SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Only 20% of patients with chronic hepatitis C (CHC) will develop cirrhosis, and fibrosis progression remains highly unpredictable. A recent genome-wide association study identified a genetic variant in the patatin-like phospholipase-3 (PNPLA3) gene (rs738409 C>G) associated with steatosis that was further demonstrated to influence severity of fibrosis in nonalcoholic fatty liver disease. The aim of this study was to assess the impact of this polymorphism on histological liver damage and response to antiviral therapy in CHC. We recruited 537 Caucasian CHC patients from three European centers (Brussels, Belgium [n = 229]; Hannover, Germany [n = 171]; Lyon, France [n = 137]); these patients were centrally genotyped for the PNPLA3 (rs738409 C>G) polymorphism. We studied the influence of rs738409 and other variants in the PNPLA3 region on steatosis and fibrosis assessed both in a cross-sectional and longitudinal manner. Seven other variants previously associated with fibrosis progression were included. Finally, we explored the impact of rs738409 on response to standard antiviral therapy using the interferon lambda 3 (IL28B) [rs12979860 C>T] variant both as a comparator and as a positive control. After adjustment for age, sex, body mass index, alcohol consumption, and diabetes, rs738409 mutant G allele homozygote carriers remained at higher risk for steatosis (odds ratio [OR] 2.55, 95% confidence interval [CI] 1.08-6.03, P = 0.034), fibrosis (OR 3.13, 95% CI 1.50-6.51, P = 0.002), and fibrosis progression (OR 2.64, 95% CI 1.22-5.67, P = 0.013). Conversely, rs738409 was not independently associated with treatment failure (OR 1.07, 95% CI 0.46-2.49, P = 0.875) and did not influence clinical or biological variables. Conclusion: The PNPLA3 (rs738409 C>G) polymorphism favors steatosis and fibrosis progression in CHC. This polymorphism may represent a valuable genetic predictor and a potential therapeutic target in CHC liver damage. (HEPATOLOGY 2011;)

Hepatitis C virus (HCV) is a major health burden with 130 to 170 million people infected, representing nearly 3% of the worldwide population.1 HCV infection progresses to chronicity in ≈80% of cases,2 and chronic hepatitis C (CHC) is one of the leading causes of cirrhosis, hepatocellular carcinoma, and liver transplantation in Western countries.3 CHC patients are at high risk of liver fibrosis, and up to 20% will develop cirrhosis and associated complications. The pace of progression to advanced/severe liver fibrosis is highly variable among individuals and varies from less than 10 years to several decades.4 Epidemiological studies have shown that clinical parameters including older age, male sex, elevated body mass index, insulin resistance, chronic alcohol consumption (>30-50 g/day), histological factors (such as steatosis), viral factors (especially HCV genotype 3), or coinfection with human immunodeficiency virus were associated with a greater risk of fibrosis progression.4-8 However, these factors, even when combined, remain poor indicators of progression, with an overall prediction rate of <30%.9 Genetic factors have long been suspected to play a role in CHC.10, 11 Thus, in earlier candidate gene studies, several single nucleotide polymorphisms (SNPs) within host genes and/or gene regions coding for the HLA system, keratin, or coagulation factors were shown to be associated with progression of HCV-induced liver fibrosis.12 Many of these studies suffered from various biases, and some yielded contradictory results.13 By contrast, using a hypothesis-independent, gene-centric functional genome scan approach, a seven-gene variant signature called the cirrhosis risk score (CRS) was shown to predict fibrosis progression more accurately.14-16 Recently, an independent genome-wide association study identified a nonsynonymous sequence variation (rs738409 C>G) encoding an isoleucine-to-methionine substitution at position 148 in the adiponutrin/patatin-like phospholipase-3 (PNPLA3) gene that appeared to be the strongest determinant of human steatosis.17 This SNP was also found to be associated not only with elevated liver enzymes in healthy subjects18 but also with disease severity, steatosis, and fibrosis, both in nonalcoholic fatty liver disease (NAFLD)19, 20 and alcoholic liver disease (ALD)21-23.

