Meta-analysis: IL28B polymorphisms predict sustained viral response in HCV patients treated with pegylated interferon-α and ribavirin

Authors

  • Y. Chen,

    1. Department of Infectious Diseases, First Affiliated Hospital of the Medical College of Xi'an Jiaotong University, Xi'an, Shanxi Province, China
    2. Institute of Liver Diseases, Huai'an Fourth People's Hospital, Huai'an, China
    Search for more papers by this author
  • H.-X. Xu,

    1. Institute of Liver Diseases, Huai'an Fourth People's Hospital, Huai'an, China
    2. Department of Infectious Diseases, Affiliated Hospital of the Xuzhou Medical College, Xuzhou, China
    Search for more papers by this author
  • L.-J. Wang,

    1. Department of Infectious Diseases, Affiliated Hospital of the Xuzhou Medical College, Xuzhou, China
    Search for more papers by this author
  • X.-X. Liu,

    1. Institute of Liver Diseases, Huai'an Fourth People's Hospital, Huai'an, China
    Search for more papers by this author
  • R. I. Mahato,

    1. Department of Pharmaceutical Sciences, University of Tennessee Health Science Center, Memphis, TN, USA
    Search for more papers by this author
  • Y.-R. Zhao

    Corresponding author
    • Department of Infectious Diseases, First Affiliated Hospital of the Medical College of Xi'an Jiaotong University, Xi'an, Shanxi Province, China
    Search for more papers by this author

  • As part of AP&T's peer-review process, a technical check of this meta-analysis was performed by Dr P. Collins.

Correspondence to:

Dr Y.-R. Zhao, Department of Infectious Diseases, First Affiliated Hospital of the Medical College of Xi'an Jiaotong University, No. 277 Yanta West Road, Xi'an 710061, Shanxi Province, China.

E-mail: zhaoyingren@sohu.com

Summary

Background

Interleukin (IL) 28B single nucleotide polymorphisms can predict sustained virological response (SVR) in hepatitis C virus (HCV) patients following pegylated interferon-alpha (PEG IFN-α) and ribavirin treatment.

Aim

To design a meta-analysis to determine IL28B genotypes', rs12979860 CC and rs8099917 TT, correlation with SVR in PEG IFN-α/ribavirin-treated HCV patients.

Methods

Meta-analysis was performed in 17 studies of rs12979860 CC vs. CT/TT and 17 of rs8099917 TT vs. TG/GG. Odds ratios (OR) and confidence intervals (95% CI) were calculated by fixed- or random-effects models. Heterogeneity, sensitivity analysis and publication bias were also assessed.

Results

Of 4252 Asian, Caucasian and African HCV patients analysed for rs12979860, SVR was more frequent in CC (vs. CT/TT; OR = 4.76, 95% CI: 3.15–7.20). Moreover, CC was associated with SVR for HCV genotype-1 or -4 infections (ORgenotype 1 = 5.52, 95% CI: 3.74–8.15; ORgenotype 4 = 8.11, 95% CI: 4.13–15.93), regardless of ethnicity. Of 4549 Caucasian and Asian HCV patients analysed for rs8099917, SVR was more frequent in TT (vs. TG/GG; OR = 3.31, 95% CI: 2.39–4.59). Moreover, TT was associated with SVR for HCV-1 (ORgenotype 1 = 4.28, 95% CI: 2.87–6.38). Rs8099917 TT predictive value was stronger in Asians (ORAsians = 8.09, 95% CI: 5.63–11.61; ORCaucasians = 3.00, 95% CI: 2.03–4.45). Ethnicity stratification revealed that rs8099917 TT had slight predictive value in Asian HCV-2/3 patients (OR = 1.99, 95% CI: 1.09–3.62).

Conclusions

IL28B rs12979860 CC and rs8099917 TT are strong SVR predictors for PEG IFN-α/ribavirin-treated HCV-1 patients, regardless of ethnicity. In HCV-2/3, rs12979860 CC has no SVR predictive value, but rs8099917 TT was slightly associated with SVR in Asians.

