Genetic variants in GCKR and PNPLA3 confer susceptibility to nonalcoholic fatty liver disease in obese individuals1–3 Yu-Cheng Lin, Pi-Feng Chang, Mei-Hwei Chang, and Yen-Hsuan Ni
INTRODUCTION
Concurrent with the epidemic of childhood obesity, pediatric nonalcoholic fatty liver disease (NAFLD)4 is an important public health issue (1, 2). The fact that not all obese children develop NAFLD suggests that environmental and/or genetic factors may influence the susceptibility of each obese individual. The first genome-wide association study (GWAS) of NAFLD from the Dallas Heart Study identified the rs738409 single
nucleotide polymorphism (SNP) in patatin-like phospholipase domain-containing-3 (PNPLA3) to be strongly associated with liver fat content (3). Subsequent studies have confirmed this association (4–6). A recent GWAS from The Genetics of Obesity-Related Liver Disease Consortium identified 4 additional genetic variants in rs2228603, rs12137855, rs780094, and rs4240624 for NAFLD in 7177 adults of European ancestry (7). The 4 newly identified susceptibility loci were located in or near neurocan (NCAN), lysophospholipase-like 1 (LYPLAL1), glucokinase regulatory protein (GCKR), and protein phosphatase 1 regulatory subunit 3b (PPP1R3B). The prevalence of NAFLD varies among different ethnic groups (8). Two recent studies attempted to verify the findings of the Genetics of Obesity-Related Liver Disease Consortium in multiethnic cohorts. Palmer et al (9) measured hepatic steatosis by using computed tomography in adult individuals of African and Hispanic ethnicities. The results showed that the allele frequency and effect size of PNPLA3 rs738409, NCAN rs2228603, LYPLAL1 rs12137855, GCKR rs780094, and PPP1R3B rs4240624 varied in different ethnic groups. Hernaez et al (10) also reported a lack of consistency in these variants in the NHANES III study population of multiethnicities. Although these studies extended the initial findings in European adults to Africans and Hispanics, there is no report in Asians. Because the effect size of genetic variants tends to be more pronounced in children than in adults because of less confounding factors, such as lifestyle habits and comorbidities (11), it would be
1
From the Department of Pediatrics, Far Eastern Memorial Hospital, Taipei, Taiwan (Y-CL and P-FC); the Departments of Pediatrics, National Taiwan University Hospital, Taipei, Taiwan (M-HC and Y-HN); and the Oriental Institute of Technology, New Taipei City, Taiwan (Y-CL). 2 Supported by research grants from the Far Eastern Memorial Hospital (FEMH-2013-C-023), the Far Eastern Memorial Hospital–National Taiwan University Hospital Joint Research Program (101-FTN-07), and the National Science Council, Taiwan (NSC-98-2628-B-002-006-MY3). 3 Address correspondence to Y-H Ni, Department of Pediatrics, National Taiwan University Hospital, No. 8, Zhongshan South Road, Zhongzheng District, Taipei City 10041, Taiwan. E-mail: [email protected]. 4 Abbreviations used: ALT, alanine aminotransferase; AST, aspartate aminotransferase; GWAS, genome-wide association study; NAFLD, nonalcoholic fatty liver disease; SNP, single nucleotide polymorphism; WHR, waist-to-hip ratio. Received November 10, 2013. Accepted for publication January 6, 2014. First published online January 29, 2014; doi: 10.3945/ajcn.113.079749.
Am J Clin Nutr 2014;99:869–74. Printed in USA. Ó 2014 American Society for Nutrition
869
Downloaded from ajcn.nutrition.org at MOUNT SAINT VINCENT UNIVERSITY LIBRARY on October 3, 2014
ABSTRACT Background: A genome-wide association study identified variants in or near patatin-like phospholipase domain-containing-3 (PNPLA3), neurocan (NCAN), lysophospholipase-like 1 (LYPLAL1), glucokinase regulatory protein (GCKR), and protein phosphatase 1 regulatory subunit 3b (PPP1R3B) that were strongly associated with nonalcoholic fatty liver disease (NAFLD) in adults of European ancestry. Objective: We examined these genetic variants in obese children and tested whether their effects on NAFLD are significant in the Taiwanese Han Chinese population. Design: We genotyped PNPLA3 rs738409, NCAN rs2228603, LYPLAL1 rs12137855, GCKR rs780094, and PPP1R3B rs4240624 in 797 obese children aged 7–18 y. NAFLD was identified by liver ultrasonography. We analyzed the effect of these genetic variants on NAFLD. Results: NAFLD was identified in 24% of the recruited obese children. We found significant associations with NAFLD at variants in PNPLA3 and GCKR but not in NCAN, LYPLAL1, and PPP1R3B. Multiple logistic regression analysis showed that, after control for the effects of age- and sex-adjusted body mass index, waist-to-hip ratio, sex, and PNPLA3 rs738409 polymorphism, the variant GCKR rs780094 TT genotype independently increased the OR of NAFLD by 1.997 (95% CI: 1.196, 3.335; P = 0.008) compared with the CC genotype. Subjects with the variant GCKR rs780094 TT genotype had a higher mean serum alanine aminotransferase concentration than did those with the CC genotype (30.8 6 34.7 compared with 22.2 6 18.6 IU/L; P = 0.01). Conclusions: By studying the genetic variants of obese Taiwanese children, we confirmed that the genetic variants in GCKR rs780094 and PNPLA3 rs738409, but not in NCAN rs2228603, LYPLAL1 rs12137855, and PPP1R3B rs4240624, are associated with an increased risk of NAFLD. GCKR and PNPLA3 variants are the common genetic factors that may confer susceptibility to NAFLD in obese individuals across multiple ethnic groups. This trial was registered at clinicaltrials.gov as NCT00274183. Am J Clin Nutr 2014;99:869–74.
870
LIN ET AL
advantageous to identify these genetic variants in children. Herein we aimed to investigate the effects of genetic variants in NCAN rs2228603, LYPLAL1 rs12137855, GCKR rs780094, and PPP1R3B rs4240624 in obese Taiwanese children.
SUBJECTS AND METHODS
Subjects
Data collection The following data were obtained from each subject: age, sex, BMI, and waist and hip circumferences. BMI was calculated as body weight (kg)/height (m)2. The ratio between waist and hip circumferences was calculated and recorded as the waist-to-hip ratio (WHR). None of the subjects had significant alcohol use. Environmental factors, such as diet and physical activity, were not recorded in this study. In fasting venous blood samples, we measured total serum bilirubin, alanine aminotransferase (ALT), aspartate aminotransferase (AST), g-glutamyltransferase, fasting glucose, triglycerides, total cholesterol, and HDL cholesterol. Insulin concentration was measured by using a chemiluminescence immunoassay (Diagnostic Products Corporation). Insulin resistance was measured by HOMA-IR and was calculated as follows: HOMA-IR = [fasting insulin (mU/mL)][fasting glucose (mmol/L)]/22.5 (13). For the subjects with abnormal ALT concentrations, we ruled out causes of hepatitis other than NAFLD by measuring hepatitis B surface antigen, anti-hepatitis C antibody, antinuclear antibody, and ceruloplasmin.
Liver ultrasonography All participants underwent liver ultrasonographic examination performed by one operator. The hand-carried machine used (TITAN; SonoSite Ltd) was equipped with a 2–5-MHz convex probe. NAFLD was defined as the presence of an ultrasonographic pattern consistent with the following criteria: liver-kidney echo discrepancy, attenuated echo penetration, visibility of diaphragm, and obscure hepatic vessel structures. The aforementioned ultrasonographic pattern was scored as described by Chan et al (14). These children were categorized to have mild, moderate, or severe steatosis if the overall score was 1–3, 4–6, or 7–9, respectively. Because ultrasonography is much more accurate at detecting moderate-severe steatosis than mild steatosis (15), we set a score $4 as the diagnostic criterion for NAFLD to avoid possible false-positive results.
