Background

Nonalcoholic fatty liver disease (NAFLD) is identified as a metabolic stress injury of the liver, as well as a burgeoning public health issue/economic burden with a high global prevalence of up to ~ 25% [1, 2]. The incidence of NAFLD corresponds to the escalation of obesity, type 2 diabetes mellitus and metabolic syndrome [3]. Various evidence has indicated the heritable contribution to the prevalence, development and progression of NAFLD. The last few decades witness the substantial progress in the genetic susceptibility of NAFLD, providing exciting insight in refining the management and treatment of NAFLD patients. A recent prospective twin study performed by Loomba R et al. concluded that hepatic steatosis and hepatic fibrosis have approximately 52% heritability and 50% heritability in multivariable-adjusted models (adjust for sex, age and ethnicity), respectively [4]. However, we cannot clarify the molecular genetic mechanisms of NAFLD by only gene polymorphism [5].

Notably, two ‘star gene’ variants, PNPLA3 I148M variant and TM6SF2 E167K variant, are recognized as the major genetic modifiers of NAFLD [6,7,8,9,10]. More recently, the EASL-EASD-EASO clinical practice guidelines for the management of NAFLD recommended that carriers of the two ‘star gene’ variants show a higher liver fat content and increased risk of nonalcoholic steatohepatitis [11]. Interestingly, recent studies have demonstrated that PNPLA3 I148M and TM6SF2 E167K variants may have an additive effect in the regulation of lipid metabolism [8, 12,13,14]. However, no study has focused on the aforementioned association in Qingdao Han Population, China, an international city with more than ~ 9.2 million people after the 18th Meeting of the Council of Heads of Member States of the Shanghai Cooperation Organization. The present study sought to investigate the potential additive effect of the two variants on the risk of NAFLD in Qingdao Han Population, China.

Methods

Subjects and methods

The present study was conducted in accordance with the principles of the World Medical Association Declaration of Helsinki. All the subjects have signed written informed consent according to the study protocol approved by the Ethics Committee of Qingdao Municipal Hospital before participation (Approval NO.2017–20).

We recruited a total of 512 unrelated Qingdao Han NAFLD patients determined by liver ultrasonography (a fairly reliable and accurate noninvasive method with sensitivity of 84.8% and specificity of 93.6% [15]) and 451 healthy controls matched for age and sex in the present study. The healthy controls were performed liver ultrasonography to rule out NAFLD. All the subjects were of Qingdao Han ethnicity, China and included from the health examination center and the Department of Gastroenterology of Qingdao Municipal Hospital. The diagnosis of NAFLD was determined by a same experienced operator in accordance with the guidelines for management of NAFLD of the Chinese Liver Disease Association in 2010 [16]. Other etiologies contributing to fatty liver disease, including alcohol consumption (≥ 20 g/day for females and ≥ 30 g/day for males), hepatitis B infection, hepatitis C infection (genotype 3), autoimmune hepatitis, Wilson’s disease and drug-induced liver injury et al. were excluded [3].

Blood samples were collected after 12-h fast for serological assays and genetic analysis. Serum concentration of albumin (ALB), fasting plasma glucose (FPG), alanine aminotransferase (ALT), aspartate aminotransferase (AST), γ-glutamyltransferase (GGT), alkaline phosphatase (ALP), triglyceride (TG), total cholesterol (TC), high-density lipoprotein (HDL) and low-density lipoprotein (LDL), uric acid (UA) were obtained using standard methods in the Central Laboratory of Qingdao Municipal Hospital.

Genotyping

Genomic DNA Purification Kit (BioTeke, Biotechnology, Beijing, China) was used to extract genomic DNA according to the manufacturer’s protocol from the noncoagulated blood samples. The genotyping of PNPLA3 I148M variant and TM6SF2 E167K variant was performed by polymerase chain reaction (PCR) with the following primers. Primers for PNPLA3 I148M and TM6SF2 E167K were: 5′- AACTTCTCTCTCCTTTGCTTTCACA -3′ (forward), 5′- GGAGGGATAAGGCCACTGTAGA -3′(reverse); 5′- TGTCTCAGAACAAACAAACAAACAGA -3′ (forward), 5′- GTAGGGGATGGTGAGGAAGAAG -3′ (reverse). The PCR amplification profile was conducted as follows: 95 °C for 10 min, 40 cycles before denaturation at 94 °C for 1 min, annealing at 58 °C for 1 min and elongation 30 s at 70 °C. The PNPLA3 and TM6SF2 genotypes were detected by ABI 3730XL (Foster City, CA, USA) and calculated by Gene Mapper 4.1 software.

