Association between rs738409 and rs2896019 single-nucleotide polymorphisms of phospholipase domain-containing protein 3 and susceptibility to nonalcoholic fatty liver disease in Greek children and adolescents.

Background: Several studies have detected a strong association linking rs738409 and rs2896019 polymorphisms in the patatin-like phospholipase domain-containing protein 3 (PNPLA3) gene with hepatic steatosis and steatohepatitis. In the present study we aimed to determine the association of those PNPLA3 variants with nonalcoholic fatty liver disease (NAFLD) susceptibility in obese and nonobese Greek children and adolescents.
Methods: The study recruited 91 children and adolescents of Greek descent with NAFLD or biopsy-proven nonalcoholic steatohepatitis, and 91 healthy subjects of normal weight (control group) with sex distribution similar to the patient group. DNA samples were amplified using polymerase chain reaction with specifically designed primers. Data were analyzed using the statistical software SPSS version 24.0.
Results: A significant correlation was shown between the rs738409 polymorphism (CG and GG genotypes) and the rs2896019 polymorphism (TG genotype) with the development of hepatic steatosis (P<0.001). The incidences of rs738409 GG, rs738409 CG and rs2896019 TG genotypes were found to be increased in patients with hepatic steatosis (obese and nonobese), but not in obese patients without liver disease. The combined expression of the 2 polymorphisms was associated with a lower age of diagnosis of hepatic steatosis in nonobese patients. Conclusions: We confirmed a strong association between the 2 polymorphisms and hepatic steatosis. The association of the rs2896019 single-nucleotide polymorphism with hepatic steatosis in obese and nonobese pediatric patients, and the combined study of both polymorphisms in a pediatric population of Greek origin are described for the first time. Copyright: © Hellenic Society of Gastroenterology.

Introduction

Nonalcoholic fatty liver disease (NAFLD) is the most common cause of chronic liver disease in children and a significant cause of liver-related morbidity and mortality [1]. The prevalence of childhood obesity is increasing, whereas the incidence of hepatic steatosis is estimated to be almost 10% in the pediatric population and may be as high as 38% in obese children [2]. Children with NAFLD are at higher risk of progressive liver disease and cirrhosis, with a consequent need for liver transplantation [3-6]. Environmental risk factors are widely known to influence the development and progression of NAFLD [7]. Genetic variability plays a critical role in NAFLD predisposition and is characterized by gene polymorphisms. The major genetic determinants of the pathogenesis and severity of liver steatosis are presented in Table 1 [8-25].

Table 1

Genetic associations implicated in the pathogenesis of NAFLD

Adiponutrin/patatin-like phospholipase domain-containing protein 3 (PNPLA3), or adiponutrin, belongs to a group of lipid-metabolizing enzymes first identified in potato tubers. Most members of the PNPLA family are soluble proteins and show nonspecific lipid acyl hydrolase activity [26]. Adiponutrin has been shown to possess both triacylglycerol lipase and acylglycerol transacylase properties [27]. Among the 9 members of the PNPLA family, PNPLA3 plays a critical role in triglyceride metabolism by mediating a rate-limiting step in triglyceride hydrolysis [28,29]. Genome-wide association studies link the nonsynonymous genetic polymorphisms rs738409 and rs2896019 in the PNPLA3 gene to hepatic triglyceride accumulation (steatosis), inflammation, fibrosis, cirrhosis, and even hepatocellular carcinoma [30-32]. Considering the morbidity associated with NAFLD in the pediatric population, it is important to identify children with hepatic steatosis who have the highest risk of developing progressive liver disease [33].

In the present study we aimed to determine: 1) the associations of the PNPLA3 rs738409 and rs2896019 single-nucleotide polymorphisms (SNPs) with NAFLD susceptibility in obese and nonobese Greek children and adolescents with hepatic steatosis; 2) the associations between the SNPs, anthropometric variables, and the laboratory findings; and 3) the associations of polymorphisms with either early-onset of steatosis or steatohepatitis in a subgroup of patients undergoing a liver biopsy.

