PNPLA3

The patatin-like phospholipase domain-containing protein 3 gene (PNPLA3), also known as adiponutrin, encodes a lipase enzyme expressed in adipocytes.9 The gene was first identified as a candidate for NAFLD susceptibility in a GWAS by Romeo et al,10 which found a significant association with the rs738409 polymorphism (not included in this review as fatty liver was assessed using 1H-MRI). The rs738409 variant is thought to eliminate the lipase activity of the enzyme.11

A number of candidate gene studies have replicated the association of PNPLA3 polymorphisms in different populations (figure 2). The effects of rs738409 on histological characteristics, disease severity and NASH diagnosis have also been investigated.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Studies evaluating the association between non-alcoholic fatty liver disease (NAFLD) and rs738409 of PNPLA3. *Combined Malaysian population including Chinese, Indian and Malay ethnic subgroups.

Sookoian et al12 performed a case–control study evaluating rs738409 polymorphism on an Argentinian population of 172 patients with NAFLD, compared with 94 controls. Discrete trait analysis showed the G allele to be associated with NAFLD compared with controls, and this relationship persisted with multivariate logistic regression, independently of age, sex, body mass index (BMI) and insulin status (HOMA-I). Of the patients with NAFLD, 103 were biopsied, 40 had SS and 63 had NASH. The G allele variant was found to be significantly associated with disease severity (GG>CG>CC) and with NASH compared with SS, with reported OR 1.88 per G allele (95% CI 1.03 to 3.43). Increased steatosis was observed with the homozygous variant (GG) compared with the heterozygous variant, which was subsequently more severe than observed for the wild type genotype.

Hotta et al13 evaluated the effects of rs738409 on a Japanese cohort of 253 patients with NAFLD and 578 healthy population controls. Of the patients with NAFLD, 189 were diagnosed with NASH and 64 with SS. The study demonstrated a significant dose-dependent relationship of SNP frequency between the NAFLD and control groups, and between patients with NASH and SS. Histologically, the variant allele had an additive effect with increased fibrosis stage, but no association was found with steatosis.

Zain et al14 evaluated the rs738409 polymorphism in a mixed ethnicity Malaysian group of 144 patients with NAFLD and 198 healthy population controls, with subgroups of Malay, Chinese and Indian ancestry. Results indicated that NAFLD was significantly associated with the risk G allele with an overall OR 2.34 which remained significant after age and gender were controlled for, and the association was different between the ethnic subgroups as demonstrated in figure 2. The NAFLD group was histologically subdivided into patients with SS, NASH with fibrosis score <2, and NASH with fibrosis score of 2+. When NASH was compared with SS, the variant G allele was found to be associated with NASH (OR 2.64). Histological characteristics of steatosis, lobular inflammation, and hepatocellular ballooning and fibrosis grade were further analysed; the GG genotype was significantly associated with fibrosis stage but no other features. Patients with at least one G allele were found to be more likely to have a fibrosis score of ≥2 (OR 1.95), suggesting a relationship with disease severity.

Valenti et al15 investigated the association of the PNPLA3 I148M variant with histological NAFLD in a European population, comprised of 321 UK and 253 Italian patients. The variant G allele was found to be significantly more prevalent in the Italian patients with NAFLD compared with 179 healthy, population-matched controls, and participants with homozygous variant genotype had an increased risk of NAFLD compared with wild type. In the combined population, the variant G allele was found to be an independent predictor for grade 2–3 steatosis, and had a dose-dependent effect on fibrosis stage >1, independently of steatosis. The variant allele and GC/GG genotypes were also associated with the presence of NASH in the combined and UK cohorts, independently of BMI, steatosis and diabetes. The study also investigated allele transmission in 71 Italian family trios, observing that the variant allele was preferentially transmitted compared with the wild type.

In the same year, Valenti et al16 studied the rs738409 polymorphism with histological severity of NAFLD in an Italian paediatric cohort (n=149). The study found that severity of steatosis was associated with the variant GG genotype (OR 18.86), independently of age, body mass and metabolic syndrome. The homozygous variant was also associated with lobular inflammation, ballooning and perivenular fibrosis, but not periportal fibrosis.

