Background

Host interleukin-28B (IL28B) genotype is associated with spontaneous clearance of acute hepatitis C virus (HCV) infection, response to interferon (IFN)-based treatment, and with the development of hepatocellular carcinoma (HCC) in patients with chronic hepatitis C (CHC) (Thomas et al. 2009; Ge et al. 2009; Zhang et al. 2016). Genome-wide association studies have identified single-nucleotide polymorphisms (SNPs) in close proximity to the IL28B gene that encode for IFN lambda (IFN-λ), with strong associations with spontaneous or IFN-induced clearance of HCV (Thomas et al. 2009; Ge et al. 2009; Rauch et al. 2010; Suppiah et al. 2009; Tanaka et al. 2009). In particular, a host IL28B rs12979860 CC genotype is associated with the highest, and TT genotype the lowest, rates of response to IFN-based therapy (Thomas et al. 2009; Ge et al. 2009; Mangia et al. 2010; Asselah et al. 2012; De Nicola et al. 2012; Poordad et al. 2012; Lawitz et al. 2013; Susser et al. 2014). The T allele also increases the risk of HCC associated with HCV (Zhang et al. 2016). In addition, differences in the distribution of IL28B genotypes explain, in part, historical observations of low sustained virological response (SVR) rates in Black or Latino patients and high SVR rates in Asian patients (Muir et al. 2004; Yu et al. 2008; Rodriguez-Torres et al. 2009).

Despite the established role of host IL28B genotype as a predictor of response to therapy, the association between IL28B polymorphisms and the natural history of HCV is currently the subject of debate. CHC patients with an IL28B CC genotype had higher alanine transaminase (ALT) levels and greater hepatic necro-inflammatory activity at baseline, with worse clinical outcomes over 4 years of follow-up than patients with non-CC genotypes (Noureddin et al. 2013). However, the same study observed no difference in the rate of fibrosis progression between patients with CC and non-CC genotypes. Other studies have yielded contradictory results on the association between IL28B genotype and fibrosis progression (Asselah et al. 2012; Falleti et al. 2011; Fabris et al. 2011; Marabita et al. 2011; Di Marco et al. 2012; D’Ambrosio et al. 2014).

While the efficacy of recently introduced, IFN-free treatment regimens is not affected by host IL28B genotype, their use is limited to patients with advanced fibrosis in many jurisdictions (Lawitz et al. 2013; Jacobson et al. 2013). Therefore, predictors of fibrosis progression may provide a valuable tool for selection of patients eligible for treatment. In addition, dual peginterferon alfa/ribavirin therapy and protease inhibitor-based triple therapy continue to be used in many regions (Pawlotsky 2014). Given that advanced fibrosis is associated with lower rates of SVR to IFN-based therapies, the identification factors predictive of fibrosis stage has implications for treatment selection and optimization (Poordad et al. 2012; Lawitz et al. 2013; Hadziyannis et al. 2004; Bonnet et al. 2014; Manns et al. 2014; Jacobson et al. 2014).

The primary objective of the Gen-C study was to investigate associations between IL28B genotype and fibrosis stage in CHC patients, in addition to gathering further information on the distribution of IL28B genotypes by HCV genotype, geographic region and ethnicity. The final results from treatment-naive patients are reported here.

Methods

Patients and study design

Gen-C is a large, international, observational study of adults with CHC (clinicaltrials.gov, trial identifier NCT01675427). Patients were excluded if they had hepatitis B virus co-infection, a history of decompensated liver disease, major organ transplantation or end-stage renal disease. Only treatment-naive patients were included in the present analysis. Enrolment was at the discretion of the investigator.

The primary objective of the Gen-C study was to investigate the relationship between IL28B genotypes and liver fibrosis stage in patients with CHC.

Secondary objectives included relationships between IL28B genotypes and liver inflammation, ALT levels and patient demographics, and the distribution of IL28B genotypes by HCV genotype, geographic region and ethnicity.

Data collection

Patients who provided written, informed consent made one study visit for blood sample collection, which was analysed in a central laboratory in Germany or Italy.