Patients and methods

Study population and data collection

This was a cross-sectional case-control study in a population of Greek origin. We studied 182 children and adolescents monitored in the Pediatric Gastroenterology Unit of the Fourth Department of Pediatrics of the Aristotle University of Thessaloniki.

The patients were divided into 3 groups (A, B, and C) according to clinical, laboratory and imaging findings. Group A included 31 overweight or obese patients (body mass index, BMI ≥85th percentile) with NAFDL or nonalcoholic steatohepatitis (NASH), of whom 21 were boys and 10 were girls (mean age 10.45±2.39 years). Group B included 33 overweight or obese subjects (BMI ≥85th percentile) without liver disease, of whom 17 were boys and 16 were girls (mean age 10.18±2.5 years). Group C included 27 non-overweight, nonobese patients (BMI <85th percentile) with NAFLD or NASH, of whom 14 were boys and 13 were girls (mean age 9.15±3.86 years). The patients in this group were selected on the basis of hepatic steatosis detected on liver ultrasonography performed as part of a diagnostic investigation. In addition, a control group (Group D) included 91 age- and sex-matched, phenotypically healthy children and adolescents with normal liver ultrasonography and normal laboratory findings.

All the subjects were tested for secondary causes of steatosis, such as the use of drugs known to precipitate steatosis, whereas viruses were ruled out using the appropriate tests. In all cases, autoimmune liver disease, metabolic liver diseases, Wilson’s disease, celiac disease, and alpha-1-antitrypsin deficiency were eliminated by standard clinical, and laboratory evaluations and liver biopsy. The study was performed in accordance with the principles outlined in the 1975 Declaration of Helsinki. Informed consent was obtained from each responsible guardian.

Anthropometric and biochemical measures

Detailed records of previous medical and family histories were reviewed. All the participants underwent a systematic physical examination by the same pediatrician, and a blood sample was collected for determination of biochemical and immunological parameters. Waist circumference (WC) was also assessed. BMI was calculated using the formula, body weight (kg)/height2 (m2). Blood pressure was measured using the appropriate cuff size, with the participant in the sitting position. The lipid profile (total cholesterol [TC], low-density lipoprotein-cholesterol [LDL-C], high-density lipoprotein-cholesterol [HDL-C], and triglycerides) was assessed in the morning blood sample obtained after a 12-h fast, using the Abbott Architect c16000 Automatic Biochemistry Analyzer. The homeostasis model assessment of insulin resistance (HOMA-IR) index was used to evaluate insulin resistance. Normal lipid levels were considered as TC values <200 mg/dL, LDL-C levels <130 mg/dL, HDL-C levels >40 mg/dL, triglycerides levels <100 mg/dL for children up to 10 years of age and <130 mg/dL for children >10 years of age.

Genotyping

DNA was extracted using the QIAamp DNA Blood Mini Kit (Qiagen, Venlo, Limburg, The Netherlands), and DNA samples were amplified by polymerase chain reaction as previously described (Dutta AK, 2012), using the sense primer 5′-TGGGCCTGAAGTCCGAGGGT-3′ and antisense primer 5′-CCGACACCAGTGCCCTGCAG-3′ for the rs738409 polymorphism and the sense primer 5′-CCTTCCCCTAAACCCACAAT-3′ and the antisense primer 5′-CATGACAGCCCTTTCCTCAT-3′ for the rs2896019 polymorphism [34,35].