Introduction

Hepatitis C virus (HCV) infection has reached epidemic proportions worldwide, with the number of chronically infected persons estimated at 200 million. The long-term hepatic impact of HCV infection can range from fibrosis to cirrhosis with or without hepatocellular carcinoma (HCC).[1] Currently, the standard therapy for chronic HCV infection is a combination regimen of pegylated interferon-alpha (PEG IFN-α) and ribavirin (RBV).[2] Patient response to therapy varies widely, and it is believed that genetic factors underlie this phenomenon. The primary goal of any HCV therapy is to achieve a sustained virological response (SVR), in which HCV RNA remains undetectable at 24 weeks after therapy ends. For PEG IFN-α/RBV, 40–50% of patients infected with HCV genotypes 1 or 4 attain SVR, whereas 70–90% of patients infected with genotypes 2 or 3 attain SVR.[3-8] Unfortunately, the combination therapy produces several serious side effects, such as a flu-like syndrome, haematological abnormalities and adverse neuropsychiatric events, which necessitate dose reduction. Between 10% and 14% of patients suffer such extensive effects that treatment termination is required.[9] Thus, it is important to identify patients who are ultimately going to be resistant to treatment, to avoid unnecessarily inducing detrimental side effects and to reduce the substantial cost of PEG-IFN-α/RBV treatment. Several studies to date have aimed to identify accurate and sensitive predictive biomarkers of treatment response. Besides virus genotype, several other factors related to the virus (e.g. viral load at treatment initiation) and host (e.g. race, age, body weight, insulin resistance, serum lipids, fibrosis stage, serum vitamin D concentration and treatment compliance) can influence antiviral-induced SVR in chronic HCV patients.[10-15]

Host genetic factors have been proposed as mediators of treatment response. Chronic HCV patients of European or Asian ancestry[6] have a higher rate of SVR than African ancestry patients,[16] reinforcing the theory of genetics as contributing to SVR risk. Previous studies have investigated many potential genetic factors, including polymorphisms of the human leucocyte antigens (HLA) and the interleukin (IL)10 promoter, but results have been inconsistent.[17, 18] Recently, several genome-wide association studies (GWAS) have indicated that single nucleotide polymorphisms (SNPs) near the interferon-gamma3-encoding IL28B gene may be associated with SVR after PEG-IFN-α/RBV therapy.[19-22] IL28B rs12979860 and rs8099917 have particularly strong linkage disequilibrium, and are considered as the most important predictive factors for treatment response. The rs12979860 CC genotype has a higher incidence of SVR than the CT/TT genotypes, and the rs8099917 TT genotype has higher SVR incidence than the TG/GG genotypes. However, the magnitude of the impact of rs12979860 and rs8099917 polymorphisms on treatment response remains controversial. Meta-analysis is a powerful method for quantitatively summarising the results from different studies. It is particularly useful when individual trials have insufficient statistical power, since pooling of trial subjects increases the sample size and decreases random error.

Therefore, we performed meta-analyses to determine the impact of IL28B rs12979860 genotype CC and rs8099917 genotype TT for chronic HCV patients infected with different virus genotypes and of different ethnicities.

Methods

Literature search

The PubMed, Web of Science, Embase and the Chinese BioMedical Literature (CBM) databases were searched for relevant articles published up to October 15, 2011 using the following search terms: ‘chronic hepatitis C’, ‘HCV’, or ‘hepatitis C virus’; ‘interleukin-28B’ or ‘IL-28B’; ‘polymorphism’ or ‘genetic polymorphism’; ‘sustained virological response’ or ‘SVR’. To identify other relevant publications, the reference lists of all retrieved articles were manually searched; in addition, cited review articles were retrieved and perused for mention of any additional potentially relevant articles. Only published studies with full text articles were included in the meta-analysis. A flowchart of the study selection process is shown in Figure 1.

