Persistent infection with the hepatitis B virus (HBV) is associated with an increased risk of cirrhosis, hepatocellular carcinoma (HCC), and anticipated liver-related mortality worldwide.1-4 To attenuate or prevent these long-term sequelae of HBV, interruption of HBV replication by means of inhibitors of HBV polymerase nucleos(t)ide analogs or interferon (IFN) is the only practical approach. IFN therapy, which is recommended as a first-line therapy in patients with less-advanced liver disease and moderately replicating HBV. However, IFN has limited application in virtue of its suboptimal tolerability and restricted antiviral activity in general, and this is particularly true in the hepatitis B e antigen (HBeAg)-negative patients who predominate in the Mediterranean area.2-14 These limitations notwithstanding, IFN therapy has gained popularity to treat HBV, because several permanent responders will ultimately achieve seroclearance of hepatitis B surface antigen (HBsAg), which is a proxy of a cure of the infection.14, 15 In view of its limited therapeutic index, many efforts have been made to optimize IFN treatment through pretreatment patient selection based on outcome predictors, but the use of both host- and virus-related variables that predict a response to therapy have not proved to be clinically useful. Indeed, constitutional variables, such as gender and age, and such disease-related predictors as elevated baseline transaminases, low baseline viremia, and moderate liver cell inflammation16 are either underrepresented in our geographical area (e.g., age and gender) or may vary with time (e.g., HBV DNA and liver inflammation), thus precluding their systematic use in pretreatment patient stratification. The polymorphisms near the IL28B gene, which encodes for IFN-λ-3 and may be implicated in the modulation of innate immune response, were found to be significantly associated with treatment-related resolution of chronic viral hepatitis C (CHC) and offer a chance to improve therapeutic algorithms of patients with CHC based on IFN.17 Because hepatitis C patients with the rs12979860 CC genotype of interleukin (IL)28B, representing approximately 30% of the Caucasian population, have the greatest chances of a sustained virological response (SVR) to IFN, compared to those carrying the T allele,18-23 we asked whether IL28B polymorphisms also have any predictive value for IFN therapy outcome in the difficult-to-cure HBeAg-seronegative patients with chronic hepatitis B (CHB). To answer this question, we retrospectively analyzed, for this genetic polymorphism, 101 patients with CHB who were followed up for a median of 11 years after a course of IFN and then correlated findings with permanent seroclearance of HBsAg.
Interleukin (IL)28B polymorphisms have been associated with interferon (IFN)-induced viral clearance in patients with chronic hepatitis C. Whether this is also true for patients with the difficult-to-cure hepatitis B e antigen (HBeAg)-negative chronic hepatitis B (CHB) is unknown. One hundred and one HBeAg-negative patients (92% genotype D) with compensated CHB (84% males, 46 years; hepatitis B virus [HBV] DNA: 6.0 log cp/mL; alanine aminotransferase [ALT]: 136 IU/L; 42% with cirrhosis) were followed up for a median of 11 years (range, 1-17) after a median of 23 months (range, 10-48) of either standard or pegylated (Peg)-IFN-alpha therapy. A post-treatment response was defined as hepatitis B surface antigen (HBsAg) clearance with or without antibody to hepatitis B surface antigen (anti-HBs) seroconversion. The rs12979860 (C>T) genotype in the IL28B locus was assessed in serum samples by using Custom TaqMan SNP Genotyping Assays (Applied Biosystems, Carlsbad, CA). During a median of 11 years of post-treatment follow-up, 21 patients (21%) cleared serum HBsAg, including 15 who developed >10 IU/mL of anti-HBs titers. Forty-eight patients (47%) had CC genotype, 42 (42%) had CT, and 11 (11%) had TT, with the allelic frequency being 68% for C allele and 32% for T allele. The rate of serum HBsAg clearance was 29% (n = 14) in CC compared to 13% (n = 7) in non-CC, genotype carriers (P = 0.039). Baseline HBV DNA levels <6 log cp/mL (odds ratio [OR], 11.9; 95% confidence interval [CI]: 2.8-50.6; P = 0.001), ALT levels >136 IU/L (OR, 6.5; 95% CI: 1.8-22.5; P = 0.003), duration of IFN (OR, 1.16; 95% CI: 1.02-1.31; P = 0.021), and genotype CC (OR, 3.9; 95% CI: 1.1-13.2; P = 0.025) independently predicted HBsAg clearance. Conclusions: IL28B polymorphism is an additional predictor of off-therapy IFN-related HBsAg seroclearance to be used in the pretreatment stratification of HBeAg-negative patients chronically infected by genotype D of HBV. (HEPATOLOGY 2013)