Genomic DNA was extracted from venous blood from each participant by using a commercial kit (Puregene; Gentra Systems) according to the manufacturer’s instructions. After extraction, the genomic DNA was immediately stored at 2808C. The TaqMan SNP genotyping assays C_7241_10 for PNPLA3 rs738409, C_16171492_10 for NCAN rs2228603, C_31403184_10 for LYPLAL1 rs12137855, C_2862873_10 for GCKR rs780094, and C_1614473_10 for PPP1R3B rs4240624 SNPs (Applied Biosystems) were performed on an ABI 7300 Real-Time PCR System (Applied Biosystems). The allele frequencies of SNP genotypes were tested for the Hardy-Weinberg equilibrium. Statistical analysis The statistical analysis was performed by using Stata statistical software (version 11.2; StataCorp). A 2-sided P value #0.05 was considered statistically significant. The standardized BMI z scores were not available for children in Taiwan. To adjust for the intrinsic effects of age and sex on BMI, we derived a variable as “adjusted BMI” by subtracting the Taiwanese children’s age- and sex-specific population median BMI from each subject’s BMI value. In univariate analysis, the chi-square test and 1-factor ANOVA were used for the comparisons of categorical and continuous variables. Multivariate analysis was conducted to identify predictive factors of pediatric NAFLD and serum ALT concentrations by fitting multiple logistic and linear regression models, respectively. To ensure the quality of the results, we used stepwise procedures for variable selection, goodness-of-fit assessment, and regression diagnostics in our regression analyses. The SNP genotypes were coded as a categorical variable as follows: 0 = 2 major alleles (the reference group), 1 = 1 minor allele + 1 major allele (heterozygous), and 3 = 2 minor alleles (homozygous). The best final regression model was identified manually by reducing the significance levels to 0.05 corresponding to the chosen a level. RESULTS
Genotype distribution of PNPLA3 rs738409, NCAN rs2228603, LYPLAL1 rs12137855, GCKR rs780094, and PPP1R3B rs4240624 Of the 797 enrolled obese subjects, 191 (24%) had an ultrasonographic score $4 and received a diagnosis of NAFLD. The genotype distributions of these SNPs between the NAFLD and non-NAFLD groups are shown in Table 1. The numbers of missing values were 0 for PNPLA3 rs738409, 0 for NCAN rs2228603, 3 for LYPLAL1 rs12137855, 1 for GCKR rs780094, and 6 for PPP1R3B rs4240624. Correspondingly, call rates for each SNP genotyping were 100%, 100%, 99.6%, 99.9%, and 99.3%, respectively. Two of 5 variants, PNPLA3 rs738409 and GCKR rs780094, were significantly associated with NAFLD. The percentage of PNPLA3 rs738409 GG genotypes was significantly higher in the subjects with NAFLD than in those without NAFLD (OR: 2.079; P = 0.003). The GCKR rs780094 TT genotype was more prevalent in the subjects with NAFLD (OR: 1.733; P = 0.018). The frequencies of the G allele of rs738409 at PNPLA3 and the T allele of rs780094 at GCKR were significantly higher in the
Downloaded from ajcn.nutrition.org at MOUNT SAINT VINCENT UNIVERSITY LIBRARY on October 3, 2014
From October 2006 to September 2012, 797 obese children aged 7–18 y were enrolled from schools in Taipei, Taiwan. All subjects were unrelated and ethnically Han Chinese. Obesity was defined as a BMI greater than the 95th percentile by different age and sex groups according to the standards of the Department of Health in Taiwan (12). The study was performed according to the principles of the Declaration of Helsinki. All parents gave their informed consent, and the study was approved by the ethics committee of the Far Eastern Memorial Hospital and National Taiwan University Hospital.
Genotyping
871
GENETIC VARIANTS IN NAFLD TABLE 1 Genotype and minor allele frequencies of PNPLA3 rs738409, NCAN rs2228603, LYPLAL1 rs12137855, GCKR rs780094, and PPP1R3B rs42406241 Genotype
Non-NAFLD
P value2
OR (95% CI)
58 (30) 95 (50) 38 (20) 0.448
238 (39.3) 293 (48.3) 75 (12.4) 0.366
0.129 0.003 0.004
1.330 (0.921, 1.921) 2.079 (1.284, 3.368) 1.407 (1.115, 1.776)
168 (88) 23 (12) 0 (0) 0.060
532 (87.8) 74 (12.2) 0 (0) 0.061
0.950 — 0.985
0.984 (0.600, 1.615) — 0.985 (0.611, 1.590)
161 (85.6) 27 (14.4) 0 (0%) 0.072
525 (86.6) 78 (12.9) 3 (0.5%) 0.069
0.129 — 0.868
1.129 (0.706, 1.804) — 1.039 (0.665, 1.624)
41 (21.5) 91 (48.0) 58 (30.5) 0.545
174 (28.7) 290 (47.9) 142 (23.4) 0.474
0.174 0.018 0.016
1.332 (0.882, 2.010) 1.733 (1.100, 2.733) 1.330 (1.056, 1.676)
182 (96.8) 6 (3.2) 0 (0) 0.016
592 (98.2) 11 (1.8) 0 (0) 0.009
0.259 — 0.2612
1.774 (0.671, 4.694) — 1.762 (0.671, 4.625)
1
The numbers of missing values were 0 for PNPLA3 rs738409, 0 for NCAN rs2228603, 3 for LYPLAL1 rs12137855, 1 for GCKR rs780094, and 6 for PPP1R3B rs4240624. MAF, minor allele frequency; NAFLD, nonalcoholic fatty liver disease. 2 As compared with the reference genotype group, the differences in genotype frequencies were analyzed by using a chi-square test. P , 0.05 was considered to be statistically significant.
subjects with NAFLD (P = 0.004 for rs738409 and P = 0.016 for rs780094). In contrast, the minor allele frequencies of NCAN rs2228603, LYPLAL1 rs12137855, GCKR rs780094, and PPP1R3B rs4240624 were low and not different between the NAFLD and non-NAFLD groups. The allele frequencies of all SNP genotypes were in Hardy-Weinberg equilibrium in both the NAFLD and non-NAFLD groups (all P . 0.05). Subject characteristics stratified by PNPLA3 rs738409 and GCKR rs780094 genotypes The basic characteristics of the study subjects stratified by PNPLA3 rs738409 genotype are shown in Table 2: 296 with CC genotype, 388 with CG genotype, and 113 with GG genotype. We found significant differences in ALT and AST concentrations. The mean (6SD) serum ALT concentration was 34.2 6 36.3 IU/L for the GG genotype (highest), 27.7 6 31.8 IU/L for the CG genotype, and 22.36 18.5 IU/L for the CC genotype (lowest) (P , 0.001). The basic characteristics of the study subjects stratified by GCKR rs780094 genotype are shown in Table 3: 215 with the CC genotype, 381 with the CT genotype, and 200 with the TT genotype. We found statistically significant differences in AST, ALT, total cholesterol, and triglyceride concentrations. The mean serum ALT concentration was 30.8 IU/L for the TT genotype (highest), 26.9 IU/L for the TC genotype, and 22.2 IU/L
for the CC genotype (lowest) (P = 0.01). Subjects carrying the T allele of GCKR rs780094 had higher serum total cholesterol and triglyceride concentrations (P , 0.001). Multivariate analysis of predicting factors for NAFLD and serum ALT concentration As shown in Table 4, the fitted final multiple logistic regression model revealed that the variant PNPLA3 rs738409 and GCKR rs780094 genotypes were independent risk factors of NAFLD after control for the effects of sex, adjusted BMI, and WHR. This final multiple logistic regression model was obtained after including age, sex, adjusted BMI, WHR, HOMA-IR, PNPLA3 rs738409, NCAN rs2228603, LYPLAL1 rs12137855, GCKR rs780094, and PPP1R3B rs4240624 genotypes in the stepwise variable selection procedure. Specifically, when compared with the wild type, the subjects with the PNPLA3 rs738409 GC genotype would have a 1.608 (95% CI: 1.058, 2.442; P = 0.026) times higher odds of having NAFLD, and those with the GG genotype would have a 2.812 (95% CI: 1.598, 4.950; P , 0.001) times higher odds of having NAFLD, whereas the values of the other covariates were fixed. The subjects carrying the GCKR rs780094 TT genotype would have a 1.997 (95% CI: 1.196, 3.335; P = 0.008) times higher odds of having NAFLD. No significant difference was found for the GCKR rs780094 TC genotype.