The primary aim of this candidate gene polymorphism study was, therefore, to assess the impact of the rs738409 variant on steatosis and fibrosis severity/progression in CHC patients. Furthermore, because advanced fibrosis and steatosis have been shown to influence response to antiviral therapy,24, 25 we also studied the potential influence of this SNP on treatment response.

Patients and Methods

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Patient Selection

Seven hundred sixty-five CHC patients were initially retrospectively identified from three European centers: Erasme Hospital, Brussels, Belgium (study cohort); Medizinische Hochschule, Hannover, Germany; and Hôtel-Dieu Hospital, Lyon, France (replication cohorts). Patients were enrolled if they had persistent anti-HCV antibodies, were HCV-RNA–positive, and had at least one histological evaluation prior to any antiviral therapy. Exclusion criteria included the presence of any coexisting chronic liver disease shown by histology or imaging, coinfection with hepatitis B virus or human immunodeficiency virus, and prior or actual immunosuppressive therapy. To minimize the likelihood of spurious association due to population stratification,26 only Caucasians were included. HCV patients with genotype 3 were also excluded, because carriage of this virus subtype might have represented a confounding factor due to its influence on both steatosis/fibrosis.7, 8 Thus, 537 patients were studied (Supporting Fig. 1).

Clinical, biological, viral, and histological characteristics for each cohort are summarized in Table 1. Written informed consent was obtained from all patients, and this study was approved by the local ethics committee of each of the three study centers.

Table 1. Patient Characteristics by Center
CharacteristicsBrussels (n = 229)Hannover (n = 171)Lyon (n = 137)P Value
  1. Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index; HDLc, high-density lipoprotein cholesterol; LDLc, low-density lipoprotein cholesterol.

  2. Variables with normal distribution are presented as the mean ± SD. Skewed variables are presented as the median (interquartile range).

Clinical    
 Age, years51.6 ± 13.046.9 ± 11.249.0 ± 9.8<0.001
 Sex, % male59.059.173.00.014
 BMI, kg/m225.7 ± 5.425.9 ± 4.424.8 ± 4.00.010
 Diabetes, %11.410.74.70.099
 Alcohol >30 g/day, %32.02.119.8<0.001
Biological    
 ALT, IU/L67 (48-110)63 (38-92)64 (41-119)0.292
 AST, IU/L57 (37-97)53 (38-83)54 (38-97)0.719
 Cholesterol, mg/dL163.3 ± 40.1166.3 ± 40.2169.7 ± 45.80.502
 HDLc, mg/dL45.2 ± 19.248.6 ± 17.445.5 ± 18.80.560
 LDLc, mg/dL100.7 ± 31.8111.5 ± 30.791.4 ± 32.20.009
 Triglycerides, mg/dL81.0 (64.0-106.0)94.5 (73.0-140.5)105.3 (77.9-135.4)<0.001
 Glucose, mg/dL94.0 (84.0-104.0)90.1 (81.1-104.5)93.7 (81.1-101.8)0.296
Viral    
 Viral load >600,000 IU/mL, %53.163.666.70.046
 Genotype, %   0.133
  189.390.685.1 
  24.76.55.2 
  44.70.66.0 
  Other1.22.43.7 
Histological    
 Steatosis, %   0.066
  <5%39.042.931.1 
  5%-33%55.649.555.5 
  >33%5.47.613.4 
 Fibrosis stage, %   0.290
  F0-F261.454.762.5 
  F3-F438.645.337.5 

Histological Classification

Only patients with a pretreatment biopsy suggesting CHC were selected. The analysis of the percentage of steatosis among total hepatocytes and fibrosis were assessed locally in each center by pathologists who were blinded to the study. Fibrosis was expressed according to the METAVIR scoring system.27