Verrijken et al17 investigated PNPLA3 rs378409 on a cohort of 470 overweight and obese Caucasian patients, of whom 287 underwent liver biopsy. Histological characteristics were evaluated in the biopsied subgroup, with heterozygotes and variant homozygotes found to have a risk of increased steatosis, lobular inflammation, severity of inflammation and ballooning, with sex-adjusted and age-adjusted OR of 7.41 (CI 3.235 to 16.697), 1.86 (CI 1.046 to 3.321), 3.5 (CI 1.712 to 7.150) and 2.78 (CI 1.514 to 5.119), respectively. Fibrosis was not found to be associated.

Guichelaar et al18 also evaluated the association of PNPLA3 rs378409 with histological severity on obese patients. A cohort of 144 predominantly female obese patients underwent biopsy; 12 had normal histology, 60 had SS and 72 were found to have NASH. The rs378409 G allele was found to be associated with NASH when compared with the combined normal and patients with SS, with multivariate OR 2.5 (CI 1.1 to 5.3).

Rotman et al19 evaluated six polymorphisms including rs738409 in 894 adult and 223 paediatric patients within the US NASH CRN cohort. The adult cohort was compared with a population control group of 336 Caucasian men aged 50 years and over. Results showed three variants: rs738409 and rs2281135 in PNPLA3 and rs2143571 in nearby SAMM50 were significantly associated with NASH compared with controls. The remaining three variants investigated: rs11597390 (CPN1), rs11597086 and rs11591741 (CHUK) were not significantly associated with NAFLD. These were associated with increased fibrosis, although whether this specifically relates to fibrosis in NASH requires further investigation.

Disease severity was further investigated, with univariate analysis showing a significant association between rs738409 allele frequency and features of steatosis and both portal and lobular inflammation. Variant allele frequency was not significantly associated with histological fibrosis or cellular injury but was associated with more severe disease in a dominant pattern. This relationship persisted after factors of age, sex, BMI, diabetes and alcohol intake were controlled for. Patients with the GG and GC genotypes were more likely to have a moderate-to-severe steatosis score (≥2; OR 1.46), Mallory-Denk bodies (OR 1.55), lobular inflammation (OR 1.84) and portal inflammation (additive OR 1.57). Multivariate analysis showed a highly significant association of the G allele with fibrosis in an additive manner, which remained after adjustment for histological parameters of steatosis, inflammation and ballooning, suggesting an independent variable. Fibrotic bridging was associated with an adjusted OR of 1.50 for each G allele. The variant allele was also associated with an increased NAS in a dominant pattern on univariate and multivariate analyses. Polymorphisms rs2281135 and rs2143571 on chromosome 22 had similar associations with NAS. Furthermore, 438 adult patients with definite histological NASH were compared with 82 with SS and the control cohort (n=336), but the rs738409 was not significantly different between SS and NASH groups. In contrast to results found by Valenti et al,16 the rs738409 SNP was not associated with any histological parameters in paediatric patients.

Speliotes et al20 studied 12 polymorphisms at seven loci in patients in the US NASH CRN cohort. A total of 592 patients of European ancestry were compared with a control group comprised of 1405 US and European patients enrolled in the Myocardial Infarctions Genetics Consortium (MIGen) study. The MIGen study used a control group to limit the influence of cardiovascular disease status. The rs738409 was found to be significantly associated with both NAFLD and overall histological components (steatosis >5%, lobular inflammation, hepatocyte ballooning, NASH diagnosis and fibrosis) compared with control. However, the histological features are highly inter-related within the NASH CRN cohort, so the case–control associations may not be reflective. PNPLA3 rs2294918 and rs2281135 SNPs were also assessed, but no association was observed after controlling for the I148M variant. One limitation to this study is that the MIGen control group was not assessed for the presence of liver disease. Case-only analysis was performed within the NASH CRN cohort, and showed that rs738409 was associated with increased lobular inflammation ≥2, but not hepatocyte ballooning or NASH diagnosis. When steatosis distribution was evaluated, patients with the variant allele were less likely to have perivenular centred steatosis. The authors hypothesise that this may be related to diffuse fat deposition and progression to advanced disease.

Petta et al21 evaluated the influence of PNPLA3 rs738409 polymorphism on NAFLD severity in a Sicilian population. Of 160 patients with NAFLD, 99 had NASH, 15 did not have NASH and 46 were described as indeterminate. The study found the PNPLA3 rs738409 variant was associated with moderate-severe steatosis (grade 2–3) by univariate analysis, but the relationship did not persist after Bonferroni correction. Multivariate logistic regression showed the homozygous variant rs738409 genotype was significantly associated with a higher score when using a recessive model.