Downloaded from ajcn.nutrition.org at MOUNT SAINT VINCENT UNIVERSITY LIBRARY on October 3, 2014
PNPLA3 rs738409 [n (%)] CC GC GG MAF (G allele) NCAN rs2228603 [n (%)] CC TC TT MAF (T allele) LYPLAL1 rs12137855 [n (%)] CC TC TT MAF (T allele) GCKR rs780094 [n (%)] CC CT TT MAF (T allele) PPP1R3B rs4240624 [n (%)] AA GA GG MAF (G allele)
NAFLD
872
LIN ET AL
TABLE 2 Basic characteristics of the subjects across PNPLA3 rs738409 genotypes1 Variable
CG (n = 388)
GG (n = 113)
11.7 6 1.93 11.7 6 1.9 11.5 6 2.0 203 (68.6) 273 (70.4) 79 (69.9) 65.7 6 17.1 64.8 6 15.9 64.2 6 16.9 154.1 6 13.2 153.3 6 12.3 153.7 6 13.7 27.2 6 3.9 27.1 6 3.5 26.7 6 3.8 89.0 6 10.7 88.9 6 9.4 87.6 6 11.4 0.90 6 0.06 0.90 6 0.05 0.90 6 0.06 118.6 6 13.8 119.8 6 15.0 116.6 6 11.8 73.6 6 10.3 74.2 6 11.4 75.1 6 10.7 22.0 6 8.7 25.0 6 14.5 27.7 6 16.8 22.3 6 18.5 27.7 6 31.8 34.2 6 36.3 17.4 6 11.7 19.2 6 17.9 20.3 6 19.4 86.0 6 6.8 86.6 6 13.3 86.8 6 9.9 17.6 6 20.6
17.1 6 21.0
P value2 0.723 0.833 0.632 0.752 0.424 0.455 0.863 0.097 0.468 ,0.001 ,0.001 0.190 0.687
16.2 6 16.4
0.815
3.82 6 4.93 3.70 6 4.77 3.48 6 3.63 158.9 6 32.7 157.3 6 34.1 158.1 6 39.1
0.802 0.839
101.7 6 54.2
93.7 6 44.3
92.6 6 46.0
0.066
46.8 6 11.6
46.6 6 10.6
47.0 6 12.3
0.938
1
ALT, alanine aminotransferase; AST, aspartate aminotransferase; DBP, diastolic blood pressure; SBP, systolic blood pressure; gGT, g-glutamyltransferase. 2 Calculated for the comparisons between the 3 genotype groups by using ANOVA for continuous variables and a chi-square test for categorical variables. 3 Mean 6 SD (all such values).
Similarly, the fitted final linear regression model showed a significant association of variant PNPLA3 rs738409 with the GCKR rs780094 genotypes with a higher mean serum ALT concentration (Table 5). After adjustment for the effects of the other covariates, the PNPLA3 rs738409 GC genotype would increase the mean serum ALT concentration by 5.43 IU/L (P = 0.011) and the GG genotype would increase it by 12.48 IU/L (P , 0.001) compared with the wild CC genotype. The mean ALT concentration of the subjects carrying the GCKR rs780094 TT genotype was 8.64 IU/L higher than that of the reference group (P = 0.001).
DISCUSSION
We found that genetic variants in PNPLA3 rs738409 and GCKR rs780094 conferred susceptibility to NAFLD in our obese schoolchildren and adolescents. To our knowledge, this is the first study to show the effect of GCKR variants on NAFLD in a pediatric Asian population. In our obese Taiwanese children and adolescents, the homozygote for the risk T allele in GCKR rs738409 has a 73% higher OR of NAFLD risk. The effect of this variant was similar in all ethnic groups, but the minor T allele frequency was lowest in Africans (0.18–0.2), intermediate in Europeans (0.39–0.41) and Hispanics (0.35–0.38), and highest in Han Chinese (0.49 in this study) (7, 9, 10). Thus, we estimated that the risk T allele in
TABLE 3 Basic characteristics of the subjects across GCKR rs780094 genotypes1 Variable
CC (n = 215)
CT (n = 381)
TT (n = 200)
Age (y) 11.7 6 1.83 11.7 6 1.9 11.6 6 2.0 Male [n (%)] 152 (70.7) 268 (70.3) 136 (68.0) Weight (kg) 65.7 6 15.8 64.8 6 16.3 64.6 6 17.2 Height (cm) 154.0 6 12.3 153.7 6 12.9 153.1 6 13.4 BMI (kg/m2) 27.3 6 3.6 27.0 6 3.8 27.0 6 3.7 Waist (cm) 88.9 6 10.1 88.7 6 10.0 88.6 6 10.6 Waist-to-hip ratio 0.90 6 0.06 0.90 6 0.06 0.90 6 0.06 SBP (mm Hg) 118.7 6 13.7 118.5 6 14.4 120.0 6 14.2 DBP (mm Hg) 74.1 6 11.8 74.6 6 10.6 73.2 6 10.5 AST (IU/L) 22.4 6 8.3 24.2 6 12.7 26.3 6 17.5 ALT (IU/L) 22.2 6 18.6 26.9 6 29.4 30.8 6 34.7 gGT (IU/L) 17.3 6 16.7 18.2 6 15.7 21.0 6 16.3 Fasting glucose 87.3 6 10.3 86.4 6 12.4 85.4 6 7.5 (mg/dL) Fasting insulin 17.4 6 19.3 16.3 6 17.2 18.4 6 25.9 (mU/mL) HOMA-IR 3.83 6 4.67 3.50 6 4.08 3.92 6 5.67 Total cholesterol 153.8 6 31.8 156.3 6 34.8 165.7 6 35.4 (mg/dL) Triglyceride 87.0 6 40.7 94.1 6 46.3 111.3 6 56.7 (mg/dL) HDL cholesterol 45.9 6 9.9 46.7 6 12.2 47.4 6 11.0 (mg/dL)
P value2 0.746 0.802 0.664 0.735 0.714 0.909 0.769 0.441 0.343 ,0.001 0.010 0.056 0.177 0.495 0.531 ,0.001 ,0.001 0.409
1 Because of one missing value in GCKR rs780094, the case number was 796. ALT, alanine aminotransferase; AST, aspartate aminotransferase; DBP, diastolic blood pressure; SBP, systolic blood pressure; gGT, g-glutamyltransferase. 2 Calculated for the comparisons between the 3 genotype groups by using ANOVA for continuous variables and a chi-square test for categorical variables. 3 Mean 6 SD (all such values).
Downloaded from ajcn.nutrition.org at MOUNT SAINT VINCENT UNIVERSITY LIBRARY on October 3, 2014
Age (y) Male [n (%)] Weight (kg) Height (cm) BMI (kg/m2) Waist (cm) Waist-to-hip ratio SBP (mm Hg) DBP (mm Hg) AST (IU/L) ALT (IU/L) gGT (IU/L) Fasting glucose (mg/dL) Fasting insulin (mU/mL) HOMA-IR Total cholesterol (mg/dL) Triglyceride (mg/dL) HDL cholesterol (mg/dL)
CC (n = 296)
GCKR rs738409 may confer a higher population-specific susceptibility to NAFLD in Asians than in Europeans and Africans. In agreement with our findings, Santoro et al (11) have shown that the functional GCKR rs1260326 SNP was associated with hepatic fat content in obese children of multiethnic groups (no Asians included). GCKR rs1260326 was in strong linkage disequilibrium with GCKR rs780094 in this study. The GCKR rs1260326 variant encodes the P446L protein, which deregulates glucose storage and disposal and facilitates de novo lipogenesis (16). Because carbohydrates are the upstream substrates in the glucokinase-regulated pathway, we speculated that the effect of GCKR variants on the development of NAFLD may be mediated by dietary sugar intake. If this hypothesis is true, we may prevent subjects with the GCKR variant from developing NAFLD by reducing dietary sugar consumption. Further studies are needed to verify this argument. The association between the GCKR variant and the ALT concentration is controversial in several studies. In this study, the multivariate analysis showed that the homozygote for the risk T allele in GCKR rs738409 significantly increased the serum ALT concentration by 8.64 IU/L as compared with the wild CC genotype. In adults, Hernaez et al (10) reported that GCKR rs780094 variants had no effect on ALT in Europeans, Africans, and Hispanics. In obese children, Santoro et al (11) found that the differences in ALT concentrations among groups stratified
873
GENETIC VARIANTS IN NAFLD TABLE 4 Multiple logistic regression analysis of predictive factors for NAFLD1 Variable
Coefficient estimate
SE
0.4931 0.2321 0.7378 0.4750 1.0340 0.3836 0.6918
0.2325 0.0267 0.1733 0.2132 0.2884 0.2371 0.2615
Male Adjusted BMI2 WHR, per 0.1 increase PNPLA3 rs738409 GC genotype3 PNPLA3 rs738409 GG genotype GCKR rs780094 TC genotype4 GCKR rs780094 TT genotype
OR (95% CI) 1.637 1.261 2.091 1.608 2.812 1.468 1.997
(1.038, (1.197, (1.489, (1.058, (1.598, (0.922, (1.196,
2.583) 1.329) 2.937) 2.442) 4.950) 2.336) 3.335)
P value 0.034 ,0.001 ,0.001 0.026 ,0.001 0.106 0.008
by GCKR rs1260326 genotype were significant only in Hispanics but not in Europeans or Africans. The genetic structure of the underlying population, sample size, and population-specific environmental interactions may account for such inconsistent results among studies. In concordance with the previous report (11), we observed a positive association between the GCKR rs780094 variant and a high serum triglyceride concentration. Furthermore, we observed an association between the GCKR rs780094 variant and a high serum cholesterol concentration. Our findings suggest that the GCKR rs780094 variant would increase the future cardiovascular disease risk in obese children and adolescents. In addition, our results show that male sex is a well-known risk factor for NAFLD. Our diagnostic modality for NAFLD was liver ultrasonography. A recent meta-analysis showed fairly good diagnostic accuracy and reliability of ultrasonography for the detection of moderate-severe fatty liver (15). In addition, we selected a relatively strict ultrasonographic score of $4 as the cutoff for NAFLD. Thus, some individuals with mild NAFLD might be TABLE 5 Multiple linear regression analysis of predictive factors for serum ALT concentration1 Variable Male Adjusted BMI2 WHR, per 0.1 increase PNPLA3 rs738409 GC genotype3 PNPLA3 rs738409 GG genotype GCKR rs780094 TC genotype4 GCKR rs780094 TT genotype
Coefficient estimate
SE
P value
8.3681 1.2330 5.3988 5.4282 12.4848 4.3725 8.6447
2.2665 0.2678 1.7947 2.1281 3.0726 2.3604 2.7093
,0.001 ,0.001 0.003 0.011 ,0.001 0.064 0.001
1 ALT, alanine aminotransferase; SNP, single nucleotide polymorphism; WHR, waist-to-hip ratio. 2 Adjusted BMI = (BMI of the case) – (age- and sex-specific median BMI in Taiwan). 3 The PNPLA3 rs738409 SNP has the GG, GC, and CC genotypes, whereas the CC genotype is the wild type and was taken as the reference group. 4 The GCKR rs780094 SNP has the CC, TC, and TT genotypes, whereas the CC genotype is the wild type and was taken as the reference group.