Genotyping

In the Brussels and Hannover cohorts, DNA was isolated from whole blood samples prospectively collected in the liver clinics concomitantly with a liver biopsy. In the Lyon cohort, DNA was extracted either from whole blood or from paraffin-embedded liver specimens.28 The genotyping procedure was performed for all three cohorts at the Laboratory of Experimental Gastroenterology, Université Libre de Bruxelles, Brussels, Belgium. Based on HapMap data,29 10 tagging SNPs in the PNPLA3 gene region (22q13.31, reference sequence: NM_025225) ±20 kb were selected (minor allele frequency >0.2 and r2 >0.6) and genotyped in our study cohort (Brussels). Twenty nanograms of DNA were used to genotype PNPLA3 (rs738409, rs4823164, rs9625960, rs6006456, rs139047, rs139051, rs1883350, rs1977081, rs3810622, rs6006591) variants. In addition, we genotyped nine other SNPs. We included seven reported to be associated with fibrosis: rs1987098 in syntaxin binding protein 5-like (STXBP5L); rs2878771 in aquaporin 2 (collecting duct) (AQP2); rs886277 in transient receptor potential cation channel, subfamily M, member 5 (TRPM5); rs2304349 and rs2570942 in antizyme inhibitor 1 (AZIN1); rs4290029 in degenerative spermatocyte homolog 1-nuclear VCP-like (DEGS1-NVL); rs4986791 in Toll-like receptor 4 (TLR4); and, finally, two variants associated with treatment response in interferon lambda 3 (IL-28B), rs12979860 and rs8099917. All SNPs were genotyped with TaqMan assays (Applied Biosystems, Foster City, CA) on a LightCycler 480-Real-Time PCR System (Roche Diagnostics GmbH, Mannheim, Germany). We included internal positive (DNA samples of known genotype) and negative controls (water) to secure the genotyping procedure. In addition, some of the genotypes were cross-checked through direct sequencing. The success rate for genotyping was 99%. All allele frequencies were in Hardy-Weinberg equilibrium.

Endpoints and Statistical Analyses

Cohort Analysis and Statistical Power.

To achieve suitable power for each study endpoint (fibrosis stage, fibrosis progression, steatosis) (Supporting Fig. 2), we examined the feasibility of pooling the three cohorts. Thus, G allele frequencies were first compared to assess potential differences between centers. Furthermore, for each study endpoint, odds ratios (ORs) in each cohort were compared using the Cochran-Mantel-Haenszel test, and homogeneity of the ORs was assessed using the Breslow-Day test. To assess the agreement between pathologists for fibrosis staging, Cohen's kappa coefficient was calculated.

Association Between PNPLA3 (rs738409 C>G) Polymorphism and Histological Damage.

We evaluated the effect of the rs738409 mutant G allele on liver damage using both a dominant genetic model comparing G allele carriers (GC or GG genotypes) with patients carrying no copy of the G allele (CC genotype) and a recessive model, comparing G allele homozygotes (GG genotype) with patients carrying one or no copy of the G allele (CG or CC genotypes). The effect on fibrosis was first assessed using a cross-sectional approach (evaluation at a single time point), comparing patients with either no fibrosis, mild fibrosis, or intermediate fibrosis (F0-F1-F2) with patients with severe fibrosis or cirrhosis (F3-F4). Fibrosis progression was also studied. We calculated a fibrosis progression rate as the ratio between fibrosis stage (METAVIR score) and the estimated duration of disease, defined as the interval between the presumed date of infection and the date of biopsy.4 A median fibrosis progression rate was determined, and patients were then classified as either progressors (patients with a fibrosis progression rate higher than the median) or nonprogressors (patients with a fibrosis progression rate equal or lower than the median). For the analysis of fibrosis and fibrosis progression, only patients for whom no antiviral therapy had been started prior to histological evaluations were considered. The presence of steatosis was studied as a qualitative (<5% versus ≥5%) variable. We conducted a multivariable logistic regression to assess the respective impact of rs738409 SNP and classical factors reported to influence fibrosis and steatosis. A recessive model of inheritance was chosen because it was suggested to be the most appropriate one by studies assessing rs738409 impact in CHC histological damage.30, 31

Mapping of the PNPLA3 Region.