Corbin et al22 investigated 21 candidate genes on 446 patients in the Duke University Health Systems NAFLD Clinical Database and Biorepository. The study found that PNPLA3 was significantly associated with increased histological steatosis only, with an additive effect. The study did not observe any associations with ADIPOQ, CHDH, MTHFD1, PEMT, PPARG and STAT3 genes; further information regarding these genes is given in relevant sections.

Gorden et al23 studied a cohort of 1092 US patients who underwent liver biopsy during bariatric surgery. Of the 1092 patients histologically assessed, 748 had evidence of NAFLD (187 of these diagnosed with NASH), with no evidence of steatosis in the remaining 344 patients who were used as the control group. A number of candidate genes including PNPLA3 were investigated, and rs738409 was found to be significantly associated with histological features of increased steatosis and hepatocyte ballooning. The LYPLAL, GCKR and PPP variants also studied were not associated with an increased risk of lobular inflammation, ballooning or fibrosis, but a trend towards increased steatosis was observed.

Sookoian et al24 performed a systematic review and meta-analysis of the PNPLA3 rs738409, comparing wild type (CC) and variant (GG) genotypes. Six of the studies with histological NAFLD assessment described above were included: Valenti,15 Valenti,16 Sookoian et al,12 Speliotes et al,20 Rotman et al,19 Hotta et al.13 Meta-analysis showed that NASH was more frequently associated with the variant genotype compared with the wild type. The variant genotype was also associated with increased inflammatory scores when four of these studies were considered, and significantly associated with fibrosis on the basis of five studies. The study also commented that the heterozygous risk was similar to the homozygous variant and suggested a dominant model for the risk of severe histological features.

LPIN1

Lipin is involved in triglyceride synthesis, lipid metabolism and adipocyte differentiation.9 The rs13412852 variant was investigated by Valenti et al25 in two Italian cohorts of 142 paediatric and 115 adults with NAFLD, compared with 337 healthy controls. With regard to the paediatric patients, the variant was significantly less frequent in NAFLD compared with control. Histologically, the homozygous variant genotype was associated with reduced fibrosis (OR 0.29, CI 0.11 to 0.66), and trended towards reduced NASH prevalence. No significant differences were noted between the adult patients and control group, although trends towards fewer NASH and reduced fibrosis were observed with the homozygous variant. The results suggest a protective role for the minor allele; however, the paediatric results should be interpreted cautiously as were compared with an adult control group.

APOC3, APOE and ORL1

The apolipoprotein C-III, encoded by APOC3 gene on chromosome 11, is a very low-density lipoprotein (VLDL) component that functions to inhibit lipoprotein lipase and hepatic lipase.9 Mutations in APOC3 are known to be associated with hypertriglyceridaemia and it is hypothesised that this protein delays catabolism of triglyceride-rich particles and is involved in hepatic lipid processing.9 The APOE gene on chromosome 19 encodes apolipoprotein E, the main apoprotein of the chylomicron that is involved in endocytosis and clearance of chylomicron and VLDL particles by the liver.9 Mutations in this protein are known to be associated with familial dysbetalipoproteinaemia.9 Oxidised low-density lipoprotein (LDL) receptor 1 is encoded by ORL1 on chromosome 12.9 The receptor binds oxidised LDL, which is then internalised and degraded in the liver.9

Valenti et al32 investigated APOC3 polymorphisms T-455C and C-482T in a cohort of 758 European (Italian and UK) patients with NAFLD. The case–control study between the Italian patients with NAFLD and 316 healthy Italian controls found no protective effect of the APOC3 wild type, and no significant association with histological characteristics or NASH diagnosis was observed. Verrijken et al17 also investigated the rs2854117 variant of APOC3, finding no significant associations with disease severity, fibrosis or other histological characteristics.

Sazci et al33 evaluated apolipoprotein E polymorphisms in 57 Turkish patients with NASH, compared with 245 healthy controls. The study found that the APOEε3 isoform was associated with an increased risk of NASH, whereas the heterozygous APOEε3ε4 was protective.

De Feo et al34 investigated the three major isoforms of APOE on an Italian population of 310 patients with NAFLD (109 biopsy proven) and 422 population controls. The case–control study found that at least one APOEε4 allele significantly reduced the risk of NAFLD (OR 0.51, CI 0.28 to 0.93) when compared with the homozygous wild type. Histological assessment in the biopsied subgroup found no significant association with APOE genotype and fibrosis.