Patients and Methods
From an initial cohort of 168 consecutive HBeAg-negative CHB patients treated with IFN-based therapy with regular follow-up at our liver unit, 101 individuals were selected who had completed a course of IFN-α-2b (Intron A; Schering-Plough, Milan, Italy), IFN-α-2a (Roferon; Roche, Milan, Italy) 6 MU three times a week (n = 69) or Peg-IFN-α-2a at 180 μg per week (PEGASYS; Roche) (n = 32) with a validated outcome and available repository DNA. Of these, 67 were excluded from analysis because of an incomplete course of IFN (n = 13), lack of a validated outcome (n = 7), missing clinical or virological data or follow-up (n = 45), or absence of adequate stored human cell DNA material for the genetic analysis or refusal to provide genetic testing consent (n = 2). All patients were positive for HBsAg and hepatitis e antigen (anti-HBe) antibody, but negative for HBeAg. Entry criteria included an HBV DNA level of >10.000 copies/mL and an alanine aminotransferase (ALT) level >1 times, but ≤10 times, the upper limit of normal. Patients were excluded if they tested positive for antibody to hepatitis D virus (HDV) or hepatitis C virus (HCV), human immunodeficiency virus (HIV), or had been treated previously with IFN. Exclusion criteria for antiviral treatment were as follows: pregnancy or lactation; drug or alcohol abuse (>40 g ethanol/day); presence of HCC; platelet count less than 100,000/mm3; white blood cell count less than 3,000/mm3; serum markers of autoimmunity; renal failure; previous episodes of hepatic decompensation; esophageal varices; or other serious medical illnesses. The extent of liver fibrosis was measured by liver biopsy according to Ishak's score.24
In all patients, ALT, aspartate aminotransferase (AST), and HBV DNA levels were measured every 3 months during treatment and every 6 months thereafter. HBsAg and anti-HBs were measured every 3 months during treatment and every year thereafter. We defined treatment outcomes according to HBsAg clearance with or without HBsAg seroconversion after the end of treatment or during follow-up. An end of therapy, response was undetectable viremia by a non–polymerase chain reaction (PCR)-based assay or HBV DNA <2,000 IU/mL by PCR assay. An SVR was HBV DNA suppression, as previously defined, starting 6 months after completion of therapy.
To participate in the study, written informed consent for genetic testing, including IL28B polymorphisms, was obtained from all individuals, and the study protocol was evaluated and approved by the local institutional review committee.
HBsAg, antibody to HBsAg (anti-HBs), HBeAg, and anti-HBe were detected by a microparticle enzyme immunoassay (AXSYM; Abbott Laboratories, North Chicago, IL). Antibody to HDV was assessed by enzyme-linked immunosorbent assay (Sorin Biomedica, Saluggia, Italy). Antibody to HIV was detected by HIV-1 third-generation assay (AXSYM HIV 1/2; Abbott Laboratories), and antibody to HCV was detected by a second-generation test system (Ortho Diagnostic Systems, Raritan, NJ). Until 2007, serum HBV DNA level was detected by non-PCR assays (Digene Hybrid Capture System; Digene Diagnostics, Inc., Beltsville, MD) and branched DNA (Bayer Corp., Tarrytown, NJ), and starting from 2007, HBV DNA was measured by a real-time PCR assay (COBAS TaqMan; Roche), which has a lower limit of quantification of 12 IU/mL. HBV genotypes were retrospectively looked for from frozen serum sample by line-probe assay (INNO-LiPA HBV Genotyping; Innogenetics N.V., Ghent, Belgium), containing specific probes for the six major genotypes (A-F). ALT and AST levels were measured by automated optimized colorimetry at 37°C.
rs12979860 Single-Nucleotide Polymorphism Genotyping.
Genotyping was conducted in a blinded fashion, relative to HBV treatment status and other characteristics, on DNA specimens collected from each individual, using the 5' nuclease assay with allele-specific TaqMan probes (TaqMan SNP Genotyping Assay on a 7900HT real-time PCR instrument; Applied Biosystems, Carlsbad, CA).