Ten tagging SNPs (including rs738409) were selected to cover the PNPLA3 region and genotyped on the Brussels cohort. Fifty-four additional SNPs were also imputated using MACH software.32 Association tests were performed under an allelic model.

Previously Described Gene Variants Associated with Fibrosis.

Six of seven variants of a reported CRS associated with fibrosis were individually tested14–16: STXBP5L (rs1987098), AQP2 (rs2878771), TRPM5 (rs886277), AZIN1 (rs2304349), DEGS1-NVL (rs4290029), and TLR4 (rs4986791). Three other SNPs in AZIN1 (rs2570942) and IL-28B (rs12979860 and rs8099917) were also tested. Association tests were performed under an allelic model. This analysis permitted the investigation of the association of the individual SNPs out of context of differentially weighted and prespecified genetic models and aggregated score.

Impact of PNPLA3 (rs738409 C>G) Polymorphism on Antiviral Therapy Outcome.

Sustained virological response (SVR) was studied in 229 patients and was defined as the presence of an HCV-RNA–negative polymerase chain reaction 6 months after cessation of antiviral therapy.25 Only patients treated with a combination of pegylated interferon and ribavarin according to standard of care were studied.24, 25 We conducted a multivariable logistic regression for the absence of SVR, including the PNPLA3 (rs738409 C>G) polymorphism and factors reported to be associated with it. In addition, we used the same model for IL-28B (rs12979860 C>T), because a recent genome-wide association study has shown that this SNP located in the IL-28B gene region is associated with SVR.33 This polymorphism was chosen as both a comparator and a positive control of the potential influence of the rs738409 variant upon SVR.

For all endpoints, categorical variables were studied using the two-sided chi-square test or the two-sided Fisher's exact test, whereas quantitative variables were analyzed using the analysis of variance or the nonparametric Kruskall-Wallis test when appropriate. The Hosmer-Lemeshow test was used to verify goodness of fit of the logistic regression models to the data. Statistical analyses were performed using SPSS version 17.0 software; P < 0.05 was considered statistically significant. Post hoc adjustment for multiple testing was performed when necessary.

Results

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Individual Cohort Analysis.

Results indicated that the G allele frequency was not statistically different between the three cohorts (29.4 versus 25.4 versus 22.6% for Brussels, Hannover, and Lyon, respectively; P = 0.111) (Table 2). Furthermore, the overall OR among G allele carriers and G allele homozygote individuals for the risk of fibrosis (cross-sectional and longitudinal) and steatosis was similar among centers (Breslow-Day test was not statistically significant; Supporting Tables 1 and 2). Therefore, these results confirmed the feasibility of pooling. Moreover, because the individual cohorts were underpowered for each main study endpoint (Supporting Fig. 2), the following results are presented as a single cohort. The concordance between pathologists assessed by Cohen's kappa coefficient (κ = 0.69, P < 0.0001) indicated substantial agreement.27

Table 2. Distribution of PNPLA3 (rs738409 C>G) Genotype by Center
 Brussels (n = 229)Hannover (n = 171)Lyon (n = 137)P Value
PNPLA3 (rs738409 C>G) genotype, no. (%)    
 CC117 (51.1)96 (56.1)78 (56.9)0.080
 CG89 (38.9)63 (36.8)56 (40.9) 
 GG23 (10.0)12 (7.0)3 (2.2) 
G allele frequency, %29.425.422.60.111

Association Between the rs738409 C>G PNPLA3 Polymorphism and Clinical/Biochemical Characteristics.

Clinical and biochemical characteristics according to patients' genotype available for the cohort are summarized in Supporting Table 3. No association was observed between the rs738409 polymorphism and key biological characteristics such as alanine aminotransferase (P = 0.204) or viral load (P = 0.457). Moreover, no significant influence of the SNP was seen with other metabolic parameters such as type 2 diabetes (P = 0.938), serum glucose level (P = 0.366), or the homeostasis model assessment of insulin resistance (HOMA-IR) score (P = 0.626).