Gambino et al35 compared 29 Italian patients with NASH with 27 controls but found no significant difference in APOE polymorphisms; however, this study has less power due to small sample size. Musso et al36 also investigated APOE isoforms on a larger Italian population, comparing 78 patients with NAFLD, 34 of whom were biopsied and diagnosed with NASH, to 156 healthy controls. APOE genotypes were not found to be significantly different. The study also compared 40 Italian patients with NASH with 40 ORL1 genotype-matched controls and concluded that the ORL1 IVS4–14 G allele was associated with the severity of steatosis, necroinflammation and fibrosis.

PEMT

Phosphatidylethanolamine N-methyltransferase (PEMT) is an enzyme in the liver that catalyses the conversion of phosphatidylethanolamine to phosphatidylcholine,9 a substance that is required for VLDL secretion. Dong et al37 evaluated the prevalence of the PEMT V175M allele in a Japanese population between 107 patients with NASH, a normal control population (n=150) and a cohort of 100 patients with chronic hepatitis C. Both the M allele frequency and M/− genotype were significantly increased in the NASH participants when compared with the control and patients with chronic hepatitis C. There were no significant differences in genotype or allele frequency between the normal and hepatitis C controls, indicating that the association observed is likely to be more specifically related to NASH than viral hepatitis.

MTTP

Microsomal triglyceride transfer protein (MTTP) is a heterodimeric protein involved in triglyceride, phospholipid and cholesterol ester transport,9 and lipoprotein assembly. The MTP gene located at 4q23 encodes the 88kD large subunit of MTTP, and mutations in this protein are associated with familial abetalipoproteinaemia.9 The encoded variant of MTP −493, with C to G transversion, undergoes less transcription than the wild type protein.38

Namikawa et al38 studied the variant −493 G allele of MTP in 63 biopsy proven patients with NASH with 150 healthy controls in a Japanese population. The MTP allele and homozygous variant genotype were both found to be significantly associated with NASH compared with controls. The variant genotype was also associated with increased steatosis and increased histological grade of NASH when compared with the heterozygous carriers.

Gambino et al35 described above also investigated the frequency of the homozygous MTP variant between patients with NASH and controls. No significant difference in allele frequency was observed between Italian patients with NASH and controls, but within the NASH group, the severity of histological steatosis, inflammation and fibrosis was increased in patients with the variant genotype when compared with heterozygotes and homozygous wild type patients. However, these results should be interpreted with caution due to the limited sample size of 27 patients.

Carulli et al39 performed a candidate gene study in an Italian population investigating the three candidate genes MTP, PCIKI and IL6. The study compared 114 patients with NAFLD, including 59 who underwent biopsy, against 79 healthy population controls. With regard to MTP, the case–control study was not significant, and there was no association with histological characteristics or NASH diagnosis within the biopsied group. The study also found no association with the PCIK I121Q variant, and results pertaining to IL6 are described in the relevant section.

Oliveira et al40 evaluated the MTP −493 (C/G) polymorphism in a cohort of 139 Brazilian patients with NAFLD, comprised of 45 with SS and 86 with NASH. The MTP G allele was not found to be associated with NAFLD when compared to 141 controls or with NASH when compared to SS.

El-Koofy et al41 investigated the association of MTTP and SOD2 SNPs with NAFLD in a paediatric Egyptian cohort. A total of 76 children with NAFLD were compared with 20 age-matched and sex-matched healthy controls. Within the NAFLD group, 33 underwent liver biopsy, and of these, 8 had SS, NASH was present in 7 cases, and the remaining 18 samples showed normal histology. The homozygous −493 MTP genotype (TT) was not present in any of the 15 patients with NASH or SS. No association was reported with the SOD2 118T/C variant, but when the homozygous wild types of MTP (GG) and SOD (TT) were combined, the study reported a significant risk of NASH compared with controls (OR 54, CI 4.1 to 707). These results should also be treated with caution as only seven patients diagnosed with NASH were studied.