Data were expressed as counts and percentages for qualitative variables and as median and range for discrete variables. Comparisons between groups were carried out using nonparametric analysis of variance. Associations between different qualitative parameters were explored using the chi-square test or Fishers's exact test, as appropriate. Univariable (i.e., chi-square test) and multivariable (i.e., logistic regression) analyses were used to determine predictors of response. Primary analysis of the association between rs12979860 and treatment response considered the following: age; gender; serum ALT levels; serum HBV DNA; HBV genotype; cirrhosis; duration of treatment; and type of IFN. Gender, cirrhosis, type of IFN, and genotype were included as categorical variables; age, serum ALT, duration of treatment, and serum HBV DNA were included as continuous variables. Those pretreatment factors found to be significant in the overall logistic regression model were assessed in stratified analyses to show rates of response across different categories of these characteristics. Statistical analysis was done with STATA software (Stata Statistical Software: Release 7.0; StataCorp LP, College Station, TX).
All patients were of European ancestry (Table 1), most were men, infected with genotype D of HBV and with advanced liver fibrosis, and IFN was given for a median of 23 months (range, 10-48). The IL28B polymorphism was genotype CC in 48 patients, CT in 42, and TT in 11, with the allelic frequency being 68% for C allele and 32% for T allele. Overall, CC, CT, and TT carriers were comparable in terms of age, gender, serum ALT levels, serum HBV DNA, HBV genotype, cirrhosis, duration of treatment, type of IFN, and duration of follow-up (Table 2). The proportion of patients with <12, 12-24, and >24 months of treatment was similar in the three IL28B groups: 8%, 61%, and 31% for the CC group; 14%, 69%, and 17% for the CT group; and 9%, 45.5%, and 45.5% for the TT group (CC versus CT versus TT, P = 0.28; CC versus non-CC, P = 0.52).
|Age, years*||46 (26-63)|
|Males, no. (%)||85 (84)|
|Cirrhosis, no. (%)||42 (42)|
|ALT, IU/L*||136 (32-932)|
|HBV DNA, log10 copies/mL*||6.0 (3.3-9.0)|
|HBV genotype D, no. (%)||93 (92)|
|IFN standard, no. (%)||69 (68)|
|Treatment duration, months*||23 (10-48)|
|Post-treatment follow-up, years*||11 (1-17)|
|IL28B genotype (%)|
|Patient Features||CC (n = 48)||CT (n = 42)||TT (n = 11)||P Value†||P Value‡|
|Age, years*||45 (26-61)||47 (28-63)||48 (29-62)||0.31§||0.60§|
|Males, no. (%)||38 (79)||37 (88)||10 (91)||0.15∥||0.41¶|
|Cirrhosis, no. (%)||20 (42)||19 (45)||3 (27)||0.57∥||0.56¶|
|ALT, IU/L*||153 (32-720)||135 (32-932)||88 (44-235)||0.07§||0.08§|
|HBV DNA, log10 copies/mL*||6.3 (3.3-9.0)||6.0 (4.0-9.0)||5.3 (5.0-7.9)||0.58§||0.54§|
|HBV genotype D, no. (%)||45 (94)||38 (90)||10 (91)||0.41∥||0.84¶|
|IFN standard, no. (%)||32 (67)||29 (69)||8 (73)||0.44∥||0.92¶|
|IFN treatment, months*||23 (10-48)||23 (12-33)||22 (12-32)||0.60§||0.78§|
|Post-treatment follow-up years*||11 (1-16)||12 (1-17)||11 (2-16)||0.86§||0.98§|
End-of-treatment response as well as SVR were significantly higher in CC than in non-CC patients (end-of-treatment response: 69% versus 45%, P = 0.014; SVR: 31% versus 13%, P = 0.025) (Table 3). At univariate analysis, duration of IFN (odds ratio [OR], 1.09; 95% confidence interval [CI]: 1.01-1.17; P = 0.02) and genotype CC (OR, 2.86; 95% CI: 1.26-6.49; P = 0.01) were associated with an increased likelihood of end-of-treatment response. Baseline HBV DNA levels (OR, 0.70; 95% CI: 0.50-0.98; P = 0.03), duration of IFN (OR, 1.09; 95% CI: 1.00-1.18; P = 0.03), and genotype CC (OR, 2.99; 95% CI: 1.24-7.21; P = 0.01) independently predicted end-of-treatment response. An SVR was associated with age (OR, 0.94; 95% CI: 0.89-0.99; P = 0.02), HBV DNA levels (OR, 0.53; 95% CI: 0.34-0.84; P = 0.007), duration of post-treatment follow-up (OR, 1.10; 95% CI: 1.00-1.21; P = 0.03), and CC genotype (OR, 2.98; 95% CI: 1.09-8.13; P = 0.03). At multivariate analysis, age (OR, 0.92; 95% CI: 0.87- 0.98; P = 0.01), baseline HBV DNA levels (OR, 0.44; 95% CI: 0.27-0.74; P = 0.002), and genotype CC (OR, 3.7; 95% CI: 1.19-11.5; P = 0.02) independently predicted an SVR.