Association Between the rs738409 C>G PNPLA3 Polymorphism and Histological Damage.

When investigating fibrosis in a cross-sectional manner, the frequency of the rs738409 G allele was significantly higher in patients with severe fibrosis/cirrhosis (F3-F4 stages) than in patients at a less advanced stage (F0-F2; P = 0.021) (Table 3). Nevertheless, among G allele carriers (CG and GG genotypes), the homozygote (60% [F3-F4]) and heterozygote (41% [F3-F4]) frequencies varied. Regardless, the risk for fibrosis was not increased in G allele carriers (OR 1.33, 95% confidence interval [CI] 0.94-1.88, P = 0.110) (Table 3) but was in GG homozygotes (OR 2.43, 95% CI 1.24-4.78, P = 0.008) (Table 3). This observation was further confirmed when studying fibrosis progression in a longitudinal manner. Indeed, 71% of G allele homozygotes were progressors (OR 2.65, 95% CI 1.28-5.51, P = 0.007) (Table 3).

Table 3. Histological Characteristics According to PNPLA3 (rs738409 C>G) Genotype and Related Odds Ratios
 GenotypeP Value
CCCGGG
Fibrosis stage (cross-sectional) (n = 537), no. (%)    
 F0-F2 (%)183 (62.9)123 (59.1)15 (39.5)0.021
 F3-F4 (%)108 (37.1)85 (40.9)23 (60.5) 
Fibrosis progression (n = 398), no. (%)    
 Nonprogressors113 (53.3)74 (50.0)11 (28.9)0.022
 Progressors99 (46.7)74 (50.0)27 (71.1) 
Steatosis (n = 435), no. (%)    
 <5%91 (40.1)64 (37.6)7 (18.4)0.038
 ≥5%136 (59.9)106 (62.4)31 (81.6) 
 Dominant Model (CG + GG Versus CC Genotypes)Recessive Model (GG Versus CG + CC Genotypes)
OR (95% CI)P ValueOR (95% CI)P Value
  1. Abbreviations: CI, confidence interval; OR, odds ratio.

Fibrosis stage (cross-sectional)1.33 (0.94-1.88)0.1102.43 (1.24-4.78)0.008
Fibrosis progression1.36 (0.91-2.01)0.1302.65 (1.28-5.51)0.007
Steatosis ≥5%1.29 (0.87-1.91)0.1992.84 (1.22-6.60)0.012

The frequency distribution of the G allele was also significantly associated with evidence of steatosis (≥5%) (P = 0.038) and, similarly to what was observed with fibrosis, this conferred a higher risk to G allele homozygotes (OR 2.84, 95% CI 1.22-6.60, P = 0.012) than to G allele carriers (OR 1.29, 95% CI 0.87-1.91, P = 0.199). Interestingly, even after correction for multiple testing given that two models were conducted, GG genotype remained a statistically significant predictor for significant fibrosis, fibrosis progression, and steatosis. These results are comparable using an allelic model (data not shown). Data for these three histological outcomes are also available for separated cohorts (Supporting Tables 1 and 2).

Of note, there was no influence of the IL-28B (rs12979860 and rs8099917) genotype assessed by a dominant or recessive model on either fibrosis (Supporting Table 6) or steatosis (data not shown)

Logistic Regression Model for Histological Damage.