A 5 ml sample of whole blood was collected in a tube containing ethylenediaminetetraacetic acid. After collection, the tube was gently inverted 10 times and was not to be centrifuged. A bar-coded label was then applied to the tube and the sample was stored at −20 °C before being packed in dry ice and shipped by courier to the central laboratory. Samples were stored at −20 °C until DNA extraction then were thawed overnight (i.e., for approximately 16 h) and rolled for approximately 30 min before opening.

Standard polymerase chain reaction technology was used to genotype SNPs near IL28B (rs12979860 [CC, TC and TT] and rs8099917 [TT, GT or GG]) and inosine triphosphate pyrophosphatase (ITPA) (rs1127354 [AA, CA or CC] and rs7270101 [CC, AC or AA]). Positive controls for IL28B rs12979860 TC and rs8099917 TG were obtained from TIB MOLBIOL GmbH, Berlin, Germany. Nucleic acid purification was performed using the MagNa Pure 96 System (Roche Diagnostics, Mannheim, Germany). Genomic DNA was extracted from a 50 µL sample by adding MagNa Pure96 to make up an eluate of 100 µL. Genotype was then determined by commercial genotyping assays (LightMix Kit rs12979860 IL28B, LightSNiP rs8099917 IL28B, LightSNiP rs7270101 ITPA, LightSNiP rs1127354 ITPA; Roche Diagnostics, GmbH) using an eluate of 5 µL.

The results of previous invasive or non-invasive fibrosis assessments were entered in the electronic Case Report Form. Investigators recorded the date, type (invasive or non-invasive) and result of the assessment. Fibrosis stage was documented categorically (Cirrhosis, Transition to cirrhosis, Advanced fibrosis-non-cirrhotic, Mild/minimal fibrosis, No fibrosis). Cirrhosis/transition to cirrhosis was defined by biopsy (Ishak 4–6, METAVIR 3–4, Batts and Ludwig 3–4, Knodell 3–4 or Scheuer 3–4) or non-invasive assessment.

Statistical analyses

Enrolment of 1500 treatment-naive and 1500 treatment-experienced patients with evaluable data was estimated to provide 80% power for a Chi square test to detect an association between liver fibrosis stage (five degrees of freedom) and IL28B genotype (three degrees of freedom) at a significance level of 0.05. The final enrolment target of 4000–6000 patients was arrived at after allowing for patient dropout and patients without evaluable data.

The analysis population was defined as all patients with blood samples and available data for IL28B SNPs. For analysis of the primary endpoint, patients were categorized as having cirrhosis/transition to cirrhosis or no cirrhosis. The relationship between cirrhosis status (cirrhosis/transition to cirrhosis vs no cirrhosis) and number of rs12979860 IL28B T alleles was investigated by Cochran-Armitage trend test. The relationship between IL28B genotype and all other fibrosis measurements (e.g. METAVIR fibrosis stage) was investigated using the Jonckheere-Terpstra trend test. The same methods were used to investigate the relationship between IL28B and other continuous, ordinal or binary variables, while the Pearson Chi square test was used to compare the association between IL28B genotypes and nominal variables.

Relationships between baseline characteristics and cirrhosis status were explored by multivariate logistic regression (MLR) analysis. Variables considered in the analysis were: alcohol consumption, ALT ratio, aspartate transaminase (AST) ratio, age, autoimmune disease, body mass index (BMI), body weight, ethnic origin, glucose intolerance/diabetes, HCV genotype, HCV RNA level, HIV–HCV co-infection, liver disease other than CHC, platelets, region of enrolment and years since HCV infection.

Results

A total of 3003 treatment-naive patients were enrolled at 213 centres across 30 countries between August 2011 (first patient) and September 2013 (last patient last visit) (Fig. 1). Overall, 87 patients were excluded, and therefore the analysis population comprised 2916 patients.

Fig. 1
figure 1

Patient flow diagram. aPatients may have been excluded for more than one reason

Full size image

Among the total population of 2916 patients, a fibrosis assessment was conducted in 2902 individuals (99.5%), of whom 26.1% were diagnosed with cirrhosis/transition to cirrhosis (Table 1). The majority of patients were male, Caucasian and had HCV G1 infection, with similar characteristics across IL28B rs12979860 genotypes (Table 1). More patients with CC or CT compared with TT genotypes had a high viral load and elevated ALT levels. The distribution of HCV genotypes varied by region (Fig. 2).