SREBP

Sterol regulatory element-binding proteins are leucine zipper transcription factors that regulate the transcription of enzymes involved in cholesterol biosynthesis.9 Musso et al42 investigated the effect of the rs11868035 variant of SREBP factor 1C on histological severity in 42 biopsy-proven Italian patients with NAFLD. The wild type and heterozygous genotypes were found to be associated with an increased severity of steatosis, necroinflammation, increased NAS, increased fibrosis score and increased prevalence of NASH when compared with the mutant genotype. Multivariate analysis showed this SREBP1C polymorphism to be independently associated with both severe steatosis and NASH. In a separate study, the same group evaluated the effect of polymorphisms in SREBP factor 2 in biopsied patients with NAFLD.43 The study reported that patients with heterozygous and homozygous variant genotypes demonstrated an increased severity of histological characteristics and NASH diagnosis compared with wild type patients. The SREBPF2 variant allele was found to be an independent predictor of NASH within these patients (OR 2.92, CI 2.08 to 4.18).

SOD2

The superoxide dismutase 2 (SOD2) gene on chromosome 6 encodes the enzyme manganese superoxide dismutase.9 This is located on the outer mitochondrial membrane, and catalyses the conversion of superoxide by-products of oxidative phosphorylation to hydrogen peroxide and oxygen.9 Mutations in SOD2 have been previously associated with idiopathic cardiomyopathy, certain cancers and motor neuron disease.9 The rs4880 C47T polymorphism of SOD2 is a C to T transversion that results in a valine to alanine substitution in the mitochondrial-targeting region of the protein and may alter the tertiary protein structure.44

Al-Serri et al44 conducted a linkage study on 71 Italian families and found that the C47T SNP was preferentially transmitted. The study then evaluated the C47T SOD2 polymorphism in a European cohort of 510 patients with NAFLD. The study showed the SOD2T variant was associated with advanced fibrosis (scores of >1) with an additive effect. When the PNPLA3 rs738409 was controlled for, the SOD2 was still associated with the degree of histological steatosis, and was an independent risk factor for advanced fibrosis. However, the SOD2 allele did not affect the diagnosis of NASH.

Namikawa et al38 performed a case–control study between 63 patients with NASH and 150 healthy controls in a Japanese population to evaluate the SOD2 C47T allele. The study found the homozygous variant genotype (TT) was significantly associated with NASH, however the T allele frequency was not. The MTP polymorphism was also investigated (described above), and the proportion of patients with NASH with both GG (MTP) and TT (SOD2) genotypes was significantly increased compared with controls.

UCP3

Uncoupling protein 3 encoded by the UCP3 gene is a mitochondrial proton carrier located at 11q13,9 primarily expressed in skeletal muscle.9 The protein is a mitochondrial anion carrier and facilitates the transport of anions from the inner to outer mitochondrial membrane, thus reducing the mitochondrial membrane potential.9 The protein is considered to be protective against oxidative stress from β oxidation of fatty acids.9 Aller et al45 investigated the −55C/T polymorphism of UCP3 in a cohort of 39 obese patients who underwent liver biopsy. The heterozygous genotype was found to be associated with an increase in both moderate-to-severe histological inflammation and moderate-to-severe steatosis.

GCLC

The catalytic subunit of glutamate-cysteine ligase (GCLC) is involved in the synthesis of glutathione (GSH).9 GSH is a tripeptide that counteracts cellular oxidative stress from reactive metabolites through thiol group reduction.46 Oliveira et al40 evaluated the −129C/T GCLC polymorphism in a cohort of 139 Brazilian patients with NAFLD, 45 with SS and 86 with NASH. The investigators observed a significant association between the T allele of GCLC with patients with NASH when compared with the SS group.

TNF

Valenti et al47 investigated two TNF polymorphisms (TNFA 238 and TNF2 308) in 99 patients with NAFLD (53 biopsied) and compared with 172 healthy controls. The allele frequency and genotype of TNF2 was not different between NAFLD and controls; however, both the TNFA variant allele and genotype were significantly increased in NAFLD. In the biopsied subgroup, 35 patients had NASH whereas 18 had SS. A trend towards a higher prevalence of TNF polymorphisms in NASH subgroup was observed when compared with those with SS, but this did not reach statistical significance, which may be due to the small sample size. Yang et al48 investigated these two TNF variants in obese children; histologically, 61 were normal and used as control, 17 had SS and 33 had NASH. No significant difference was observed.

Hu et al49 also investigated the same variants, comparing a cohort of 189 Chinese patients with NAFLD, 44 diagnosed histologically, to 138 healthy controls. Neither variant was homozygous in either the patient or control groups. The heterozygous genotype and A allele frequency of the TNFA variant was found to be significantly associated with NAFLD compared with control, with relative risk 2.19. No histological associations were observed in the biopsied subgroup, although the sample size for comparison was small.