In light of the above results, a greater influence of rs738409 in homozygote G allele carriers, we conducted a univariate and multivariable logistic regression analysis for fibrosis (cross-sectional and progression) and the presence of steatosis that includes the rs738409 genotype using recessive homozygote (GG versus CG and CC genotypes) as a reference and included reported clinical factors (Table 4). After adjustment, G allele homozygosity remained independently associated with histological liver damage. Sex was also independently associated with fibrosis studied in both manners, and excessive alcohol consumption remained associated with its cross-sectional assessment (P = 0.015). Body mass index was an independent predictor of steatosis (P = 0.002). For all three models presented, the Hosmer-Lemeshow test was nonsignificant (indicating suitable adjustment of models to the data). A HOMA-IR index >2 was associated with a greater fibrosis stage assessed in a cross-sectional manner (OR 2.81, 95% CI 1.43-5.52, P = 0.002) and the presence of steatosis (OR 4.62, 95% CI 2.20-9.71, P < 0.001). Nevertheless, this observation concerned a univariate subgroup analysis (given that these data were only available in 164 and 149 patients, respectively) and did not enable us to perform further analyses.

Table 4. Univariate and Multivariable Logistic Regression Analysis of PNPLA3 (rs738409 C>G) Influence on Histological Liver Damage Characteristics
VariablesUnivariateMultivariable
OR95% CIP ValueOR95% CIP Value
  • Abbreviations: BMI, body mass index; CI, confidence interval; OR, odds ratio.

  • *

    Stage F0-F2 versus stage F3-F4.

Significant fibrosis* (n=537)      
 PNPLA3 (rs738409) GG2.431.24-4.780.0083.131.50-6.510.002
 Age, years1.041.02-1.05<0.0011.051.03-1.07<0.001
 Sex (male)1.961.36-2.84<0.0012.661.56-4.52<0.001
 BMI, kg/m21.051.01-1.090.0181.040.98-1.090.165
 Diabetes2.211.22-4.000.0091.160.52-2.580.710
 Alcohol >30 g/day1.941.23-3.080.0052.051.15-3.650.015
 Steatosis ≥5%1.510.99-2.280.0511.190.72-1.970.505
Fibrosis progression (n = 398)      
  PNPLA3 (rs738409) GG2.651.28-5.510.0072.641.22-5.670.013
 Age, years1.010.99-1.030.0991.020.99-1.050.060
 Sex (male)1.651.10-2.480.0152.061.20-3.550.009
 BMI, kg/m21.030.98-1.080.1961.010.96-1.070.674
 Diabetes1.220.62-2.390.5600.650.26-1.650.368
 Alcohol >30 g/day1.590.90-2.830.1131.580.79-3.150.197
 Steatosis ≥5%1.611.01-2.580.0471.500.86-2.610.150
Steatosis ≥5% (n = 435)      
 PNPLA3 (rs738409) GG2.841.22-6.600.0122.551.08-6.030.034
 Age, years1.010.99-1.030.1941.010.99-1.030.180
 Sex (male)1.541.03-2.290.0341.440.89-2.320.136
 BMI, kg/m21.101.05-1.15<0.0011.081.03-1.140.002
 Diabetes1.600.78-3.300.2041.120.48-2.590.796
 Alcohol >30 g/day1.300.77-2.180.3301.350.73-2.500.34

Mapping of the PNPLA3 Region.

Associations with fibrosis and fibrosis progression are reported in Fig. 1, Supporting Table 4, and Supporting Table 5. Neither the nine other genotyped tagging SNPs (Fig. 1 and Supporting Table 4) nor the 54 imputated ones (Supporting Table 5) were more associated than rs738409. However, rs139047 appeared to be more strongly associated with steatosis (nominal P = 0.003) than rs738409 (nominal P = 0.009). After correction for multiple testing using a permutation procedure, only rs139047 remained significantly associated with steatosis (corrected P = 0.021), and a clear trend to significance was observed for rs738409 (corrected P = 0.068). Of note, the linkage of rs139047 and rs738409 was not detected (r2 = 0.021).

thumbnail image

Figure 1. Mapping of the PNPLA3 region in the Brussels cohort (tagging SNPs, n = 10). The upper panel shows tagging SNPs, positions, and linkage disequilibrium structure. The lower panel shows nominal P values from analysis for fibrosis (assessed in a cross-sectional manner), fibrosis progression, and steatosis (≥5%). The rs738409 SNP was associated with fibrosis (nominal P = 0.027), fibrosis progression (nominal P = 0.015), and steatosis (nominal P = 0.009). The rs139047 SNP was also associated with steatosis (▴) (nominal P = 0.003). After corrections for multiple testing, only the association of rs139047 with steatosis remained statistically significant (corrected P = 0.021).