categorized as non-NAFLD. Such categorization would generally decrease the power to detect differences rather than bias toward a significant difference. In addition, we recognize the concern for multiple comparisons in the statistical analysis. Because the results of this study were mainly derived from the multivariate analysis, multiple testing adjustment of the significance of individual parameters in the regression model was not needed. In conclusion, we showed that the genetic variants in GCKR rs780094, but not in NCAN rs2228603, LYPLAL1 rs12137855, and PPP1R3B rs4240624, are associated with an increased risk of NAFLD independent of the effect of PNPLA3 rs738409 SNP in our population of obese Taiwanese children. It is known that genetic variants that have a significant association with disease in some studies in one population may not necessarily have the same association in another study. Therefore, our finding has important implications for global comparisons with the genetic data of different ethnic groups, including Europeans, Africans, and Hispanics. We are indebted to Kevin Liu for laboratory work, to Li-Chin Fan for help with the data collection, and to Chien-Hao Chen for assistance with the statistical analysis. The authors’ responsibilities were as follows—Y-CL and Y-HN: designed and conducted the research; Y-CL and P-FC: analyzed the data; Y-CL, M-HC, and Y-HN: wrote the manuscript; and Y-HN: had primary responsibility for the final content of the manuscript. All authors read and approved the final manuscript. None of the authors had a conflict of interest.
REFERENCES 1. Matteoni CA, Younossi ZM, Gramlich T, Boparai N, Liu YC, McCullough AJ. Nonalcoholic fatty liver disease: a spectrum of clinical and pathological severity. Gastroenterology 1999;116:1413–9. 2. Janssen I, Katzmarzyk PT, Boyce WF, Vereecken C, Mulvihill C, Roberts C, Currie C, Pickett W. Comparison of overweight and obesity prevalence in school-aged youth from 34 countries and their relationships with physical activity and dietary patterns. Obes Rev 2005;6: 123–32. 3. Romeo S, Kozlitina J, Xing C, Pertsemlidis A, Cox D, Pennacchio LA, Boerwinkle E, Cohen JC, Hobbs HH. Genetic variation in PNPLA3 confers susceptibility to nonalcoholic fatty liver disease. Nat Genet 2008;40:1461–5. 4. Sookoian S, Castano GO, Burgueno AL, Gianotti TF, Rosselli MS, Pirola CJ. A nonsynonymous gene variant in the adiponutrin gene is
Downloaded from ajcn.nutrition.org at MOUNT SAINT VINCENT UNIVERSITY LIBRARY on October 3, 2014
1 Logistic regression model: Pearson goodness-of-fit test (P = 0.5548 . 0.05) and Hosmer-Lemeshow goodness-of-fit test (P = 0.2963 . 0.05). NAFLD, nonalcoholic fatty liver disease; SNP, single nucleotide polymorphism; WHR, waist-tohip ratio. 2 Adjusted BMI = (BMI of the case) – (age- and sex-specific median BMI in Taiwan). 3 The PNPLA3 rs738409 SNP has the GG, GC, and CC genotypes, whereas the CC genotype is the wild type and was taken as the reference group. 4 The GCKR rs780094 SNP has the CC, TC, and TT genotypes, whereas the CC genotype is the wild type and was taken as the reference group.
874
5.
6. 7.
8.
10.
associated with nonalcoholic fatty liver disease severity. J Lipid Res 2009;50:2111–6. Goran MI, Walker R, Le KA, Mahurkar S, Vikman S, Davis JN, Spruijt-Metz D, Weigensberg MJ, Allayee H. Effects of PNPLA3 on liver fat and metabolic profile in Hispanic children and adolescents. Diabetes 2010;59:3127–30. Lin YC, Chang PF, Hu FC, Yang WS, Chang MH, Ni YH. A common variant in the PNPLA3 gene is a risk factor for non-alcoholic fatty liver disease in obese Taiwanese children. J Pediatr 2011;158:740–4. Speliotes EK, Yerges-Armstrong LM, Wu J, Hernaez R, Kim LJ, Palmer CD, Gudnason V, Eiriksdottir G, Garcia ME, Launer LJ, et al. Genome-wide association analysis identifies variants associated with nonalcoholic fatty liver disease that have distinct effects on metabolic traits. PLoS Genet 2011;7:e1001324. Browning JD, Szczepaniak LS, Dobbins R, Nuremberg P, Horton JD, Cohen JC, Grundy SM, Hobbs HH. Prevalence of hepatic steatosis in an urban population in the United States: impact of ethnicity. Hepatology 2004;40:1387–95. Palmer ND, Musani SK, Yerges-Armstrong LM, Feitosa MF, Bielak LF, Hernaez R, Kahali B, Carr JJ, Harris TB, Jhun MA, et al. Characterization of european ancestry nonalcoholic fatty liver diseaseassociated variants in individuals of African and Hispanic descent. Hepatology 2013;58:966–75. Hernaez R, McLean J, Lazo M, Brancati FL, Hirschhorn JN, Borecki IB, Harris TB, Genetics of Obesity-Related Liver Disease (GOLD)
11.
12. 13.
14. 15.
16.
Consortium, Nguyen T, Kamel IR, et al. Association between variants in or near PNPLA3, GCKR, and PPP1R3B with ultrasound-defined steatosis based on data from the third National Health and Nutrition Examination Survey. Clin Gastroenterol Hepatol 2013;11:1183–90 e2. Santoro N, Zhang CK, Zhao H, Pakstis AJ, Kim G, Kursawe R, Dykas DJ, Bale AE, Giannini C, Pierpont B, et al. Variant in the glucokinase regulatory protein (GCKR) gene is associated with fatty liver in obese children and adolescents. Hepatology 2012;55:781–9. Chen W, Chang MH. New growth charts for Taiwanese children and adolescents based on World Health Organization standards and healthrelated physical fitness. Pediatr Neonatol 2010;51:69–79. Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, Turner RC. Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia 1985;28:412–9. Chan DF, Li AM, Chu WC, Chan MH, Wong EM, Liu EK, Chan IH, Yin J, Lam CW, Fok TF, et al. Hepatic steatosis in obese Chinese children. Int J Obes Relat Metab Disord 2004;28:1257–63. Hernaez R, Lazo M, Bonekamp S, Kamel I, Brancati FL, Guallar E, Clark JM. Diagnostic accuracy and reliability of ultrasonography for the detection of fatty liver: a meta-analysis. Hepatology 2011;54:1082– 90. Rees MG, Wincovitch S, Schultz J, Waterstradt R, Beer NL, Baltrusch S, Collins FS, Gloyn AL. Cellular characterisation of the GCKR P446L variant associated with type 2 diabetes risk. Diabetologia 2012;55:114–22.