Chowdhury et al50 evaluated the effect of the −308 and −238 TNF variants on fibrosis in NAFLD, investigating 29 Indian patients with biopsy-proven NAFLD. The variants were not found to be associated with fibrosis severity; however, DNA amplification only worked in 25 cases, reducing the sample size.

Aller et al51 investigated the −308TNF polymorphism on histological characteristics in 66 patients with NAFLD and 213 obese controls. Variant frequency was not significantly different between patients and controls, although within the patient group those carrying the heterozygous variant (n=15) had significantly more severe inflammation and fibrosis than those with wild type genotype; however, the sample size is small.

Tokushige et al52 investigated six TNF polymorphisms in a Japanese population of 102 patients with NAFLD and 100 controls, and found no significant differences between the NAFLD and control groups. Of the patients with NAFLD, 36 had SS and 66 had NASH: the −1031C and −863A variants were significantly more frequent in the NASH group compared with SS, but there were no associations with histological features.

Wong et al53 investigated the role of three variants of TNF in a Chinese population. A cohort of 79 patients with NAFLD, comprising 61 individuals with SS and 18 patients diagnosed with NASH, was compared with 40 healthy population controls. The results showed no significant differences in the frequency of the three TNF polymorphisms investigated.

MIF

Akyildiz et al55 investigated the association of the −173 G/C MIF variant with NAFLD and severity. Ninety-one patients with NAFLD stratified as 37 with NASH, 44 probable NASH and 10 SS were compared with 104 healthy controls. MIF variant genotype and allele were not significantly different between NAFLD cases and control, or within the NAFLD groups.

Hormones

Adiponectin, encoded by ADIPOQ, and leptin are hormones secreted by adipocytes. Adiponectin is involved in the regulation of energy homoeostasis and metabolism, particularly through enhancement of insulin sensitivity.9 Reduced plasma levels of adiponectin have been associated with features of the metabolic syndrome including NAFLD. Leptin acts through its receptor, encoded by LEPR, to stimulate satiety and reduce food intake, and also increase energy expenditure.9 Studies have investigated polymorphisms in ADIPOQ and LEPR for association with NAFLD.

Transcription factors

HFE

The HFE gene on chromosome 6 encodes a membrane protein that is hypothesised to regulate iron absorption through the interaction of transferrin with its receptor; homozygotes for mutations (C282Y or H63D) in this gene develop hereditary haemochromatosis.9

Valenti et al68 evaluated polymorphisms in HFE in an Italian population comprising 134 patients with NAFLD and 291 healthy controls. Within the NAFLD cohort, 67 underwent biopsy, and NASH was found in 42 patients, whereas 25 had SS. The case–control analysis showed that C282Y (but not H63D) was significantly more prevalent in patients with NAFLD compared with controls. There were no significant differences in frequency of either polymorphism between patients with and without NASH, although only a percentage of the patients with NAFLD were histologically assessed. The same group69 studied the HFE C282Y and H63D mutations in a larger Italian population. The study compared a cohort of 587 histologically diagnosed patients with NAFLD to 184 healthy controls. The study found the frequency of variant alleles was not significant. It was noted that hepatocellular iron deposition was an independent risk factor for moderate-to-severe fibrosis, but no significant association was found between HFE genotype and severity of fibrosis.

Deguti et al70 evaluated 32 Brazilian patients with histological NASH and found no evidence of association between either C282Y or H63D variant, HFE mutations with histological injury and did not observe significant iron deposition. Zamin et al71 also studied the effects of these HFE variants on histological features. A cohort of 29 Brazilian patients with NASH was compared with 20 healthy controls and 20 patients with hepatitis C. The study found the prevalence of the mutation was similar between groups, with no significant association between mutation and fibrosis score or other histological characteristics. An association between liver iron overload and mutation was noted, but this did not correlate with fibrosis score.

Raszeja-Wyszomirska et al72 studied the effects of C282Y and H63D HFE mutations on a Polish cohort of 62 biopsy-proven patients with NAFLD. The authors scored patients on fibrosis as described and compared those with no or mild fibrosis (score 0–2, n=48) against those with advanced fibrosis (score 3–4, n=14). It was observed that in cases with severe fibrosis, the variant H63D allele was associated with increased serum levels of total cholesterol and LDL cholesterol, but no other associations between histological disease severities were apparent.