Download figure to PowerPoint

Previously Described Gene Variants Associated with Fibrosis.

On an individual basis, none of the nine SNPs selected could be associated with fibrosis or fibrosis progression (Supporting Table 6).

Impact of the PNPLA3 (rs738409 C>G) Polymorphism on Antiviral Therapy Outcomes.

On univariate analysis, rs738409 G allele carriers were associated with failure to achieve SVR relative to G allele homozygosity (OR 1.89, 95% CI 1.06-3.38, P = 0.030 and OR 1.25, 95% CI 0.23-6.70, P = 0.794, respectively) (Supporting Table 7). However, after adjustment, PNPLA3 (rs738409 C>G) G allele carriers did not remain independently associated with non-SVR (OR 1.07, 95% CI 0.46-2.49, P = 0.875). Routine risk factors, including fibrosis (OR 18.64, 95% CI 4.01-86.60, P < 0.001), male sex (OR 2.52, 95% CI 1.05-6.06, P = 0.039), and viral load >600,000 IU/mL (OR 2.55, 95% CI 1.09-5.94, P = 0.031), were included in the model.

In contrast, still using a dominant model, the IL-28B (rs12979860 C>T) mutant T allele was, as expected, associated with therapy failure (OR 4.35, 95% CI 1.67-11.31, P = 0.003) (Supporting Table 7).

Discussion

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

This study demonstrates that an SNP in the PNPLA3 gene (rs738409 C>G) significantly influences fibrosis and fibrosis progression in CHC. In addition, this SNP appears to be a risk factor for steatosis. Conversely, rs738409 does not seem to influence the outcome of standard antiviral therapy. This is the first study to report the significant impact of rs738409 on fibrosis progression using liver biopsy specimens. In addition, we report that another SNP within the PNPLA3 region (rs139047) was associated with steatosis.

The SNP rs738409 has been associated with both NAFLD17, 19, 20 and ALD.21-23 The role of this polymorphism was originally revealed as a major determinant of steatosis in NAFLD.17, 19, 20 Our results in CHC are consistent with these studies emphasizing rs738409 as an independent predictor of liver fat accumulation. Steatosis is influenced by viral genotype. Indeed, HCV genotype 3 has been reported to be an independent predictor of liver fat.7 Unfortunately, this was not investigated in the present study. However, two recent studies reported that the impact of rs738409 on steatosis was limited to patients with non–genotype 3 HCV.30, 34 Remarkably, the rs139047 PNPLA3 variant was even more strongly associated with steatosis than rs738409. However, these findings merit further investigation in subsequent studies. Furthermore, rs738409 has been reported to be associated with advanced disease course in ALD,21-23 and more specifically with fibrosis severity in cross-sectional studies in NAFLD, independently of steatosis.19, 20 Similarly, in our study, rs738409 was the strongest independent determinant of fibrosis. Moreover, older age, male sex, and alcohol consumption were confirmed to be associated with severe fibrosis/cirrhosis. Again, our results are in line with those of investigators reporting the link between rs738409 and fibrosis studied in a cross-sectional manner in CHC.30 Remarkably, the impact of rs738409 was also obvious when studying fibrosis progression.