Downloaded from ajcn.nutrition.org at MOUNT SAINT VINCENT UNIVERSITY LIBRARY on October 3, 2014
9.
LIN ET AL
INTRODUCTION
Concurrent with the epidemic of childhood obesity, pediatric nonalcoholic fatty liver disease (NAFLD)4 is an important public health issue (1, 2). The fact that not all obese children develop NAFLD suggests that environmental and/or genetic factors may influence the susceptibility of each obese individual. The first genome-wide association study (GWAS) of NAFLD from the Dallas Heart Study identified the rs738409 single
nucleotide polymorphism (SNP) in patatin-like phospholipase domain-containing-3 (PNPLA3) to be strongly associated with liver fat content (3). Subsequent studies have confirmed this association (4–6). A recent GWAS from The Genetics of Obesity-Related Liver Disease Consortium identified 4 additional genetic variants in rs2228603, rs12137855, rs780094, and rs4240624 for NAFLD in 7177 adults of European ancestry (7). The 4 newly identified susceptibility loci were located in or near neurocan (NCAN), lysophospholipase-like 1 (LYPLAL1), glucokinase regulatory protein (GCKR), and protein phosphatase 1 regulatory subunit 3b (PPP1R3B). The prevalence of NAFLD varies among different ethnic groups (8). Two recent studies attempted to verify the findings of the Genetics of Obesity-Related Liver Disease Consortium in multiethnic cohorts. Palmer et al (9) measured hepatic steatosis by using computed tomography in adult individuals of African and Hispanic ethnicities. The results showed that the allele frequency and effect size of PNPLA3 rs738409, NCAN rs2228603, LYPLAL1 rs12137855, GCKR rs780094, and PPP1R3B rs4240624 varied in different ethnic groups. Hernaez et al (10) also reported a lack of consistency in these variants in the NHANES III study population of multiethnicities. Although these studies extended the initial findings in European adults to Africans and Hispanics, there is no report in Asians. Because the effect size of genetic variants tends to be more pronounced in children than in adults because of less confounding factors, such as lifestyle habits and comorbidities (11), it would be
1
From the Department of Pediatrics, Far Eastern Memorial Hospital, Taipei, Taiwan (Y-CL and P-FC); the Departments of Pediatrics, National Taiwan University Hospital, Taipei, Taiwan (M-HC and Y-HN); and the Oriental Institute of Technology, New Taipei City, Taiwan (Y-CL). 2 Supported by research grants from the Far Eastern Memorial Hospital (FEMH-2013-C-023), the Far Eastern Memorial Hospital–National Taiwan University Hospital Joint Research Program (101-FTN-07), and the National Science Council, Taiwan (NSC-98-2628-B-002-006-MY3). 3 Address correspondence to Y-H Ni, Department of Pediatrics, National Taiwan University Hospital, No. 8, Zhongshan South Road, Zhongzheng District, Taipei City 10041, Taiwan. E-mail: [email protected]. 4 Abbreviations used: ALT, alanine aminotransferase; AST, aspartate aminotransferase; GWAS, genome-wide association study; NAFLD, nonalcoholic fatty liver disease; SNP, single nucleotide polymorphism; WHR, waist-to-hip ratio. Received November 10, 2013. Accepted for publication January 6, 2014. First published online January 29, 2014; doi: 10.3945/ajcn.113.079749.
Am J Clin Nutr 2014;99:869–74. Printed in USA. Ó 2014 American Society for Nutrition
869
Downloaded from ajcn.nutrition.org at MOUNT SAINT VINCENT UNIVERSITY LIBRARY on October 3, 2014
ABSTRACT Background: A genome-wide association study identified variants in or near patatin-like phospholipase domain-containing-3 (PNPLA3), neurocan (NCAN), lysophospholipase-like 1 (LYPLAL1), glucokinase regulatory protein (GCKR), and protein phosphatase 1 regulatory subunit 3b (PPP1R3B) that were strongly associated with nonalcoholic fatty liver disease (NAFLD) in adults of European ancestry. Objective: We examined these genetic variants in obese children and tested whether their effects on NAFLD are significant in the Taiwanese Han Chinese population. Design: We genotyped PNPLA3 rs738409, NCAN rs2228603, LYPLAL1 rs12137855, GCKR rs780094, and PPP1R3B rs4240624 in 797 obese children aged 7–18 y. NAFLD was identified by liver ultrasonography. We analyzed the effect of these genetic variants on NAFLD. Results: NAFLD was identified in 24% of the recruited obese children. We found significant associations with NAFLD at variants in PNPLA3 and GCKR but not in NCAN, LYPLAL1, and PPP1R3B. Multiple logistic regression analysis showed that, after control for the effects of age- and sex-adjusted body mass index, waist-to-hip ratio, sex, and PNPLA3 rs738409 polymorphism, the variant GCKR rs780094 TT genotype independently increased the OR of NAFLD by 1.997 (95% CI: 1.196, 3.335; P = 0.008) compared with the CC genotype. Subjects with the variant GCKR rs780094 TT genotype had a higher mean serum alanine aminotransferase concentration than did those with the CC genotype (30.8 6 34.7 compared with 22.2 6 18.6 IU/L; P = 0.01). Conclusions: By studying the genetic variants of obese Taiwanese children, we confirmed that the genetic variants in GCKR rs780094 and PNPLA3 rs738409, but not in NCAN rs2228603, LYPLAL1 rs12137855, and PPP1R3B rs4240624, are associated with an increased risk of NAFLD. GCKR and PNPLA3 variants are the common genetic factors that may confer susceptibility to NAFLD in obese individuals across multiple ethnic groups. This trial was registered at clinicaltrials.gov as NCT00274183. Am J Clin Nutr 2014;99:869–74.
870
LIN ET AL
advantageous to identify these genetic variants in children. Herein we aimed to investigate the effects of genetic variants in NCAN rs2228603, LYPLAL1 rs12137855, GCKR rs780094, and PPP1R3B rs4240624 in obese Taiwanese children.
SUBJECTS AND METHODS
Subjects
Data collection The following data were obtained from each subject: age, sex, BMI, and waist and hip circumferences. BMI was calculated as body weight (kg)/height (m)2. The ratio between waist and hip circumferences was calculated and recorded as the waist-to-hip ratio (WHR). None of the subjects had significant alcohol use. Environmental factors, such as diet and physical activity, were not recorded in this study. In fasting venous blood samples, we measured total serum bilirubin, alanine aminotransferase (ALT), aspartate aminotransferase (AST), g-glutamyltransferase, fasting glucose, triglycerides, total cholesterol, and HDL cholesterol. Insulin concentration was measured by using a chemiluminescence immunoassay (Diagnostic Products Corporation). Insulin resistance was measured by HOMA-IR and was calculated as follows: HOMA-IR = [fasting insulin (mU/mL)][fasting glucose (mmol/L)]/22.5 (13). For the subjects with abnormal ALT concentrations, we ruled out causes of hepatitis other than NAFLD by measuring hepatitis B surface antigen, anti-hepatitis C antibody, antinuclear antibody, and ceruloplasmin.
Liver ultrasonography All participants underwent liver ultrasonographic examination performed by one operator. The hand-carried machine used (TITAN; SonoSite Ltd) was equipped with a 2–5-MHz convex probe. NAFLD was defined as the presence of an ultrasonographic pattern consistent with the following criteria: liver-kidney echo discrepancy, attenuated echo penetration, visibility of diaphragm, and obscure hepatic vessel structures. The aforementioned ultrasonographic pattern was scored as described by Chan et al (14). These children were categorized to have mild, moderate, or severe steatosis if the overall score was 1–3, 4–6, or 7–9, respectively. Because ultrasonography is much more accurate at detecting moderate-severe steatosis than mild steatosis (15), we set a score $4 as the diagnostic criterion for NAFLD to avoid possible false-positive results.