Table 1 Patient demographics and background disease characteristics
Full size table
Fig. 2
figure 2

Distribution of IL28B rs12979860 genotype by HCV genotype and geographic region. The distribution of HCV genotype showed significant variation (p < 0.0001) by region of enrolment. p < 0.0001b for association between HCV genotype and region of enrolment. aPatients with multiple genotypes were assigned to disjoint groups according to the following hierarchy: G1, G4, G3, G2, other bPearson’s Chi square test

Full size image

IL28B genotype and liver fibrosis stage

Among HCV G1-infected patients, the prevalence of cirrhosis/transition to cirrhosis increased with the number of rs12979860 T alleles overall (p = 0.0018) (Fig. 3a), in Caucasians (p = 0.0010) (Fig. 3b) and in European Caucasians (p = 0.0023) (Fig. 3c).

Fig. 3
figure 3

The association between IL28B rs12979860 genotype and liver fibrosis by HCV genotype overall (a), in Caucasian patients with HCV G1 infection (b), and in European Caucasian patients with HCV G1 infection (c). Significant associations were found between IL28B rs12979860 genotype and liver fibrosis in patients with HCV G1 infection (p = 0.0018) (a), in Caucasian treatment-naive patients with HCV G1 infection (p = 0.0010) (b), and in treatment-naive European Caucasian patients with HCV G1 infection (p = 0.0023) (c). aCochran-Armitage Trend Test for a trend in binomial proportions across the three IL28B genotypes

Full size image

No statistically significant associations were observed for other HCV genotypes. Among HCV G1 patients enrolled at European study sites (n = 1245), an association was observed between the prevalence of cirrhosis/transition to cirrhosis and the number of rs12979860 T alleles (p = 0.0030, Table 2). For patients enrolled in Asia, Latin America or the Middle East, no significant associations were observed (Table 2; Additional file 1: Tables S1–S4). However, these results should be interpreted with caution due to low patient numbers.

Table 2 Association between liver fibrosis and IL28B rs12979860 genotype in patients with HCV G1 infection, by region of enrolment
Full size table

MLR analyses

Older age, higher BMI, HCV G1 infection (vs G2), higher AST ratio, lower platelet count, enrolment at an Asian/Middle Eastern site (vs European), liver disease other than CHC and glucose intolerance/diabetes were significantly associated with an increased risk of cirrhosis/transition to cirrhosis (Fig. 4a). IL28B genotype was not associated with cirrhosis/transition to cirrhosis in the final MLR model. In contrast, when only HCV G1-infected patients were considered, rs12979860 genotype (CT or TT vs CC) was significantly associated with cirrhosis/transition to cirrhosis (Fig. 4b). When the analysis was further restricted to European HCV G1 patients, rs8099917 GG and TG genotype (vs TT) was significantly associated with cirrhosis/transition to cirrhosis (Fig. 4c).

Fig. 4
figure 4

Multiple logistic regression analysis for cirrhosis or transition to cirrhosis in all patients (a), all patients with HCV G1 infection (b), and all European Caucasian patients with HCV G1 infection (c). AST aspartate transaminase; MLR multiple logistic regression

Full size image

IL28B rs12979860 genotype, serum ALT levels and necro-inflammatory grade

A statistically significant decrease in mean ALT ratio was observed with an increasing number of rs12979860 T alleles for G1 (p = 0.0007), G2 (p = 0.0061) and G3 (p < 0.0001) but not G4 patients (Additional file 1: Table S5).

No association was observed between IL28B genotype and METAVIR fibrosis stage or necro-inflammatory grade (Additional file 1: Tables S6 and S7), or liver stiffness (Additional file 1: Table S8) overall or when stratified by HCV genotype. There was however significant association between IL28B genotype and AST to platelet ratio index (APRI) score and Fibrosis-4 (FIB-4) score (age, AST, platelet count and ALT) in the overall population and in the subgroup of patients with G2 and G3 infection (Additional file 1: Tables S9 and S10).

rs12979860 genotype distributions

In general, an rs12979860 CT genotype was more common than a CC or TT genotype (Fig. 5a) and there was a significant association between HCV genotype and IL28B genotype (p < 0.0001), with the CC genotype occurring more frequently in G2 and G3 patients than in G1 and G4 patients.