Nelson et al73 characterised the C282Y and H63D HFE SNPs in 786 participants in the NASH CRN cohort. The relationship between genotype and histological features, including steatosis, inflammation, hepatocyte ballooning, fibrosis stage and NAS,3 was evaluated. It was found that participants with the C282Y mutant allele were more likely to have increased iron deposition and hepatocellular pattern, but H63D was not associated with increased hepatic iron accumulation. Both HFE SNPs were shown to be associated with an increased steatosis grade when compared with wild type, and H63D allele was also associated with an increase in NAS. A significant reduction in hepatocyte ballooning and definitive NASH associated with the C282Y SNP was also observed; however, there was no association between HFE SNPs and fibrosis stage. The authors concluded that the H63D allele was an independent risk factor for the highest grade steatosis, lobular inflammation and NAS ≥5, and definitive diagnosis of NASH.

TMPRSS6

The TMPRSS6 gene encodes transmembrane protease, serine 6, a cell surface enzyme involved in liver matrix remodelling.9 The encode enzyme also cleaves a co-receptor necessary for the transcription of hepcidin; therefore, mutations in this gene can affect cellular iron status.74

Valenti et al74 compared the frequency of the TMPRSS6 AV736 polymorphism in a cohort of 216 Italian patients against a 271 population control. The V/V genotype was observed to be associated with reduced hepatic iron and less severe hepatocyte ballooning, but other histological characteristics of NASH were similar. The study concluded that TMPRSS6 is not a risk factor for the development of NAFLD.

SERPINE 1

Espino et al78 compared the 4G/5G variant of SERPINE1 (PAI-1) in 50 obese Chilean patients undergoing bariatric surgery and 71 controls. Within the obese patient group, 21 were histologically normal, 9 had steatosis and 20 met criteria for NASH. No differences in variant frequency were observed between patients and controls, or those with and without NASH-related fibrosis.

INSR

The insulin receptor is a tyrosine kinase membrane receptor involved in glucose homoeostasis. The protein encoded by ENPP1 on chromosome 6 is a cell membrane protein that may be involved in insulin sensitivity and IRS-1 encodes a substrate involved in the insulin receptor pathway.9

Dongiovanni et al79 evaluated the effect of polymorphisms ENPP1 rs1044498 and IRS-1 rs1801278 in a European cohort. A case–control study of 240 Italian patients with NAFLD and healthy population controls showed no significance. Histological features were assessed on a cohort of 702 Italian patients; both polymorphisms were associated with increased fibrosis and the rs1044498 variant allele of ENPP1 was associated with fibrosis severity.

GCKR

GCKR encodes glucokinase regulator, a protein that modulates the activity of the enzyme glucokinase in liver cells by reversibly binding to form an inactive complex.9 Tan et al80 investigated the GCKR rs780094 (C>T) and rs1260326 (C>T) SNPs on a mixed ethnicity Malaysian cohort of 144 patients with NAFLD and 198 controls. The variant T allele frequencies of each SNP was found to be more prevalent in patients with both NAFLD and NASH when compared with controls; however, when the ethnic subgroups were considered, the relationship was only significant in the Indian population, not those of Chinese or Malay ethnicity. Owing to the mixed ethnicity of this cohort, the results should be interpreted with caution.

TCF7L2

Transcription factor 7-like 2 is a transcription factor expressed in T cells containing a high mobility group box that is thought to be involved in glucose homoeostasis.9 Musso et al81 investigated the rs7903146 variant of the TCF7L2 gene on an Italian population. A total of 78 patients with NAFLD, 34 of whom were biopsied and diagnosed with NASH, were compared with 156 healthy population controls. It was found that the wild type genotype was less frequent in patients with NAFLD compared with controls. The study reports that severe hepatic steatosis was predicted by TCF (OR 2.2, CI 1.7 to 4.9) and grade 3 fibrosis was predicted by TCF (OR 2.0, CI 1.7 to 4.3); however, the analysis of histological characteristics is limited due to small sample size and because all patients had NASH.

NCAN

Gorden et al23 studied a cohort of 1092 US patients undergoing bariatric surgery, as described above. The variant in NCAN was significantly associated with increased histological features of steatosis, lobular inflammation and perivenular fibrosis, and trended towards increased ballooning. However, it is unclear whether these associations are independent of the association with the PNPLA3 rs738409 shown to be significantly associated with steatosis and ballooning in the same study.