Genomic DNA was extracted from venous blood from each participant by using a commercial kit (Puregene; Gentra Systems) according to the manufacturer’s instructions. After extraction, the genomic DNA was immediately stored at 2808C. The TaqMan SNP genotyping assays C_7241_10 for PNPLA3 rs738409, C_16171492_10 for NCAN rs2228603, C_31403184_10 for LYPLAL1 rs12137855, C_2862873_10 for GCKR rs780094, and C_1614473_10 for PPP1R3B rs4240624 SNPs (Applied Biosystems) were performed on an ABI 7300 Real-Time PCR System (Applied Biosystems). The allele frequencies of SNP genotypes were tested for the Hardy-Weinberg equilibrium. Statistical analysis The statistical analysis was performed by using Stata statistical software (version 11.2; StataCorp). A 2-sided P value #0.05 was considered statistically significant. The standardized BMI z scores were not available for children in Taiwan. To adjust for the intrinsic effects of age and sex on BMI, we derived a variable as “adjusted BMI” by subtracting the Taiwanese children’s age- and sex-specific population median BMI from each subject’s BMI value. In univariate analysis, the chi-square test and 1-factor ANOVA were used for the comparisons of categorical and continuous variables. Multivariate analysis was conducted to identify predictive factors of pediatric NAFLD and serum ALT concentrations by fitting multiple logistic and linear regression models, respectively. To ensure the quality of the results, we used stepwise procedures for variable selection, goodness-of-fit assessment, and regression diagnostics in our regression analyses. The SNP genotypes were coded as a categorical variable as follows: 0 = 2 major alleles (the reference group), 1 = 1 minor allele + 1 major allele (heterozygous), and 3 = 2 minor alleles (homozygous). The best final regression model was identified manually by reducing the significance levels to 0.05 corresponding to the chosen a level. RESULTS
Genotype distribution of PNPLA3 rs738409, NCAN rs2228603, LYPLAL1 rs12137855, GCKR rs780094, and PPP1R3B rs4240624 Of the 797 enrolled obese subjects, 191 (24%) had an ultrasonographic score $4 and received a diagnosis of NAFLD. The genotype distributions of these SNPs between the NAFLD and non-NAFLD groups are shown in Table 1. The numbers of missing values were 0 for PNPLA3 rs738409, 0 for NCAN rs2228603, 3 for LYPLAL1 rs12137855, 1 for GCKR rs780094, and 6 for PPP1R3B rs4240624. Correspondingly, call rates for each SNP genotyping were 100%, 100%, 99.6%, 99.9%, and 99.3%, respectively. Two of 5 variants, PNPLA3 rs738409 and GCKR rs780094, were significantly associated with NAFLD. The percentage of PNPLA3 rs738409 GG genotypes was significantly higher in the subjects with NAFLD than in those without NAFLD (OR: 2.079; P = 0.003). The GCKR rs780094 TT genotype was more prevalent in the subjects with NAFLD (OR: 1.733; P = 0.018). The frequencies of the G allele of rs738409 at PNPLA3 and the T allele of rs780094 at GCKR were significantly higher in the
Downloaded from ajcn.nutrition.org at MOUNT SAINT VINCENT UNIVERSITY LIBRARY on October 3, 2014
From October 2006 to September 2012, 797 obese children aged 7–18 y were enrolled from schools in Taipei, Taiwan. All subjects were unrelated and ethnically Han Chinese. Obesity was defined as a BMI greater than the 95th percentile by different age and sex groups according to the standards of the Department of Health in Taiwan (12). The study was performed according to the principles of the Declaration of Helsinki. All parents gave their informed consent, and the study was approved by the ethics committee of the Far Eastern Memorial Hospital and National Taiwan University Hospital.
Genotyping
871
GENETIC VARIANTS IN NAFLD TABLE 1 Genotype and minor allele frequencies of PNPLA3 rs738409, NCAN rs2228603, LYPLAL1 rs12137855, GCKR rs780094, and PPP1R3B rs42406241 Genotype
Non-NAFLD
P value2
OR (95% CI)
58 (30) 95 (50) 38 (20) 0.448
238 (39.3) 293 (48.3) 75 (12.4) 0.366
0.129 0.003 0.004
1.330 (0.921, 1.921) 2.079 (1.284, 3.368) 1.407 (1.115, 1.776)
168 (88) 23 (12) 0 (0) 0.060
532 (87.8) 74 (12.2) 0 (0) 0.061
0.950 — 0.985
0.984 (0.600, 1.615) — 0.985 (0.611, 1.590)
161 (85.6) 27 (14.4) 0 (0%) 0.072
525 (86.6) 78 (12.9) 3 (0.5%) 0.069
0.129 — 0.868
1.129 (0.706, 1.804) — 1.039 (0.665, 1.624)
41 (21.5) 91 (48.0) 58 (30.5) 0.545
174 (28.7) 290 (47.9) 142 (23.4) 0.474
0.174 0.018 0.016
1.332 (0.882, 2.010) 1.733 (1.100, 2.733) 1.330 (1.056, 1.676)
182 (96.8) 6 (3.2) 0 (0) 0.016
592 (98.2) 11 (1.8) 0 (0) 0.009
0.259 — 0.2612
1.774 (0.671, 4.694) — 1.762 (0.671, 4.625)
1
The numbers of missing values were 0 for PNPLA3 rs738409, 0 for NCAN rs2228603, 3 for LYPLAL1 rs12137855, 1 for GCKR rs780094, and 6 for PPP1R3B rs4240624. MAF, minor allele frequency; NAFLD, nonalcoholic fatty liver disease. 2 As compared with the reference genotype group, the differences in genotype frequencies were analyzed by using a chi-square test. P , 0.05 was considered to be statistically significant.
subjects with NAFLD (P = 0.004 for rs738409 and P = 0.016 for rs780094). In contrast, the minor allele frequencies of NCAN rs2228603, LYPLAL1 rs12137855, GCKR rs780094, and PPP1R3B rs4240624 were low and not different between the NAFLD and non-NAFLD groups. The allele frequencies of all SNP genotypes were in Hardy-Weinberg equilibrium in both the NAFLD and non-NAFLD groups (all P . 0.05). Subject characteristics stratified by PNPLA3 rs738409 and GCKR rs780094 genotypes The basic characteristics of the study subjects stratified by PNPLA3 rs738409 genotype are shown in Table 2: 296 with CC genotype, 388 with CG genotype, and 113 with GG genotype. We found significant differences in ALT and AST concentrations. The mean (6SD) serum ALT concentration was 34.2 6 36.3 IU/L for the GG genotype (highest), 27.7 6 31.8 IU/L for the CG genotype, and 22.36 18.5 IU/L for the CC genotype (lowest) (P , 0.001). The basic characteristics of the study subjects stratified by GCKR rs780094 genotype are shown in Table 3: 215 with the CC genotype, 381 with the CT genotype, and 200 with the TT genotype. We found statistically significant differences in AST, ALT, total cholesterol, and triglyceride concentrations. The mean serum ALT concentration was 30.8 IU/L for the TT genotype (highest), 26.9 IU/L for the TC genotype, and 22.2 IU/L
for the CC genotype (lowest) (P = 0.01). Subjects carrying the T allele of GCKR rs780094 had higher serum total cholesterol and triglyceride concentrations (P , 0.001). Multivariate analysis of predicting factors for NAFLD and serum ALT concentration As shown in Table 4, the fitted final multiple logistic regression model revealed that the variant PNPLA3 rs738409 and GCKR rs780094 genotypes were independent risk factors of NAFLD after control for the effects of sex, adjusted BMI, and WHR. This final multiple logistic regression model was obtained after including age, sex, adjusted BMI, WHR, HOMA-IR, PNPLA3 rs738409, NCAN rs2228603, LYPLAL1 rs12137855, GCKR rs780094, and PPP1R3B rs4240624 genotypes in the stepwise variable selection procedure. Specifically, when compared with the wild type, the subjects with the PNPLA3 rs738409 GC genotype would have a 1.608 (95% CI: 1.058, 2.442; P = 0.026) times higher odds of having NAFLD, and those with the GG genotype would have a 2.812 (95% CI: 1.598, 4.950; P , 0.001) times higher odds of having NAFLD, whereas the values of the other covariates were fixed. The subjects carrying the GCKR rs780094 TT genotype would have a 1.997 (95% CI: 1.196, 3.335; P = 0.008) times higher odds of having NAFLD. No significant difference was found for the GCKR rs780094 TC genotype.