Figure 1.

Flowchart of the process for selecting studies for inclusion in the meta-analysis.

Inclusion and exclusion criteria

Studies were selected based on the following inclusion criteria: (a) HCV-infected patients received PEG-IFN-α/RBV combination therapy; (b) HCV genotype 1/4-infected patients received treatment for 48 weeks, and genotype 2/3-infected patients received treatment for 24 weeks; (c) SVR rates were compared with IL28B rs12979860 CC genotype and CT/TT genotype, or with rs8099917 TT genotype and TG/GG genotype; and (d) the genotype was tested to ensure the Hardy–Weinberg equilibrium (HWE). Studies were excluded according to the following criteria: (i) non-human population; (ii) co-infection with hepatitis B or human immunodeficiency virus HIV; (iii) co-existence of any other liver diseases, such as autoimmune hepatitis, alcoholic liver disease, drug hepatitis, Wilson's disease, decompensated liver cirrhosis or HCCs; (iv) liver transplantation recipients; (v) treatment with non-PEG IFN, or without RBV; (vi) study population deviated from HWE; or (vii) absent information on SVR rates in patients with IL28B SNPs rs12979860 CC vs. CT/TT or rs8099917 TT vs. TG/GG. When several publications reported data from a single study, the most recent and complete publication was selected.

Data extraction and study design

Data extraction was carried out by two independent investigators (Yong Chen and He-Xiang Xu) using predefined forms. All discrepancies were resolved by discussion, and, if required, participation by the third author (Ying-Ren Zhao). For all included studies, the following information was extracted: first author, publication year, ethnicity and geographical area, HCV genotype, number of chronic HCV cases, number of SVR cases, genotype frequency of IL28B in chronic HCV cases, genotype frequency of IL28B in SVR HCV cases. SVR was defined as undetectable HCV RNA at the end of treatment.

All meta-analyses were performed using the Review Manager (RevMan version 5.1; http://ims.cochrane.org/revman). The first overall analysis included all patients infected with HCV, regardless of virus genotype. The second analysis stratified patients by virus genotype. Since genotypes 2 and 3 require the same standard therapy time and attain similar SVR rates, these two genotypes were included in the same analysis. Within each of the genotype stratification analyses, additional stratification by ethnicity (Caucasian, Asian and African) was analysed.

Data analysis

Statistical analyses were performed using the Review Manager version 5.1 for Windows (Cochrane Collaboration, Oxford, UK) and Stata version 10.0 (Stata Corp LP, College Station, TX, USA) software programs. The association strength between IL28B SNPs (rs12979860 or rs8099917) and SVR in HCV patients treated with PEG-IFN-α/RBV was determined by calculating the respective odds ratio (OR) with 95% confidence intervals (CI). The significance of the pooled OR was determined by the Z-test, and a P-value of less than 0.05 was considered significant. Two meta-analysis models for dichotomous outcomes were used: the random-effects model (using DerSimonian and Laird's method[23]) and the fixed-effects model (using Mantel-Haenszel's method[24]). To assess between-study heterogeneity, both the Chi-squared based Q statistic test (Cochran's Q statistic) and the I 2 statistic[25] were calculated. For Cochran's Q statistic, heterogeneity was considered significant for < 0.10 and the results were pooled by the random-effects model. When > 0.10, the fixed-effects model was used. In addition, the Galbraith plot was used to detect potential sources of heterogeneity (outliers).[26] To validate the meta-analysis results, a sensitivity analysis was performed by sequential omission of individual studies or by omitting outlier studies identified by the Galbraith plotting method. Publication bias was investigated by funnel plot, Egger's linear regression method and Begg's rank correlation method.[27] All P-values were two-sided. To ensure the reliability and the accuracy of the results, two authors independently uploaded the data into the statistical software programs and verified that the results were identical.

