of liver steatosis are more likely to be associated with ballooning, Mallory bodies and bridging fibrosis than zone 3 steatosis.17 In the case-only analysis, the G allele of rs738409 was associated with a less detrimental metabolic profile (lower BMI, weight, waist circumference, plasma TG, risk of type 2 diabetes, and increased HDL-cholesterol). This finding is in line with the observation by Kantartzis et al., that among patients with fatty liver, GG homozygotes (but not C allele carriers) were not statistically different in insulin sensitivity than control individuals without fatty liver, despite the large difference in liver fat content.18
In summary, the results of the 3 cross-sectional studies published in Hepatology and previous work uniformly show that the rs738409 G allele is associated with liver steatosis independently of traditional risk factors (i.e., adiposity, age, sex) and insulin sensitivity. Therefore, the rs738409 G allele predicts steatosis above and beyond well-known metabolic risk factors, although, only a small percent of the variation in liver steatosis is accounted for by this genetic information (∼5%).5 More importantly, among NAFLD patients, the rs738409 G allele appears to predict severity of NAFLD independent of steatosis. This is despite the observation that the G allele is also associated with a metabolic profile and insulin sensitivity that is similar or even more favorable than that of the C allele.18, 8 Because the contribution of G allele in hepatic steatosis is likely small (see above)5 its use in clinical practice is currently limited. However, it is anticipated that future algorithms that could predict individuals at risk for liver steatosis would probably include this and other yet un-identified alleles in combination with non-genetic factors.
As obesity and insulin resistance increase in the western societies, NAFLD is expected to increase, calling for better ways of identifying those at high risk of progression. Cross-sectional studies can not establish cause-effect or temporal relations among variables, therefore results from this type of studies should always be interpreted with caution. Furthermore, when assessing predictors of disease progression (genetic or environmental), a large number of NAFLD patients with a wide spectrum of disease severity are necessary to avoid the problem of co-linearity between histologic parameters. For instance, if the majority of study patients have an advanced form of the disease, it becomes difficult to appropriately assess the effects on individual parameters of liver damage. Finally, the fact that liver steatosis changes throughout the disease progression further complicates the interpretation of cross sectional studies that look for predictors of NAFLD severity using steatosis as an independent predictor. To this end, longitudinal studies are necessary to adequately delineate the key components and predictors of NAFLD progression.
Proteins like PNPLA3 open up new horizons in our understanding of the genetic architecture of NAFLD not to mention its possible contribution to alcoholic liver disease.4 While functional studies of the rs738409 C→G polymorphism are warranted to elucidate its contribution to the pathogenesis of NAFLD, several other genetic variants still remain to be discovered. Now that several large cohorts of NAFLD patients have been established linked to biospecimens (i.e., genomic DNA, plasma/serum specimens, liver biopsies) there is opportunity for large GWAS studies along with replication to study the non-coding variants across the human genome. Such data will provide opportunities to examine not only individual variations but also gene pathways including gene-gene interaction (i.e., epistasis). It will be also important to further enrich the existing NAFLD cohorts with environmental exposures of the participants for future gene-environment interaction studies. Finally, novel imaging techniques coupled with innovative biochemical tests of liver inflammation and fibrosis have to be discovered for a less invasive and more comprehensive assessment of NAFLD, which will bring us beyond the current limitations imposed by liver biopsy. Despite the inherent challenges of these approaches/efforts, collaboration in a multidisciplinary setting and inventive thinking are necessary to move forward the field of NAFLD. Identification of the molecular mechanisms that bring about liver steatosis and progression to NASH and cirrhosis will make possible the development of interventions aimed at preventing NAFLD altogether.