Downloaded from ajcn.nutrition.org at MOUNT SAINT VINCENT UNIVERSITY LIBRARY on October 3, 2014
PNPLA3 rs738409 [n (%)] CC GC GG MAF (G allele) NCAN rs2228603 [n (%)] CC TC TT MAF (T allele) LYPLAL1 rs12137855 [n (%)] CC TC TT MAF (T allele) GCKR rs780094 [n (%)] CC CT TT MAF (T allele) PPP1R3B rs4240624 [n (%)] AA GA GG MAF (G allele)
NAFLD
872
LIN ET AL
TABLE 2 Basic characteristics of the subjects across PNPLA3 rs738409 genotypes1 Variable
CG (n = 388)
GG (n = 113)
11.7 6 1.93 11.7 6 1.9 11.5 6 2.0 203 (68.6) 273 (70.4) 79 (69.9) 65.7 6 17.1 64.8 6 15.9 64.2 6 16.9 154.1 6 13.2 153.3 6 12.3 153.7 6 13.7 27.2 6 3.9 27.1 6 3.5 26.7 6 3.8 89.0 6 10.7 88.9 6 9.4 87.6 6 11.4 0.90 6 0.06 0.90 6 0.05 0.90 6 0.06 118.6 6 13.8 119.8 6 15.0 116.6 6 11.8 73.6 6 10.3 74.2 6 11.4 75.1 6 10.7 22.0 6 8.7 25.0 6 14.5 27.7 6 16.8 22.3 6 18.5 27.7 6 31.8 34.2 6 36.3 17.4 6 11.7 19.2 6 17.9 20.3 6 19.4 86.0 6 6.8 86.6 6 13.3 86.8 6 9.9 17.6 6 20.6
17.1 6 21.0
P value2 0.723 0.833 0.632 0.752 0.424 0.455 0.863 0.097 0.468 ,0.001 ,0.001 0.190 0.687
16.2 6 16.4
0.815
3.82 6 4.93 3.70 6 4.77 3.48 6 3.63 158.9 6 32.7 157.3 6 34.1 158.1 6 39.1
0.802 0.839
101.7 6 54.2
93.7 6 44.3
92.6 6 46.0
0.066
46.8 6 11.6
46.6 6 10.6
47.0 6 12.3
0.938
1
ALT, alanine aminotransferase; AST, aspartate aminotransferase; DBP, diastolic blood pressure; SBP, systolic blood pressure; gGT, g-glutamyltransferase. 2 Calculated for the comparisons between the 3 genotype groups by using ANOVA for continuous variables and a chi-square test for categorical variables. 3 Mean 6 SD (all such values).
Similarly, the fitted final linear regression model showed a significant association of variant PNPLA3 rs738409 with the GCKR rs780094 genotypes with a higher mean serum ALT concentration (Table 5). After adjustment for the effects of the other covariates, the PNPLA3 rs738409 GC genotype would increase the mean serum ALT concentration by 5.43 IU/L (P = 0.011) and the GG genotype would increase it by 12.48 IU/L (P , 0.001) compared with the wild CC genotype. The mean ALT concentration of the subjects carrying the GCKR rs780094 TT genotype was 8.64 IU/L higher than that of the reference group (P = 0.001).
DISCUSSION
We found that genetic variants in PNPLA3 rs738409 and GCKR rs780094 conferred susceptibility to NAFLD in our obese schoolchildren and adolescents. To our knowledge, this is the first study to show the effect of GCKR variants on NAFLD in a pediatric Asian population. In our obese Taiwanese children and adolescents, the homozygote for the risk T allele in GCKR rs738409 has a 73% higher OR of NAFLD risk. The effect of this variant was similar in all ethnic groups, but the minor T allele frequency was lowest in Africans (0.18–0.2), intermediate in Europeans (0.39–0.41) and Hispanics (0.35–0.38), and highest in Han Chinese (0.49 in this study) (7, 9, 10). Thus, we estimated that the risk T allele in
TABLE 3 Basic characteristics of the subjects across GCKR rs780094 genotypes1 Variable
CC (n = 215)
CT (n = 381)
TT (n = 200)
Age (y) 11.7 6 1.83 11.7 6 1.9 11.6 6 2.0 Male [n (%)] 152 (70.7) 268 (70.3) 136 (68.0) Weight (kg) 65.7 6 15.8 64.8 6 16.3 64.6 6 17.2 Height (cm) 154.0 6 12.3 153.7 6 12.9 153.1 6 13.4 BMI (kg/m2) 27.3 6 3.6 27.0 6 3.8 27.0 6 3.7 Waist (cm) 88.9 6 10.1 88.7 6 10.0 88.6 6 10.6 Waist-to-hip ratio 0.90 6 0.06 0.90 6 0.06 0.90 6 0.06 SBP (mm Hg) 118.7 6 13.7 118.5 6 14.4 120.0 6 14.2 DBP (mm Hg) 74.1 6 11.8 74.6 6 10.6 73.2 6 10.5 AST (IU/L) 22.4 6 8.3 24.2 6 12.7 26.3 6 17.5 ALT (IU/L) 22.2 6 18.6 26.9 6 29.4 30.8 6 34.7 gGT (IU/L) 17.3 6 16.7 18.2 6 15.7 21.0 6 16.3 Fasting glucose 87.3 6 10.3 86.4 6 12.4 85.4 6 7.5 (mg/dL) Fasting insulin 17.4 6 19.3 16.3 6 17.2 18.4 6 25.9 (mU/mL) HOMA-IR 3.83 6 4.67 3.50 6 4.08 3.92 6 5.67 Total cholesterol 153.8 6 31.8 156.3 6 34.8 165.7 6 35.4 (mg/dL) Triglyceride 87.0 6 40.7 94.1 6 46.3 111.3 6 56.7 (mg/dL) HDL cholesterol 45.9 6 9.9 46.7 6 12.2 47.4 6 11.0 (mg/dL)
P value2 0.746 0.802 0.664 0.735 0.714 0.909 0.769 0.441 0.343 ,0.001 0.010 0.056 0.177 0.495 0.531 ,0.001 ,0.001 0.409
1 Because of one missing value in GCKR rs780094, the case number was 796. ALT, alanine aminotransferase; AST, aspartate aminotransferase; DBP, diastolic blood pressure; SBP, systolic blood pressure; gGT, g-glutamyltransferase. 2 Calculated for the comparisons between the 3 genotype groups by using ANOVA for continuous variables and a chi-square test for categorical variables. 3 Mean 6 SD (all such values).
Downloaded from ajcn.nutrition.org at MOUNT SAINT VINCENT UNIVERSITY LIBRARY on October 3, 2014
Age (y) Male [n (%)] Weight (kg) Height (cm) BMI (kg/m2) Waist (cm) Waist-to-hip ratio SBP (mm Hg) DBP (mm Hg) AST (IU/L) ALT (IU/L) gGT (IU/L) Fasting glucose (mg/dL) Fasting insulin (mU/mL) HOMA-IR Total cholesterol (mg/dL) Triglyceride (mg/dL) HDL cholesterol (mg/dL)
CC (n = 296)
GCKR rs738409 may confer a higher population-specific susceptibility to NAFLD in Asians than in Europeans and Africans. In agreement with our findings, Santoro et al (11) have shown that the functional GCKR rs1260326 SNP was associated with hepatic fat content in obese children of multiethnic groups (no Asians included). GCKR rs1260326 was in strong linkage disequilibrium with GCKR rs780094 in this study. The GCKR rs1260326 variant encodes the P446L protein, which deregulates glucose storage and disposal and facilitates de novo lipogenesis (16). Because carbohydrates are the upstream substrates in the glucokinase-regulated pathway, we speculated that the effect of GCKR variants on the development of NAFLD may be mediated by dietary sugar intake. If this hypothesis is true, we may prevent subjects with the GCKR variant from developing NAFLD by reducing dietary sugar consumption. Further studies are needed to verify this argument. The association between the GCKR variant and the ALT concentration is controversial in several studies. In this study, the multivariate analysis showed that the homozygote for the risk T allele in GCKR rs738409 significantly increased the serum ALT concentration by 8.64 IU/L as compared with the wild CC genotype. In adults, Hernaez et al (10) reported that GCKR rs780094 variants had no effect on ALT in Europeans, Africans, and Hispanics. In obese children, Santoro et al (11) found that the differences in ALT concentrations among groups stratified
873
GENETIC VARIANTS IN NAFLD TABLE 4 Multiple logistic regression analysis of predictive factors for NAFLD1 Variable
Coefficient estimate
SE
0.4931 0.2321 0.7378 0.4750 1.0340 0.3836 0.6918
0.2325 0.0267 0.1733 0.2132 0.2884 0.2371 0.2615
Male Adjusted BMI2 WHR, per 0.1 increase PNPLA3 rs738409 GC genotype3 PNPLA3 rs738409 GG genotype GCKR rs780094 TC genotype4 GCKR rs780094 TT genotype
OR (95% CI) 1.637 1.261 2.091 1.608 2.812 1.468 1.997
(1.038, (1.197, (1.489, (1.058, (1.598, (0.922, (1.196,
2.583) 1.329) 2.937) 2.442) 4.950) 2.336) 3.335)
P value 0.034 ,0.001 ,0.001 0.026 ,0.001 0.106 0.008
by GCKR rs1260326 genotype were significant only in Hispanics but not in Europeans or Africans. The genetic structure of the underlying population, sample size, and population-specific environmental interactions may account for such inconsistent results among studies. In concordance with the previous report (11), we observed a positive association between the GCKR rs780094 variant and a high serum triglyceride concentration. Furthermore, we observed an association between the GCKR rs780094 variant and a high serum cholesterol concentration. Our findings suggest that the GCKR rs780094 variant would increase the future cardiovascular disease risk in obese children and adolescents. In addition, our results show that male sex is a well-known risk factor for NAFLD. Our diagnostic modality for NAFLD was liver ultrasonography. A recent meta-analysis showed fairly good diagnostic accuracy and reliability of ultrasonography for the detection of moderate-severe fatty liver (15). In addition, we selected a relatively strict ultrasonographic score of $4 as the cutoff for NAFLD. Thus, some individuals with mild NAFLD might be TABLE 5 Multiple linear regression analysis of predictive factors for serum ALT concentration1 Variable Male Adjusted BMI2 WHR, per 0.1 increase PNPLA3 rs738409 GC genotype3 PNPLA3 rs738409 GG genotype GCKR rs780094 TC genotype4 GCKR rs780094 TT genotype
Coefficient estimate
SE
P value
8.3681 1.2330 5.3988 5.4282 12.4848 4.3725 8.6447
2.2665 0.2678 1.7947 2.1281 3.0726 2.3604 2.7093
,0.001 ,0.001 0.003 0.011 ,0.001 0.064 0.001
1 ALT, alanine aminotransferase; SNP, single nucleotide polymorphism; WHR, waist-to-hip ratio. 2 Adjusted BMI = (BMI of the case) – (age- and sex-specific median BMI in Taiwan). 3 The PNPLA3 rs738409 SNP has the GG, GC, and CC genotypes, whereas the CC genotype is the wild type and was taken as the reference group. 4 The GCKR rs780094 SNP has the CC, TC, and TT genotypes, whereas the CC genotype is the wild type and was taken as the reference group.