Results

Characteristics of included studies

Among the 60 initially identified studies with potential relevance, 24 were excluded based on the study population having: spontaneous clearance of HCV (n = 6[28-33]); HIV co-infection (n = 6[34-39]); liver transplant (n = 6[40-45]); decompensated liver cirrhosis or HCC (n = 3[46-48]); treatment with non-PEG IFN (n = 1[49]); and no SVR data for IL28B rs12979860 or rs8099917 polymorphisms (n = 2[50, 51]). In addition, 11 articles were excluded for being: reviews (n = 10[52-61]); and an earlier publication of a multi-published trial (n = 1[28]). Consequently, eight studies[62-69] evaluated IL28B rs12979860 genotype CC vs. CT/TT only, eight studies[19, 21, 22, 70-74] evaluated rs8099917 TT vs. TG/GG only, and nine studies[75-83] evaluated IL28B rs12979860 genotype CC and rs8099917 genotype TT simultaneously. In total, 25 studies[19, 21, 22, 62-83] evaluating IL28B rs12979860 genotype CC (n = 17; 4252 cases) and rs8099917 genotype TT (n = 17; 4549 cases) were included in the meta-analyses (Figure 1).

The main characteristics of these studies are presented in Table 1. In all studies, HCV genotype 1/4-infected patients received subcutaneous PEG IFN-α-2a (180 μg/week) or PEG IFN-α-2b (1.5 μg/kg/week) and oral RBV (according to body weight) for 48 weeks, and HCV genotype 2/3-infected patients received subcutaneous PEG IFN and oral RBV for 24 weeks.

Table 1. Characteristics of IL28B rs12979860 and rs8099917 polymorphisms' genotype distributions in studies included in the meta-analysis
IL28B polymorphismNo.ReferenceHCV genotypeEthnicity (region)No. of patientsGenotype for patientsNo. of SVRGenotype for SVR
CC(TT)CT/TT(TG/GG)CC(TT)CT/TT(TG/GG)
rs129798601Thompson et al.[62]1Caucasian, African (US)15875121075639340299
    Caucasian,1287470817582320262
    African30042258572037
 2McCarthy et al.[63]1, 2, 3Caucasian, African (US)23176155724329
 3Mangia et al.[64]2/3Caucasian (Italy)26810016820182119
 4Lagging et al.[75]1Caucasian (Europe)16844124882959
 5Stattermayer et al.[76]1, 2, 3, 4Caucasian (Austria)682250432426201225
   1 372129243208102105
   2/3 208871211607090
   4 1023468592930
 Halfon et al.[77]1, 2, 3Caucasian (Israel)198611371084464
   1 15646110773245
   2/3 421527311219
 7Lindh et al.[78]1Caucasian (Sweden)1102882372413
 8Lin et al.[79]1Asian (Taiwan)191171201311247
 9Liao et al.[65]1, 2, 3, 6Asian (China)92821058562
 10Moghaddam et al.[80]3Caucasian (Norway)28112915222699127
 11De Nicola et al.[66]4Caucasian, African (Italy)1032479502129
 12Asselah et al.[67]4Caucasian, African (France)822260431825
 13Venegas et al.[81]1Caucasian (Chile)992079501931
 14Lyoo et al.[82]1Asian (Korea)6557842402
 15Arends et al.[68]1, 4Caucasian (Netherlands)1367651
 16Alestig et al.[69]1Caucasian (Sweden)501832291613
 17Honda et al.[83]1Asian (Japan)32181415123
rs80999171Tanaka et al.[19]1Asian (Japan)31419611814012515
 2Suppiah et al.[21]1Caucasian (Australia)848442406392247145
 3Rauch et al.[22]1, 2, 3, 4Caucasian (Switzerland)46527219329720196
 4Lagging et al.[75]1Caucasian (Europe)1689771885434
 5Stattermayer et al.[76]1, 2, 3, 4Caucasian (Austria)682409273426289137
   1 37221915320714859
   2/3 208118901609565
   4 1027230594613
 6Halfon et al.[77]1, 2, 3Caucasian (Israel)198115831087335
   1 1568571775126
   2/3 42301231229
 7Lindh et al.[78]1Caucasian (Sweden)1106644635013
 8Lin et al.[79]1Asian (Taiwan)191170211311238
 9Moghaddam et al.[80]3Caucasian (Norway)2812018022616165
 10Venegas et al.[81]1Caucasian (Chile)992970502525
 11Lyoo et al.[82]1Asian (Korea)6556942393
 12Ridruejo et al.[70]1Caucasian (Argentina)1024161372017
 13Hashimoto et al.[71]1Asian (Japan)84632143385
 14Sakamoto et al.[72]2Asian (Japan)12910029988117
 15Yu et al.[73]2Asian (Taiwan)4824325042938643
 16Hayashi et al.[74]1Asian (Japan)2992198013812711
 17Honda et al.[83]1Asian (Japan)32181415123