Fig. 5
figure 5

Distribution of IL28B rs12979860 genotype according to HCV genotype. The distribution of IL28B rs12979860 genotype varied significantly by HCV genotype (a), region of enrolment (b), and ethnicity (c). p < 0.0001a for the association between HCV genotype and IL28B rs12979860 genotype, p < 0.0001a for the association between geographic region and IL28B rs12979860 genotype, and p < 0.0001a for the association between ethnicity and IL28B rs12979860 genotype. aPearson’s Chi square test

Full size image

rs12979860 genotype was significantly associated with region of enrolment overall (p < 0.0001, Fig. 5b) and in HCV G1 and G4 patients [p < 0.0001 and p = 0.0467, respectively (Table 3)]. rs12979860 distribution was not, however, homogenous across the countries within each region (Additional file 1: Table S8).

Table 3 Distribution of IL28B rs12979860 genotype by region of enrolment and HCV RNA genotype
Full size table

rs12979860 genotype was significantly associated with patient ethnic origin (p < 0.0001); CC was the most common genotype in Asian patients, but was infrequent in Black patients (Fig. 5c).

Discussion

The primary finding of the Gen-C study was a positive association between an increasing number of rs12979860 T alleles and the prevalence of cirrhosis/transition to cirrhosis in treatment-naïve G1 patients. While the positive association between T allele and fibrosis stage observed for HCV G1 broadly supports findings from several previous studies, it should be taken into account that these studies did not differentiate between HCV genotypes, and also included relatively small numbers of patients (Falleti et al. 2011; Fabris et al. 2011; Di Marco et al. 2012).

There is also evidence that contradicts the results of the present study: analyses in G1 patients have reported that a CC genotype is associated with a higher prevalence of cirrhosis/transition to cirrhosis (Abe et al. 2010), and that there is no relationship between IL28B and fibrosis stage (D’Ambrosio et al. 2014; Bochud et al. 2012). The studies by D’Ambrosio et al. and Abe et al. measured fibrosis by biopsy; however, they included relatively small numbers of patients (D’Ambrosio et al. 2014; Abe et al. 2010). While the study by Bochud et al. (2012) included a similar number (n = 919) of Caucasian G1-infected, treatment-naive patients, the median duration of HCV infection was longer (21 vs ~14 years), and more patients had HIV–HCV co-infection (5 vs 3.5%). Two further studies found no association between rs12979860 genotype and fibrosis stage (Noureddin et al. 2013; Marabita et al. 2011).

Associations between IL28B genotype and liver fibrosis were restricted to HCV G1 infection in the present study. When the data were analysed by geographic region, the association between IL28B genotype and fibrosis status remained statistically significant in the large subgroup of European Caucasian patients with G1 infection but was not statistically significant in other populations. The lack of a relationship in the smaller subgroup of Asian, Latin American and Middle Eastern patients with G1 infection may simply be due to the smaller sample size, or may be due to factors related to ethnicity that were not measured in this analysis. Interestingly there was also a significant association between IL28B genotype and APRI score and FIB-4 score in the overall population and in patients with G2 and G3 infection, but not in those with G1 infection. Strong conclusions cannot be drawn regarding the overall association between IL28B genotype and fibrosis status, as significant associations between C allele frequency and cirrhosis/transition to cirrhosis have been reported previously (Bochud et al. 2012; Rembeck et al. 2012).