categorized as non-NAFLD. Such categorization would generally decrease the power to detect differences rather than bias toward a significant difference. In addition, we recognize the concern for multiple comparisons in the statistical analysis. Because the results of this study were mainly derived from the multivariate analysis, multiple testing adjustment of the significance of individual parameters in the regression model was not needed. In conclusion, we showed that the genetic variants in GCKR rs780094, but not in NCAN rs2228603, LYPLAL1 rs12137855, and PPP1R3B rs4240624, are associated with an increased risk of NAFLD independent of the effect of PNPLA3 rs738409 SNP in our population of obese Taiwanese children. It is known that genetic variants that have a significant association with disease in some studies in one population may not necessarily have the same association in another study. Therefore, our finding has important implications for global comparisons with the genetic data of different ethnic groups, including Europeans, Africans, and Hispanics. We are indebted to Kevin Liu for laboratory work, to Li-Chin Fan for help with the data collection, and to Chien-Hao Chen for assistance with the statistical analysis. The authors’ responsibilities were as follows—Y-CL and Y-HN: designed and conducted the research; Y-CL and P-FC: analyzed the data; Y-CL, M-HC, and Y-HN: wrote the manuscript; and Y-HN: had primary responsibility for the final content of the manuscript. All authors read and approved the final manuscript. None of the authors had a conflict of interest.
REFERENCES 1. Matteoni CA, Younossi ZM, Gramlich T, Boparai N, Liu YC, McCullough AJ. Nonalcoholic fatty liver disease: a spectrum of clinical and pathological severity. Gastroenterology 1999;116:1413–9. 2. Janssen I, Katzmarzyk PT, Boyce WF, Vereecken C, Mulvihill C, Roberts C, Currie C, Pickett W. Comparison of overweight and obesity prevalence in school-aged youth from 34 countries and their relationships with physical activity and dietary patterns. Obes Rev 2005;6: 123–32. 3. Romeo S, Kozlitina J, Xing C, Pertsemlidis A, Cox D, Pennacchio LA, Boerwinkle E, Cohen JC, Hobbs HH. Genetic variation in PNPLA3 confers susceptibility to nonalcoholic fatty liver disease. Nat Genet 2008;40:1461–5. 4. Sookoian S, Castano GO, Burgueno AL, Gianotti TF, Rosselli MS, Pirola CJ. A nonsynonymous gene variant in the adiponutrin gene is
Downloaded from ajcn.nutrition.org at MOUNT SAINT VINCENT UNIVERSITY LIBRARY on October 3, 2014
1 Logistic regression model: Pearson goodness-of-fit test (P = 0.5548 . 0.05) and Hosmer-Lemeshow goodness-of-fit test (P = 0.2963 . 0.05). NAFLD, nonalcoholic fatty liver disease; SNP, single nucleotide polymorphism; WHR, waist-tohip ratio. 2 Adjusted BMI = (BMI of the case) – (age- and sex-specific median BMI in Taiwan). 3 The PNPLA3 rs738409 SNP has the GG, GC, and CC genotypes, whereas the CC genotype is the wild type and was taken as the reference group. 4 The GCKR rs780094 SNP has the CC, TC, and TT genotypes, whereas the CC genotype is the wild type and was taken as the reference group.
874
5.
6. 7.
8.
10.
associated with nonalcoholic fatty liver disease severity. J Lipid Res 2009;50:2111–6. Goran MI, Walker R, Le KA, Mahurkar S, Vikman S, Davis JN, Spruijt-Metz D, Weigensberg MJ, Allayee H. Effects of PNPLA3 on liver fat and metabolic profile in Hispanic children and adolescents. Diabetes 2010;59:3127–30. Lin YC, Chang PF, Hu FC, Yang WS, Chang MH, Ni YH. A common variant in the PNPLA3 gene is a risk factor for non-alcoholic fatty liver disease in obese Taiwanese children. J Pediatr 2011;158:740–4. Speliotes EK, Yerges-Armstrong LM, Wu J, Hernaez R, Kim LJ, Palmer CD, Gudnason V, Eiriksdottir G, Garcia ME, Launer LJ, et al. Genome-wide association analysis identifies variants associated with nonalcoholic fatty liver disease that have distinct effects on metabolic traits. PLoS Genet 2011;7:e1001324. Browning JD, Szczepaniak LS, Dobbins R, Nuremberg P, Horton JD, Cohen JC, Grundy SM, Hobbs HH. Prevalence of hepatic steatosis in an urban population in the United States: impact of ethnicity. Hepatology 2004;40:1387–95. Palmer ND, Musani SK, Yerges-Armstrong LM, Feitosa MF, Bielak LF, Hernaez R, Kahali B, Carr JJ, Harris TB, Jhun MA, et al. Characterization of european ancestry nonalcoholic fatty liver diseaseassociated variants in individuals of African and Hispanic descent. Hepatology 2013;58:966–75. Hernaez R, McLean J, Lazo M, Brancati FL, Hirschhorn JN, Borecki IB, Harris TB, Genetics of Obesity-Related Liver Disease (GOLD)
11.
12. 13.
14. 15.
16.
Consortium, Nguyen T, Kamel IR, et al. Association between variants in or near PNPLA3, GCKR, and PPP1R3B with ultrasound-defined steatosis based on data from the third National Health and Nutrition Examination Survey. Clin Gastroenterol Hepatol 2013;11:1183–90 e2. Santoro N, Zhang CK, Zhao H, Pakstis AJ, Kim G, Kursawe R, Dykas DJ, Bale AE, Giannini C, Pierpont B, et al. Variant in the glucokinase regulatory protein (GCKR) gene is associated with fatty liver in obese children and adolescents. Hepatology 2012;55:781–9. Chen W, Chang MH. New growth charts for Taiwanese children and adolescents based on World Health Organization standards and healthrelated physical fitness. Pediatr Neonatol 2010;51:69–79. Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, Turner RC. Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia 1985;28:412–9. Chan DF, Li AM, Chu WC, Chan MH, Wong EM, Liu EK, Chan IH, Yin J, Lam CW, Fok TF, et al. Hepatic steatosis in obese Chinese children. Int J Obes Relat Metab Disord 2004;28:1257–63. Hernaez R, Lazo M, Bonekamp S, Kamel I, Brancati FL, Guallar E, Clark JM. Diagnostic accuracy and reliability of ultrasonography for the detection of fatty liver: a meta-analysis. Hepatology 2011;54:1082– 90. Rees MG, Wincovitch S, Schultz J, Waterstradt R, Beer NL, Baltrusch S, Collins FS, Gloyn AL. Cellular characterisation of the GCKR P446L variant associated with type 2 diabetes risk. Diabetologia 2012;55:114–22.
Downloaded from ajcn.nutrition.org at MOUNT SAINT VINCENT UNIVERSITY LIBRARY on October 3, 2014
9.
LIN ET AL