Meta-analyses results

For IL28B rs12979860, the between-study heterogeneity was significant when all 17 studies were pooled for analysis (I 2 = 79%, < 0.001); thus, the random-effects model was used. The pooled results indicated that IL28B rs12979860 genotype CC was associated with SVR in PEG IFN-α/RBV-treated HCV patients (OR = 4.76, 95% CI: 3.15–7.20; Figure 2a). Virus genotype stratification analyses indicated that rs12979860 CC was associated with SVR in HCV patients infected with genotypes 1 or 4 (ORgenotype 1 = 5.52, 95% CI: 3.74–8.15; ORgenotype 4 = 8.11, 95% CI: 4.13–15.93; Figure 2b and d respectively). No association was found for genotype 2/3 (ORgenotype 2/3 = 1.23, 95% CI: 0.71–2.14; Figure 2c). Ethnicity stratification analysis of genotype 1 showed that rs12979860 CC was associated with SVR in Asians, Caucasians and Africans (ORAsians = 5.79, 95% CI: 2.42–12.22; ORCaucasians = 5.48, 95% CI: 3.34–9.01; ORAfricans = 5.43, 95% CI: 2.70–10.92; Table 2). Since the genotype 2/3 and genotype 4 sub-groups had smaller sample sizes and were primarily composed of Caucasians, ethnicity stratification analyses were not performed.

Figure 2.

Association of IL28B rs12979860 genotype CC and SVR in PEG IFN-α/RBV-treated HCV patients, compared to genotype CT/TT. Random-effects model-forest plots of pooled ORs with 95% CI are shown for (a) total pooled analysis, (b) studies evaluating genotype 1 and (c) studies evaluating genotype 2/3. Fixed- effects model-forest plot of pooled ORs with 95% CI is shown for (d) studies evaluating genotype 4.

Table 2. Summary of IL28B polymorphism influence on SVR in PEG IFN-α/RBV-treated HCV patients in this meta-analysis
IL28B polymorphismMajor genotype vs. minor genotypeaNo. of studiesOdds ratioMbHeterogeneityBegg′s test, PEgger's test, P
OR [95% CI]PQI2 (%)Pc
  1. HCV, hepatitis C virus; IL 28B, interleukin 28B; PEG IFN-α pegylated interferon-alpha; SVR, single nucleotide polymorphisms.

  2. a

    Comparison of IL28B rs12979860 genotype CC with genotype CT/TT or comparison of IL28B rs8099917 genotype TT with genotype TG/GG.

  3. b

    M, model of meta-analysis: R, random-effects model; F, fixed-effects model.

  4. c

    P, the P-value of heterogeneity.

  5. d

    Values could not be calculated.

  6. e

    Adjustment for heterogeneity was performed by excluding the outlier studies (potential soures of heterogeneity) from Lindh et al. and Moghaddam et al.

  7. f

    Adjustment for heterogeneity was performed by excluding the outlier study from Lindh et al.

  8. g

    Adjustment for heterogeneity was performed by excluding the outlier studies from Tanaka et al., Suppiah et al., Lagging et al. and Hayashi et al.