TM6SF2

TM6SF2 encodes a transmembrane protein of unknown function;9 the rs58542926 SNP in TM6SF2 is in strong linkage disequilibrium with the rs2228603 NCAN variant discussed above, and associations attributed to NCAN may relate instead to this variant.84

Liu et al84 studied the effect of the rs58542926 variant of TM6SF2 on two cohorts with histologically characterised NAFLD. Quantitative analysis was performed within a discovery cohort of 349 UK patients with NAFLD, and results were validated using a cohort of 725 European Caucasian patients. Within the NAFLD discovery cohort, minor allele was significantly more frequent than the reference population, OR 1.72 (CI 1.16 to 2.57) and OR 4.84 (CI 1.01 to 22.9) for heterozygous and homozygous variants, respectively, when compared with healthy control group (n=379).

Histological characteristics were further investigated: degree of steatosis was not significantly associated with either rs2228603 in NCAN or rs58542926 in TM6SF2, in either cohort, although a trend was observed when the cohorts were combined. The authors suggest a possible relationship with a small effect size, which was supported by an association which was noted when the combined group was categorised into mild or pronounced steatosis. Severity of steatohepatits, assessed by characteristics of necroinflammation and hepatocyte ballooning, was associated with the TM6SF2 variant within the discovery cohort, but was not replicated in the validation or combined cohorts.

Importantly, the variants of NCAN and TM6SF2 were both significantly associated with the stage of histological fibrosis. On further investigation, conditioning the results on TM6SF2 negated the effect of the NCAN variant, indicating TM6SF2 was responsible for the association relationship. This relationship was replicated in the validation and combined cohorts, and was independent of confounding factors including age, gender, BMI, presence of diabetes and PNPLA3 phenotype. When mild fibrosis was compared with advanced fibrosis in the combined cohort, each minor allele was associated with OR 1.88 (CI 1.41 to 2.50). The study also compared 99 UK patients with NAFLD-related HCC to the combined cohort. Although the homozygous variant of TM6SF2 was found associated with an increased risk of HCC in univariate analysis, this did not persist in multivariate analysis adjusted for confounding factors.

Since this study, two other studies have investigated the role of this polymorphism on histologically proven NAFLD: Dongiovanni et al85 studied a large, combined cohort of adult and paediatric European patients undergoing liver biopsy for hepatology or bariatric purposes. Of these, 112 patients had no fatty liver disease and served as the control group, 412 had SS and 677 fulfilled histological NASH diagnosis. The results showed the variant to be associated with severity of all histological characteristics (steatosis, necroinflammation, ballooning and fibrosis stage), and this persisted after adjustment for age, gender, BMI, diabetes, recruitment and PNPLA3 genotype. The variant was associated with NASH compared with SS (OR 1.84, CI 1.23 to 2.79) and stage 3–4 fibrosis (OR 2.08, CI 1.20 to 3.55), although association with fibrosis did not remain after adjustment for NASH.

Sookoian et al86 performed a case–control study investigating the role of on TM6SF2 in an Argentinian population of 226 patients with histological NAFLD. In comparison to a control group of 135 individuals with no ultrasonographic evidence of fatty liver, the rs58542926 variant was found to be significantly associated with fatty liver (OR 1.37, CI 1.02 to 1.84); however, this relationship did not persist after adjusting for age, BMI and rs738409 in PNPLA3. The NAFLD group was subdivided into 96 patients with SS and 130 with NASH; the T allele was found to be associated with disease progression when the control group was included in the analysis; however, this was not significant when the SS group was compared with NASH. Within the patients with NAFLD, degree of steatosis was found to be significantly associated with this variant, independently of PNPLA3 genotype, age, sex and BMI. In contrast, no associations were observed with other histological characteristics of lobular inflammation, ballooning or fibrosis stage (F1–2 vs F3–4), and there was no association with increased NAS (defined as >5).

SPINK1

The SPINK 1 gene encodes serine peptidase inhibitor, Kazal type 1.9 This enzyme inhibits trypsin, and is a component of pancreatic juice, secreted by pancreatic acinar cells; mutations result in familial calcific pancreatitis.9 Oruc et al87 evaluated the SPINK1 N34S P555 polymorphism in 50 Turkish patients with NAFLD and 44 controls. Of the NAFLD group, 30 patients had SS and 20 had NASH. No significant association was found between the SPINK1 variant with NAFLD or histological NASH.