The primary immune response to HCV infection is mediated predominantly through IFN-λ cytokines and may be influenced by IL28B genotype (Thomas et al. 2012; Watanabe et al. 2013). Patients with a “favourable” rs12979860 CC genotype are more likely to experience spontaneous viral clearance and respond to IFN-based therapy (Thomas et al. 2009; Ge et al. 2009), while those with the T allele may be at increased risk of HCV-related HCC (Zhang et al. 2016). Interestingly, there is a well-established association between this “favourable” genotype for viral clearance (CC), and increased markers of intra-hepatic inflammatory response to HCV infection (Noureddin et al. 2013; D’Ambrosio et al. 2014; Abe et al. 2010; Bochud et al. 2012; Rembeck et al. 2012). The presence of significantly higher necro-inflammatory activity and serum ALT levels in patients with CC genotypes has been interpreted as an indication of an enhanced immune response in patients with CC genotypes as compared with non-CC genotypes (Noureddin et al. 2013; D’Ambrosio et al. 2014; Rembeck et al. 2012). A more vigorous immune response would be expected to lead to more rapid fibrosis progression over time in patients who do not experience spontaneous viral clearance. However, an association with liver fibrosis stage and IL28B genotype is less well established (Noureddin et al. 2013; D’Ambrosio et al. 2014; Abe et al. 2010; Bochud et al. 2012; Rembeck et al. 2012). In the current study, the CC genotype was associated with higher ALT ratios and a lower prevalence of cirrhosis/transition to cirrhosis in patients with HCV G1 infection. These results support the interpretation that the “favourable” CC genotype is associated with a more vigorous antiviral immune response but not with higher incidence of fibrosis. Unfortunately, inflammatory grade was not available for many patients, which limits the strength of this conclusion.

Recently, it has been shown that different combinations of IL28B genotype and toll-like receptor-2 (TLR-2) variants are associated with different manifestations of HCV-related liver disease and lymphoproliferative diseases (De Re et al. 2016). The IL28B CC genotype exerts a protective effect on CHC and the development of cirrhosis and HCC, while the TLR-2 del/del genotype was associated with an increased risk of HCC. In combination, the presence of one or more TLR-2 del alleles abolished the protective effect of the CC genotype. These data suggest that IL28B and TLR-2 are functionally interconnected in HCV disease-specific phenomena.

Baseline factors predictive of cirrhosis/transition to cirrhosis are similar to those reported elsewhere (Noureddin et al. 2013; Falleti et al. 2011; Bochud et al. 2012; Rueger et al. 2015; Wright et al. 2003). It should be noted that attendance at a clinic in Asia (vs Europe) was significantly associated with cirrhosis/transition to cirrhosis in G1 patients, which seems counterintuitive given that the CC genotype was more prevalent in Asian patients. However, this result may be explained by the greater proportion of Asian than European patients with cirrhosis at baseline (~53 vs ~25%).

Overall, the Gen-C study supports previously published frequency distributions of IL28B genotypes. The CC genotype was less common in G1 and G4 than G2 and G3 patients, and the proportions of CC patients in each HCV genotype agreed with previous estimates (De Nicola et al. 2012; Falleti et al. 2011; Bochud et al. 2012; Mottola et al. 2015). The geographic distribution of rs12979860 genotypes also agrees with previous reports, with the highest prevalence of the C allele in Asia, and the highest prevalence of the TT genotype in Latin America (Thomas et al. 2009; Ge et al. 2009). Finally, ethnic associations with s12979860 genotypes agree with previous reports, with the lowest prevalence of the C allele in Black patients, and the highest in Asian patients (Thomas et al. 2009; Ge et al. 2009).

Limitations include the small amount of METAVIR data collected, which tempers conclusions regarding IL28B genotype, inflammatory grade and fibrosis stage. Rapid fibrosis progression is associated with factors other than IL28B genotype, such as alcohol consumption, liver disease other than CHC, HIV–HCV co-infection and glucose intolerance, which, notwithstanding our MLR analysis, could have confounded the results.

In summary, an increasing number of rs12979860 T alleles is associated with an increased prevalence of cirrhosis/transition to cirrhosis in treatment-naïve patients with HCV G1 infection. When the G1 population was categorized geographically, this association was evident in the large subgroup of European Caucasians with HCV G1 infection but not in the smaller subgroups of G1-infected patients from Asia, Latin America or the Middle East. Thus, the presence of the T allele may be used to prioritize treatment for G1 patients who are at increased risk of progressing to cirrhosis.