  9. h

    Adjustment for heterogeneity was performed by excluding the outlier studies from Lindh et al. and Venegas et al.

rs12979860Overall174.76 [3.15, 7.20]<0.001R75.4579<0.0010.0770.450
 Overall with adjustment for heterogeneitye154.49 [3.37, 5.98]<0.001R26.28470.0240.0750.312
 Genotype 1105.52 [3.74, 8.15]<0.001R19.61540.0200.0740.336
 Genotype 1 with adjustment for heterogeneityf94.92 [4.10, 5.91]<0.001F10.55240.2290.1750.644
 Ethnicity stratification analysis
  Asians35.79 [2.42, 12.22]<0.001F0.2500.8841.0000.095
  Caucasians75.48 [3.34, 9.01]<0.001R19.56690.0030.2300.306
  Caucasians with adjustment for heterogeneityf64.30 [2.95, 6.26]<0.001R9.50470.0910.2600.645
  Africans15.43[2.70, 10.92]<0.001F-d-d-d-d-d
 Genotype 2/3, for all Caucasians41.23 [0.71, 2.14]0.455R6.63550.0851.0000.755
 Genotype 438.11 [4.13, 15.93]<0.001F0.5700.7531.0000.587
IL-28B rs8099917Overall173.31 [2.39, 4.59]<0.001R68.4177<0.0010.0530.146
 Genotype 1134.28 [2.87, 6.38]<0.001R50.2176<0.0010.1610.074
 Genotype 1 with adjustment for heterogeneity g93.89 [2.99, 5.07]<0.001F9.41150.3090.0760.099
 Ethnicity stratification analysis
  Asians68.09 [5.63, 11.61]<0.001F4.8100.4390.7070.121
  Caucasians73.00 [2.03, 4.45]<0.001R18.21670.0060.1330.235
  Caucasians with adjustment for heterogeneityh52.38 [1.95, 2.91]<0.001F5.59280.2321.0000.857
 Genotype 2/351.40 [0.98, 2.00]0.062F4.75160.3141.0000.914
 Ethnicity stratification analysis
  Asians21.99 [1.09, 3.62]0.024F1.57360.2101.000-d
  Caucasians31.19 [0.77, 1.84]0.438F1.4200.4931.0000.796

For IL-28B rs8099917, the between-study heterogeneity was significant when all 17 studies were pooled for meta-analysis (I 2 = 77%, < 0.001); thus, the random-effects model was used. The pooled results indicated that rs8099917 TT was associated with SVR in PEG IFN-α-treated HCV patients (OR = 3.31, 95% CI: 2.39–4.59; Figure 3a). In genotype 1 stratification analysis, rs8099917 TT was also associated with SVR in HCV patients (OR = 4.28, 95% CI: 2.87–6.38; Figure 3b). In the ethnicity stratification analysis for genotype 1, the association remained for Asians and Caucasians (ORAsians = 8.09, 95% CI: 5.63–11.61; ORCaucasians = 3.00, 95% CI: 2.03–4.45). In genotype 2/3 stratification analysis, however, rs8099917 genotype TT was not associated with SVR in HCV patients (OR = 1.40, 95% CI: 0.98–2.00; Figure 3c). Ethnicity stratification analysis of genotype 2/3 found no association among Caucasians (OR = 1.19, 95% CI: 0.77–1.84), but did find association among Asians (OR = 1.99, 95% CI: 1.09–3.62) (Figure 3d and Table 2).

Figure 3.

Association of IL28B rs8099917 genotype TT and SVR in PEG IFN-α/RBV -treated HCV patients, compared to genotype TG/GG. Random-effects model-forest plots of pooled ORs with 95% CI are shown for (a) total pooled studies and (b) studies evaluating genotype 1. Fixed-effects model-forest plots of pooled ORs with 95% CI are shown for (c) studies evaluating genotype 2/3 and (d) Asian ethnicity stratification analysis for genotype 2/3.