Statistical analysis

A descriptive statistical analysis was performed for all the study data. Phenotypic quantitative data were expressed as mean ± standard deviation. The Kolmogorov-Smirnov test was used to test the normality of the distributions of the quantitative variables. For the qualitative variables, frequencies were expressed in absolute values and percentages. Categorical variables were presented as frequencies and percentages (n, [%]). The parametric method of the Student’s t test was used when the data of the variables followed a normal curve, and the non-parametric method of the Mann-Whitney U test was used otherwise. The independence test between the categorical variables was performed using the Fisher χ2 test for dichotomous variables and with the Pearson χ2 test for variables characterized by more than 2 levels (3 genotypes to polymorphism). To assess the association between the genotypes and NAFLD or the quantitative traits, such as alanine aminotransferase (ALT) and aspartate aminotransferase (AST), we used the χ2 test and logistic regression or multiple regression, adjusting for covariables such as age, HOMA-IR index and BMI. Finally, we used operating characteristic (receiver-operating characteristic, [ROC]) analyses to assess the usefulness of the laboratory values, clinical measures, and the PNPLA3 rs738409 and rs2896019 SNPs as predictors for discriminating between the NAFLD cases and controls. P-values <0.05 were considered statistically significant and, where appropriate, were adjusted for multiple comparisons. For this purpose, the criterion χ2 was used as a goodness-of-fit test. Data were analyzed using the statistical software Statistical Package for Social Science version 24.0 software for Windows (SPSS Inc., Chicago, Illinois, USA). The HOMA-IR index was calculated as the product of the fasting plasma insulin concentration and the fasting plasma glucose (mg/dL) concentration divided by 405 (HOMA-IR = [insulin × glucose]/405).

Results

Case control association study

The frequency distribution of PNPLA3 rs738409 and rs2896019 SNPs was in Hardy-Weinberg equilibrium. In groups A and C, 58 patients were found to have hepatic steatosis; 11 of them underwent a liver biopsy and were diagnosed as having steatohepatitis. The criteria for liver biopsy in patients with hepatic steatosis were a positive family history for NAFLD, NASH, or diabetes mellitus; abnormal ALT values (>80 IU/mL); and hepatosplenomegaly during clinical examination [36].

rs738409 and rs2896019 polymorphisms in the study population

The frequencies of the PNPLA3 rs738409 and the rs2896019 SNPs in the study population are shown in Table 2. The frequency of normal homozygotes (rs738409 CC) was found to be higher in the obese patients without liver disease (75.8% in group B and 73.6% in the control group) but was lower in the patients with hepatic steatosis (groups A and C: 35.5% and 7.4%, respectively). The incidence of rs738409 polymorphism (CG+GG) was found to be 58.2% in the patient population (groups A, B, and C), higher than that in the control subjects (27.5%). The frequency of the rs738409 GG genotype was found to be higher (33%) in the patient groups and much lower in the control group (2.2%). In the patients with hepatic steatosis (groups A and C), the frequency of rs738409 polymorphism (CG+GG) was 77.6%, and the frequency of G allele was 46.6% (Table 2).

Table 2

Frequency of the rs738409 and rs2896019 SNPs in the study population

Hepatic steatosis and steatohepatitis were diagnosed in 90% of the rs738409 GG carriers and in 78.3% of the rs738409 CG carriers. The carriers of the PNPLA3 risk allele (rs378409 GG) showed the highest probability for developing liver disease (odds ratio [OR] 20.73, 95% confidence interval [CI] 7.38-58.23, P<0.001; and OR 7.4, 95%CI 0.4-125.8, P<0.001 after logistic regression), in contrast to the carriers of the wild allele (CC: OR 0.1, 95%CI 0.05-0.21; P<0.001, and OR 0.01, 95%CI 0.001-0.29; P<0.001 after logistic regression; Table 3).

Table 3

Correlations of the rs738409 genotypes with the clinical and laboratory findings

Regarding the rs2896019 polymorphism, the incidence of normal homozygotes (TT) was found to be greater in the obese patients (groups A and B) and in the control group (71%, 72.7% and 80.2%, respectively), and smaller in the normal-weight patients with NAFLD (33.3%). Correspondingly, the incidence of the rs2896019 polymorphism (TG+GG) was 39.6% in the patient population (group A, B, and C), greater than in control subjects (19.8%). In the patients with hepatic steatosis (groups A and C), the frequency of the rs2896019 polymorphism (TG+GG) was 46.6% (Table 2).