Statistical analysis

SPSS 20.0 statistical software (SPSS Inc., Chicago, IL, USA) was performed for statistical analysis. The baseline characteristics between healthy controls and NAFLD patients were showed as mean ± standard deviation, and the differences were examined using student’s t test or paired samples t test. The Hardy-Weinberg equilibrium was measured using the χ2 test. Genotype and allele frequencies between healthy controls and NAFLD patients were analyzed by chi-square test. The DNA distributions between the two groups were determined by χ2 test and Fisher’s exact test where appropriate. The association between the PNPLA3 I148M and/or TM6SF2 E167K variants and NAFLD was evaluated by the odds ratio (OR) with 95% confidence interval (CI) and performed by logistic regression analysis. In addition, an additive model (by coding the genotypes 0, 1 and 2 for CC, CG and GG, respectively) for PLPLA3 gene and a dominant model (by coding the genotypes 0 and 1 for CC and CT + TT, respectively) for TM6SF2 gene were assumed to assessed the potential additive effect of the two variants. Statistical significance was defined as P-value less than 0.05.

Results

Demographic and clinical characteristics

Demographic and clinical characteristics of healthy controls and NAFLD patients were presented in Table 1. When compared to the healthy controls, NAFLD patients had higher BMI, serum levels of FPG, ALT, AST, GGT, ALP, TG, TC, LDL, UA and decreased ALB, HDL. All the difference reached significance (all P < 0.05).

Table 1 Demographic and clinical characteristics of healthy controls and nonalcoholic fatty liver disease patients
Full size table

Associations between PNPLA3 I148M and TM6SF2 E167K variants and NAFLD

The genotypic distribution of PNPLA3 I148M and TM6SF2 E167K were both in Hardy-Weinberg equilibrium in both two groups (P control = 0.243, 0.773; P NAFLD = 0.113, 0.313). The minor allele frequencies of the two single-nucleotide polymorphisms (SNPs) in the whole subjects were 45.48% for PNPLA3 I148M and 6.69% for TM6SF2 E167K, respectively. As listed in Table 2, both the differences in the genotypes and allele frequencies of the PNPLA3 I148M and TM6SF2 E167K reached significance (all P < 0.05). Moreover, Table 3 showed that the PNPLA3 I148M variant was significantly associated with the risk of NAFLD in an additive model (CG, OR = 2.092, 95% CI: 1.551–2.820, P = 0.000; GG, OR = 4.566, 95% CI: 3.141–6.638, P = 0.000, respectively). Interestingly, it suggested a strong link between the TM6SF2 E167K variant and the risk of NAFLD in a dominant model (CT + TT, OR = 2.327, 95% CI: 1.542–3.513, P = 0.000).

Table 2 Distribution of genotypes and allele frequencies of I148M and E167K variants in the whole subjects
Full size table
Table 3 Associations between I148M and E167K variants and nonalcoholic fatty liver disease
Full size table

Furthermore, Table 4 demonstrated the association between the two variants and clinical features. Subjects with the G allele of PNPLA3 I148M had higher BMI, serum levels of ALT, AST, GGT, ALP, TG, TC, LDL and UA (all P < 0.05). The serum level of HDL did not reached significance (P = 0.967). However, only serum concentrations of some lipid profiles (TG and TC; both P < 0.05) showed inversely strong links between carriers of the T allele of TM6SF2 E167K and the noncarriers. The T allele of TM6SF2 E167K was associated with lower levels of serum TG and TC (P = 0.003, P = 0.006).

Table 4 Association between I148M and E167K variants and clinical features in the whole subjects
Full size table

Additive effects of PNPLA3 I148M and TM6SF2 E167K variants on the risk of NAFLD

Moreover, the potential additive effect of the two variants on the risk of NAFLD was conducted in the study subjects. We counted risk alleles at the two gene locus (for PNPLA3 148 locus, by coding the genotypes 0, 1 and 2 for CC, CG and GG; for TM6SF2 167 locus, by coding the genotypes 0 and 1 for CC and CT + TT, respectively). The group with 0 risk allele was regarded as the control group. Excitingly, the prevalence of NAFLD increased following with the increase of the number of risk alleles (36.77, 49.53, 71.56 and 77.78% of subjects with 0, 1, 2 and 3 risk alleles, respectively, Fig. 1). In addition, the increasing of the number of risk alleles showed a strong link with the risk of NAFLD (1 risk allele, OR = 1.687, 95% CI: 1.226–2.321, P = 0.001; 2 risk alleles, OR = 4.326, 95% CI: 3.100–6.037, P = 0.000; 3 risk alleles, OR = 6.018, 95% CI: 1.229–29.459, P = 0.027, respectively, Table 5).

Fig. 1
figure 1

Prevalence of NAFLD according to the number of risk alleles

Prevalence of NAFLD according to the number of risk alleles is shown. We counted risk alleles at the two gene locus (for PNPLA3 148 locus, by coding the genotypes 0, 1 and 2 for CC, CG and GG; for TM6SF2 167 locus, by coding the genotypes 0 and 1 for CC and CT + TT, respectively). The prevalence of NAFLD increased following with the increase of the number of risk allele.