Heterogeneity analysis

For IL28B rs12979860, the between-study heterogeneity was not only significant for the pooled analyses (n = 17), but also for the virus genotype stratification analyses (genotype 1: = 0.020, I 2 = 54%; genotype 2/3: = 0.085, I 2 = 55%). Furthermore, ethnicity stratification analysis of patients with genotype 1 indicated that the between-study heterogeneity was significant for Caucasians (= 0.003, I 2 = 69%), but was not significant for the Asians or Africans (Table 2). In addition, Galbraith plotting of all 17 studies identified two outliers[78, 80] as possible sources of heterogeneity (Figure 4a). Galbraith plotting of patients with genotype 1 identified one study[78] as an outlier (Figure 4b), and it was one of the two previously identified for the Caucasians (Figure 4c). Excluding these two outlier studies did not significantly adjust the pooloed heterogeneity (< 0.10). Genotype 1 stratification analysis with the corrected pool (heterogeneity source[78] removed) indicated that heterogeneity was adjusted and reduced (= 0.229, I 2 = 24%). Caucasian stratification analysis with the corrected pool, however, indicated that heterogeneity was not significantly adjusted (= 0.091, I 2 = 47%).

Figure 4.

Identification of studies acting as sources of heterogeneity for the associations between IL28B polymorphisms and SVR in PEG IFN-α/RBV-treated HCV patients. Each author listed represents a separate study (shown in Tables 1 and 2) for the indicated association. Galbraith plots of the associations between IL28B (a, b, c) rs12979860 or (d, e, f) rs8099917 polymorphism genotype and SVR in (a, d) overall populations, (b, e) all genotype 1 patients, (c, f) Caucasian genotype 1 patients.

For IL-28B rs8099917, the between-study heterogeneity was not only significant for the pooled analysis (n = 17), but also for the virus genotype stratification analysis of genotype 1 (< 0.001, I 2 = 77%). In contrast, heterogeneity was not significant for genotype 2/3 (= 0.314, I 2 = 16%). In patients with genotype 1, the between-study heterogeneity was significant for ethnicity stratification analysis of Caucasians (= 0.006, I 2 = 67%), but not for any of the other ethnicities. In addition, Galbraith plotting of the overall pooled studies (n = 17) identified five outliers[19, 74, 78, 80, 81] as potential sources of heterogeneity (Figure 4d). Galbraith plotting of patients with genotype 1 identified four studies as outliers (Figure 4e), while plotting of Caucasian patients with genotype 1 identified two studies (Figure 4f). Exclusion of the four heterogeneity sources[19, 21, 74, 75] adjusted and reduced the heterogeneity for patients with genotype 1 (= 0.309, I 2 = 15%), as did removal of the two sources[78, 81] from the Caucasian genotype 1 analysis (= 0.232, I 2 = 28%). We did not try to reduce the obvious between-study heterogeneity of the overall analyses (n = 17) because it would have required excluding five studies (35%) and may have caused bias.

Sensitivity analysis

Sensitivity analysis was performed by sequential omission of individual studies. For pooled analyses of more than three individual studies, the significance of ORs was not influenced excessively by omitting any single study (data not shown). For the IL28B rs12979860 polymorphism in overall populations (n = 17), the two outlier studies[78, 80] from Galbraith plotting were removed and the significance of OR was not altered. Sensitivity analysis in patients with genotype 1 was performed by omitting the Galbraith plotting outlier study,[78] and the significance of ORs were not altered (Table 2). For the IL28B rs8099917 polymorphism in overall populations (n = 17), sensitivity analysis was not performed because too many studies (n = 5) would have to be excluded. However, sensitivity analysis was performed for patients with genotype 1 by omitting the four Galbraith plotting outlier studies,[19, 21, 74, 75] and for Caucasian patients with genotype 1 by omitting the two Galbraith plotting outlier studies[67, 78]; for each, the significance of ORs were not altered (Table 2).