NAFLD was diagnosed in 45.5% of the normal homozygotes (TT) and 64.7% of the heterozygotes (TG). The TG genotype was strongly associated with NAFLD (OR 3.06, 95%CI 1.56-6.01; P<0.001, and OR 2.07, 95%CI 0.1-38.7; P<0.001 after logistic regression; Table 4). The combined presence of the 2 polymorphisms, as found in 27 patients, regardless of the combination of the alleles, increased the probability for developing liver disease from 2 to 8 times higher (OR 4.27, 95%Cl 2.12-8.58; P<0.001).

Table 4

Correlations of the rs2896019 genotypes with the clinical and laboratory findings

The mean ages at NAFLD and NASH diagnosis were 9.85±3.25 years (n=47) and 9.82±3.13 years (n=11), respectively. The mean age at NAFLD diagnosis in the carriers of both risk alleles of rs738409 SNP was 1 year younger than that in the non-carriers, whereas the mean age at NASH diagnosis was 2.7 years (P<0.001). Regarding the rs2896019 polymorphism, comparisons between the carriers and non-carriers of both risk alleles showed an earlier age at diagnosis of liver disease (1.5 years earlier for NAFLD and 1.6 years for NASH, P=0.754) The patients with a combination of genotypes of the 2 polymorphisms (heterozygous or homozygous form) had a younger age at NAFLD and NASH diagnosis, at the statistical significance level of P=0.096. The age difference in the diagnosis of hepatic steatosis and steatohepatitis was more pronounced in the group of patients with lean NAFLD (group C).

Association of rs738409 and rs2896019 with the clinical and anthropometric parameters

The patients with the rs738409 CG and rs738409 GG genotype showed a statistically significantly lower mean BMI than the patients with the rs738409 CC genotype (21.99±6.1 kg/m2 and 21.33±7.7 kg/m2 vs. 26.97±4.2 kg/m2, P<0.001). The mean BMI for the carriers of rs2896019 polymorphism was 22.4±9.4 kg/m2 for those carrying the rs2896019 GG genotype and 21.96±8.2 kg/m2 for those carrying the rs2896019 TG genotype. A higher mean BMI (25.1±4.99 kg/m2) was found in the patients with genotype rs2896019 TT (Tables 3 and 4).

No significant correlation was detected between the rs738409 and rs2896019 polymorphisms and the mean WC, systolic blood pressure (SBP) or diastolic blood pressure (Tables 3 and 4).

Association of rs738409 and rs2896019 with the laboratory findings

Children homozygous for the PNPLA3 minor allele (rs738409 GG) showed a stronger correlation between AST, ALT, TC, triglycerides, and LDL-C levels. The carriers of the rs738409 G allele were found to have moderate-to-severe insulin resistance as compared with the carriers of the wild allele (OR 6.89, 95%CI 2.97-16.03; P<0.001 vs. OR 0.35, 95%CI 0.17-0.76; P=0.008; Table 3). The rs2896019 TG allele was observed to increase the probability of elevated ALT values (OR 2.95, 95%CI 1.36-6.44; P=0.010). A statistically significant correlation was found between abnormal HOMA-IR index values and the TG genotype (OR 2.59, 95%CI 1.21-5.57; P=0.021). No statistically significant correlation was found between high AST, TC, triglycerides or HDL-C levels and the 3 genotypes of the rs2896019 polymorphism (Table 4).