Conversely, six other variants, a subset of the seven CRS SNPs signature reported to influence fibrosis severity14 and progression,15 were not associated on an individual basis. This was an expected finding, because the CRS was designed and validated in previous studies as a unique combination of genetic variants. We have recently replicated the CRS as a predictive score in fibrosis progression in this population when used in combination and not on an individual basis.16

Not surprisingly, the PNPLA3 rs738409 G allele was associated with SVR failure in univariate but not in multivariable analysis. This finding was anticipated, because severe fibrosis/cirrhosis was a major predictor of treatment nonresponse in our study. The rs738409 variant is possibly influencing fibrosis, a strong established factor of therapy failure.24, 25 In this regard, our findings are not concordant with previous studies.30 Indeed, in our logistic model, we included more confounders (sex, viral load, alcohol consumption) previously associated with SVR,25 which may have substantially reduced the variant's impact. Moreover, a recent study reported that rs738409 had a striking impact on fibrosis, a major treatment confounder, in patients drinking >30 g/day.35 Nevertheless, our study included fewer patients for this outcome than these authors and more insights are required to unravel this issue. Conversely, the IL-28B rs12979860 T allele, which modulates the interferon response, was a strong independent predictor of therapy failure with various viral genotypes as demonstrated.33, 36 Nevertheless, it is notable that the rs12979860 variant had no impact on fibrosis severity, as reported in a novel candidate SNP analysis from the IDEAL study.37 We did not find any association with this variant on steatosis ≥5%. Another SNP in the IL-28B region (rs12979075) has been reported to also influence steatosis in Caucasian CHC patients with non–genotype 3.34 We did not assess this variant, which was in high linkage with rs12979860 (r2 = 0.88), in American Caucasian patients.33 Moreover, rs12979075 was shown in that study to have a much lower impact than PNPLA3 (rs738409). A mapping of the IL-28B region will inform future efforts.

In humans, PNPLA3 encodes a transmembrane protein expressed in stellate cells but more prominently in hepatocytes,38 where it interacts with lipids.39 The G allele of rs738409 variant leads to an isoleucine to methionine substitution at position 148 that alters catalytic activity, leading to triglyceride accumulation in hepatocytes.39 The exact function of the enzyme and the mechanisms underlying liver injury remain hypothetical.40 Steatosis can promote inflammatory mediators and oxidative stress and has been shown to favor hepatocyte apoptosis in CHC.41 Steatosis is also closely associated with metabolic syndrome and insulin resistance, a well-established cause of fibrosis progression in hepatitis C.42 Thus, insulin resistance, and more specifically its surrogate marker HOMA-IR index, was associated with both steatosis and higher fibrosis stage. The cutoff for this index, chosen as 2, has been reported to reflect insulin resistance in both steatosis and fibrosis in CHC.43 Interestingly, the rs738409 genotype did not influence insulin resistance as reported in NAFLD.44 Although the HCV life cycle is integrally linked to lipid metabolism,45 the rs738409 genotype was not associated with viral load. We did not observe any influence of this variant on parameters of lipid metabolism in this work, and these results are consistent with those of a recent well-powered genome-wide association study in NAFLD.46

In contrast to studies in both NAFLD and ALD, liver damage (steatosis and fibrosis) related to rs738409 in CHC appears to primarily involve the homozygote G allele carriers as reported by others30, 31 and was found in 7% of our CHC population. Therefore, the potential clinical relevance appears limited. This advocates for the use of combined genetic markers in future individualized management decisions.

In conclusion, the PNPLA3 (rs738409 C>G) polymorphism is associated with both steatosis and fibrosis progression in CHC. Further studies are required to confirm our results and to evaluate the potential of PNPLA3 to serve both as predictor and a therapeutic target in CHC. Notably, the PNPLA3 polymorphism has now been associated with liver diseases with different etiologies (NAFLD, ALD, and CHC) and therefore must perturb a common underlying biological pathway leading to fibrosis. Therapeutic intervention of this common pathway may therefore portend efficacy in liver disease associated with viral, lipid, and alcohol insults.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

We thank Michel Georges for valuable advice in genetic analyses; Delphine Degré, Romy Ouziel, and Vincent Vercruysse for logistic assistance; Paule Guilloreau for valuable help in data collection; and Béatrice Gulbis, François Demoor, and Ann Vann Roost for technical assistance.

References

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
HEP_24350_sm_suppinfo.doc329KSupporting Information

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.