Publication bias

Begg's test and Egger's test were performed to access the publication bias in this meta-analysis. Funnel plot shapes did not reveal obvious evidence of asymmetry and the P-values of both Begg's and Egger's tests were >0.05 (Table 2), providing statistical evidence of the funnel plots' symmetry. Thus, publication bias was not evident in this meta-analysis.

Discussion

Recent GWAS have identified genetic variations near the IL28B gene that are strongly associated with spontaneous and treatment-induced clearance of HCV infection. IL28B variations have been reported as strongly associated with on- treatment viral kinetics and with ~twofold increased SVR rates in HCV genotype 1 and in four patients. In fact, IL28B variations were shown to be the strongest pre-treatment predictor of virological response in HCV genotype 1 patients, but ineffective predictors for genotype 2/3-infected patients.[56] Variations in IL28B may also influence the kinetics of viral response to therapy. In one study, PEG-IFN/RBV-treated patients carrying the rs12979860 CC genotype had a greater HCV RNA decline than patients with the CT or TT genotype.[62] IL28B rs12979860 has also been associated with liver fibrosis in chronic HCV infection. In particular, the T allele affects the severity of fibrosis.[47] In our meta-analysis, pooled analysis of all studies showed that both rs12979860 CC and rs8099917 TT were associated with SVR in PEG IFN-α/RBV- treated HCV patients. Genotype stratification analyses showed that rs12979860 CC and rs8099917 TT were both associated with SVR in genotype 1, and not genotype 2/3, HCV patients. Ethnicity stratification analyses of genotype 1 patients showed that rs12979860 CC and rs8099917 TT were both associated with SVR in Asians, Caucasians and Africans. Thus, for genotype 1-infected patients, tests for IL28B rs12979860 or rs8099917 polymorphisms may help guide the physician in treatment regimen and design and/or the patient's decision to undergo therapy. In the ethnicity stratification analyses for genotype 2/3-infected patients, rs12979860 CC was not associated with SVR in Asians or Caucasians, but rs8099917 TT was associated with SVR in Asians.

It should be noted when interpreting the findings from our study that the meta-analysis design used has some possible limitations. First, meta-analyses are most powerful when performed with all available individual patient data, including data from trials that have not been published. In our meta-analysis, only published studies were included. As such, publication bias may have occurred, even though Begg's test and Egger's test were performed to access the publication bias and indicated no such bias existed. Second, the genotype 2/3 and genotype 4 groups were relatively small in size, and this may impact the generalisability of our conclusions. Future studies with larger groups of genotype 2/3 and genotype 4 patients, including those from different ethnicities, should be conducted to validate the associations with IL28B polymorphisms. Third, meta-analysis is an observational study that is subject to methodological deficiencies of the included studies. Fourth, the amount of published studies on genotype 2/3 and genotype 4 patients were insufficient to conduct ethnicity stratification analysis. Finally, it is very likely that non-genetic factors, such as viral load at treatment initiation and patient physical and physiological features, may influence SVR of antiviral-treated chronic HCV patients; however, very limited data on such potential factors were reported in the studies included in our analysis.

Nonetheless, our meta-analysis suggests that both IL28B rs12979860 CC and rs8099917 TT genotypes are strong predictors of SVR in PEG IFN-α/RBV-treated HCV genotype 1-infected patients, regardless of ethnicity. In contrast, rs12979860 CC has no predictive impact of SVR in genotype 2/3-infected patients, but rs8099917 TT is slightly associated with SVR in genotype 2/3-infected Asians.

Acknowledgements

Declaration of personal interests: None. Declaration of funding interests: This study was funded in full by the Major National Science and Technology Projects for Infectious Diseases (11th Five-Year Plan, China) (2008ZX10002-007), the National Natural Science Fund of China (Project Number 30872219) and Jiangsu Health International Exchange Program sponsorship (JSH-2011-051).

Ancillary