Patients with NAFLD and lean NAFLD

Comparison of the clinical and laboratory parameters of groups A and C (obese and nonobese children with hepatic steatosis or steatohepatitis) revealed several differences, some of which were assessed as statistically significant. The frequencies of the rs738409 CC and rs2896019 TT genotypes were higher in the obese group (P<0.001), whereas that of the rs2896019 TG genotype was higher in the normal-weight group (P<0.001). No significant statistical difference was found between the mean age at diagnosis and the sex of patients in the 2 groups (P=0.123). We found no significant differences between the AST and ALT levels in both groups (P=0.723 and P=0.526 respectively; Table 5). The lowest HOMA-IR index values were observed in the patients with lean NAFLD (P=0.067; Table 5).

Table 5

Comparison of various quantitative phenotypes among the different genotypes at rs738409 and rs2896019 in PNPLA3 in the obese and nonobese patients with NAFLD

ROC curve for the rs738409 and rs2896019 SNPs

To identify the children at higher risk for developing liver disease, we performed a ROC curve analysis. The first approach concerned the assessment of prognostic capacity based individually on gene testing (genotypes CG and GG of the rs738409 polymorphism and genotypes TG and GG for the rs2896019 polymorphism). The area under the curve (AUC) index was 0.804, with a minimum value of 0.5 and an excellent value of 1.00. Predictiveness assessment was performed with a χ2 test (OR 24.37, 95%CI 7.9-74.9; P<0.001; Fig. 1).

Figure 1

Receiver operating characteristic curve for predictive ability of the rs738409 and rs2896019 single-nucleotide polymorphisms

The second approach assessed the prognostic ability by combining genetic testing with laboratory findings (ALT and HOMA-IR index). The AUC was 0.998 (OR 3477, 95%CI 308.9-39139.6; P<0.001; Fig. 2).

Figure 2

Receiver operating characteristic curve for predictive ability of the rs738409 and rs2896019 single-nucleotide polymorphisms combined with alanine aminotransferase level and homeostatic model assessment for insulin resistance

Discussion

In this study, we evaluated the effects of PNPLA3 rs738409 and rs2896019 SNPs on the development of hepatic steatosis in children and adolescents of Greek origin and the probability of early detection of patients at higher risk for developing liver disease. To our knowledge, this is the first report of the association of rs2896019 polymorphism with hepatic steatosis in obese and nonobese pediatric patients, and the combined study of rs738409 and rs2896019 polymorphisms exclusively in the pediatric population.

The total frequencies of the rs738409 (CG+GG) and rs2896019 (TG+GG) genotypes in the control group (27.5% and 19.8%, respectively) are in line with the frequencies reported for the European population (23.3% and 20.4%, respectively) by the National Centre for Biotechnology Information [37,38]. The frequency of the rs738409 G allele in our study agrees with other studies conducted on children worldwide (Table 6) [39-43]. The frequencies of the rs738409 polymorphism genotypes and their association with hepatic steatosis and steatohepatitis are consistent with studies involving other white populations from Italy [33], Finland [10], Germany [44], England and Switzerland [45], as well in populations of Spanish [46] and Asian origins [47,48]. There are no relevant literature data for the comparison of the frequency of the rs2896019 polymorphism in European pediatric populations. Most references relate mainly to adult patients of Asian descent, such as the studies by Islek et al, Kitamoto et al, and Kawaguchi et al [35,49,50].

Table 6

The frequency of the rs738409 G allele in the pediatric population

In the multivariate logistic regression analysis, we observed that both rs738409 and rs2896019 were significantly associated with fatty liver, which indicates that the carriers of the G allele (OR 7.4, 95%CI 0.4-125.8, P<0.001; and OR 2.07, 95%CI 0.1-38.7, P<0.001, respectively) were more likely to have NAFLD than the non-carriers, independently of age, sex, BMI and HOMA-IR index. The conclusions of the present study agree with the literature on the association of the rs738409 polymorphism with hepatic steatosis, as found in several studies in both adult and pediatric patients [10,33,44,51-53]. Although the literature data on the rs2896019 polymorphism are limited, studies in Caucasian, Hispanic, and Asian populations also correlate the TG and GG genotypes of the rs2896019 polymorphism with hepatic steatosis [49,50,54].