Full size image
Table 5 OR (95% CI) for nonalcoholic fatty liver disease in subjects with different number of risk alleles
Full size table

Discussion

Our study aimed to assess the potential additive effect of the PNPLA3 I148M and TM6SF2 E167K variants on the risk of NAFLD in Qingdao Han Population, China. We confirmed the significant association between both the PNPLA3 I148M and TM6SF2 E167K variants and the risk of NAFLD in a cohort consisting of a total of 512 NAFLD patients and 451 age and sex matched healthy controls. Moreover, combining the I148M and E167K variants could improve risk prediction for NAFLD in a Qingdao Han Population.

NAFLD ranked as the most concerned chronic liver disease in the coming decades worldwide, and, nonalcoholic steatohepatitis is recognized as the second leading indication for liver transplantation in the USA [17,18,19]. Notably, there is a genetic susceptibility for NAFLD [1, 3, 11]. Mahdessian H et al. firstly reported that TM6SF2 inhibition induced significant reduction of the expression of PNPLA3 gene in both Huh7 and HepG2 cells, which, to some extent, indicating a joint/additive effect between PNPLA3 gene and TM6SF2 gene in determining lipid metabolism [20]. More recently, a multiethnic study [12] of obese children and adolescents (including Caucasians, African Americans, and Hispanics) summarized that there is a joint effect among PNPLA3 I148M, TM6SF2 E167K, and GCKR rs1260326 single nucleotide polymorphisms in regulating intrahepatic fat accumulation. Similarly, Wang and colleagues [13] demonstrated the additive effect of PNPLA3 I148M and TM6SF2 E167K variants on NAFLD in a Chinese cohort. In the present study, we observed that PNPLA3 I148M variant and TM6SF2 E167K variant were both major and independent genetic determinants of NAFLD in a Qingdao Han Population, consistent with previous studies [21, 22]. More importantly, subjects in our study with increasing number of risk alleles (PNPLA3 148 locus G allele and TM6SF2 167 locus T allele) showed an additive effect of the two variants, compared to individuals with 0 risk allele (1 risk allele, OR = 1.687, P = 0.001; 2 risk alleles, OR = 4.326, P = 0.000; 3 risk alleles, OR = 6.018, P = 0.027, respectively). Furthermore, our pilot study [23] performed a Bayesian analysis and evaluated the additive effect of the two variants on hepatocyte lipid metabolism as well as the underlying mechanism in vitro. Our date suggested that the PNPLA3 I148M and TM6SF2 E167K variants may increase hepatic lipid content by upregulating the expression of sterol regulatory element-binding transcription factor 1c and fatty acid synthase. However, to the best of our knowledge, all other previous studies have not reported the underlying molecular mechanism of this potential additive effect. These encouraging findings provide the potential perspective in new NAFLD classification and the probability of offering a new important tool for risk prediction of NAFLD in the coming years.

We additionally investigated the relations between PNPLA3 I148M and TM6SF2 E167K variants and serum concentrations of lipid profiles. Several studies showed inconsistent metabolic associations of NAFLD risk alleles. For instance, some studies demonstrated a strong link between TM6SF2 E167K variant and lower plasma levels of TC, TG and LDL [8, 24,25,26], but the differences in several other studies did not reach the significance [9, 27, 28]. Similarly, the association between the PNPLA3 I148M variant and plasma concentrations of lipid profiles was also inconsistent. [6, 29,30,31]. Our data demonstrated that subjects with the G allele of PNPLA3 I148M had higher serum levels of TG, TC and LDL (all P < 0.05). And, the serum concentrations of some lipid profiles (TG and TC; both P < 0.05) showed inversely strong links between carriers of the T allele of TM6SF2 E167K and the noncarriers. The T allele of TM6SF2 E167K was associated with lower concentrations of serum levels of TG and TC (P = 0.003, P = 0.006). These different findings may inasmuch as the population ethnicity, diagnostic method of NAFLD and demographic and clinical characteristics. Therefore, the relationship between the two ‘star gene’ variants and clinical features of NAFLD population needs further investigation.

This study has some limitations. First one is the perform of ultrasonography to determine NAFLD, instead of magnetic resonance imaging and/or liver biopsy. Notably, liver ultrasonography cannot provide reliable quantitative data [32]. In addition, we conducted the present study in a single center and the findings may have limited application value in other different populations. Furthermore, the size of subjects in the study was not sufficiently large to comprehensively explore the associations.

Conclusions

In conclusion, we confirmed the significant association between both the PNPLA3 I148M and TM6SF2 E167K variants and the risk of NAFLD in a Qingdao Han Population cohort. More importantly, combining the I148M and E167K variants could improve risk prediction for NAFLD in a manner of an additive effect. This study may provide the potential perspective in new NAFLD classification and the identification of high risk NAFLD populations in the future.