In the present study, we also focused on the role of adiponutrin gene polymorphisms in both obese and nonobese patients with hepatic steatosis. The incidence of normal homozygotes of the 2 polymorphisms was greater in the obese group without liver disease (75.8% and 72.7%), whereas higher incidences of the rs738409 CG+GG and rs2896019 TG+GG genotypes were observed in the patients with normal BMI values (BMI <85th percentile) and hepatic steatosis or steatohepatitis (92.6% and 66.7%, respectively). Adiponutrin gene polymorphisms were not correlated with the clinical features of metabolic syndrome (BMI, WC, SBP, DBP). Although obesity continues to be a major risk factor for hepatic steatosis, the genetic background appears to burden both obese and nonobese patients in terms of both the early onset and the severity of the disease.

The incidence of hepatic steatosis in the normal-weight patients in this study population was 14.8%, consistent with the incidences of hepatic steatosis reported in other studies of normal-weight patients (lean NAFLD) belonging to white or Asian populations [55-61]. The presence of the rs738409 CG and rs738409 GG genotypes could be related to the increased morbidity observed in the patients with normal BMI and hepatic steatosis, considering the role of the polymorphism in the process of lipogenesis, in the accumulation of triglycerides in the hepatic cells, and in the inhibition of lipophagy, which occurs regardless of total body fat.

Regarding the patients’ biochemical test results, the rs738409 G allele was significantly associated with the serum levels of both AST and ALT, whereas the rs2896019 G allele was associated only with the ALT levels. Several studies have correlated elevated aminotransferase levels, particularly ALT, with the presence of the 2 polymorphisms, although significant population variations have been observed [33,47,62-65]. In studies in the pediatric population, Goran et al [46] and Carrasco et al [66] also found a significant correlation between liver enzyme levels, especially ALT, and the rs738409 GG genotype, in both obese and normal-weight children (P<0.001).

Drawing conclusions from the present study, we can say that the sample of the present study was quantitatively satisfactory for conducting a monocentric study in children and adolescents of Greek descent who have hepatic steatosis. However, indications of the low frequency of the rs2896019 GG genotype in the Greek population certainly exist, and a multicenter study may be required to draw safe conclusions about this particular genotype.

The clinical usefulness of our findings lies in the investigation of the role of adiponutrin polymorphisms, a powerful genetic factor in liver steatosis, in children and adolescents of Greek origin. Such a study is needed in the Greek population to determine the frequency of the polymorphisms of the adiponutrin gene, given its large geographical variation. Understanding the contribution of the adiponutrin gene to the natural history of hepatic steatosis could be crucial for developing future preventive or therapeutic strategies aimed at preventing the progression of hepatic steatosis to more severe forms of liver disease. The high prevalence of hepatic steatosis in the general population and the progression of the disease to steatohepatitis, cirrhosis, and hepatocellular carcinoma, which increases morbidity and mortality in adulthood, are a major challenge for the existing health infrastructure.

What is already known:

Children with obesity have at least a 38% chance of developing fatty liver disease; variations in several genes contribute to the risk of developing hepatic steatosis

Data indicate that the rs738409 patatin-like phospholipase domain-containing protein 3 (PNPLA3) single nucleotide polymorphism (SNP) is associated with hepatic steatosis

The prevalence of PNPLA3 polymorphisms differs between geographic areas

The rs2896019 polymorphism may also be involved in the pathogenesis of hepatic steatosis

What the new findings are:

The frequencies of the rs738409 and rs2896019 PNPLA3 variants were greater in obese and nonobese pediatric patients of Greek origin with hepatic steatosis

Earlier diagnosis of hepatic steatosis and steatohepatitis was necessary in nonobese carriers of both SNPs

Obese children with a normal genotype were less likely to develop hepatic steatosis early in life

Nonobese pediatric patients with hepatic steatosis had predominantly abnormal genotypes