|Response||CC (n = 48)||CT/TT (n = 53)||P Value|
|End-of-therapy virological response (%)||33 (69)||24 (45)||0.01|
|SVR (%)||15 (31)||7 (13)||0.02|
|HBsAg clearance (%)||14 (29)||7 (13)||0.04|
In the post-treatment follow-up, 95 patients (94%) were seen at least every year at the outpatient clinic, and 75 (74%) fulfilled the predefined strict post-treatment surveillance program of yearly HBsAg testing. Among the 26 patients with suboptimal HBsAg testing rates, 16 (62%) had this marker being performed for at least 5 years. During a median of 11 years of post-treatment follow-up, 21 patients (21%) cleared HBsAg and 15 developed >10 IU/mL of anti-HBs titers. HBsAg clearance rates were 29% (14 patients) in CC, 7% (3 patients) in CT, and 36% (4 patients) in TT.
By univariate analysis, baseline HBV DNA, ALT, duration of treatment, post-treatment follow-up, and IL28B polymorphisms, but not age, gender, cirrhosis, HBV genotype, and type of IFN, were significantly associated with HBsAg clearance (Table 4). At multivariate analysis, baseline HBV DNA levels (OR, 0.31; 95% CI: 0.15-0.62; P = 0.001), ALT levels (OR, 1.0; 95% CI: 1.0-1.0; P = 0.03), duration of IFN therapy (OR, 1.20; 95% CI: 1.04-1.39; P = 0.012), and genotype CC (OR, 3.6; 95% CI: 1.05-12.5; P = 0.04) predicted HBsAg seroclearance (Table 5). When baseline levels of HBV DNA and ALT were expressed as dichotomous variables by using median values, HBV DNA levels <6 log cp/mL (OR, 11.9; 95% CI: 2.8-50.6; P = 0.001), ALT levels >136 IU/L (OR, 6.5; 95% CI: 1.8-22.5; P = 0.003), duration of IFN therapy (OR, 1.16; 95% CI: 1.02-1.31; P = 0.021), and genotype CC (OR, 3.9; 95% CI: 1.1-13.2; P = 0.025) were confirmed to predict HBsAg seroclearance. By combining HBV DNA, ALT, and IL28B genotype, patients were classified into four different groups characterized by increasing rates of HBsAg loss (Table 6).
|Patient Features||HBsAg Seroclearance||P Value|
|Yes (n = 21)||No (n = 80)|
|Age, years*||40 (26-62)||47 (26-63)||0.11†|
|Males, no. (%)||17 (81)||68 (85)||0.43‡|
|Cirrhosis, no. (%)||9 (43)||33 (41)||0.54‡|
|ALT, IU/L*||201 (45-720)||129 (32-932)||0.04†|
|HBV DNA, log10 copies/mL*||5.0 (4.0-9.0)||6.3 (3.3-9.0)||0.002†|
|HBV genotype D, no.* (%)||19 (90)||74 (92)||0.53‡|
|IFN standard, no. (%)||18 (86)||51 (63)||0.05‡|
|Duration of IFN treatment, months*||24 (16-48)||22 (10-33)||0.01†|
|Post-treatment follow-up, years*||12 (1-16)||7 (1-17)||0.02†|
|IL28B CC genotype (%)||14 (67)||34 (43)||0.04‡|
|Baseline Variable||End-of-Treatment Response OR (95% CI); P Value||SVR OR (95% CI); P Value||HBsAg Clearance OR (95% CI); P Value|
|Age, years||1.01 (0.97-1.05); 0.54||0.94 (0.89-0.99); 0.02||0.92 (0.87-0.98); 0.01||0.95 (0.90-1.00); 0.10|
|Male||0.70 (0.23-2.12); 0.53||0.80 (0.23-2.80); 0.73||0.75 (0.21-2.61); 0.65|
|Cirrhosis||0.58 (0.26-1.29); 0.18||0.75 (0.28-2.00); 0.57||1.06 (0.40-2.82); 0.89|
|ALT levels, IU/L||1.00 (0.99-1.00); 0.61||1.00 (0.99-1.00); 0.16||1.00 (1.00-1.00); 0.02||1.00 (1.00-1.00); 0.03|
|HBV-DNA levels, log cp/mL||0.74 (0.54-1.01); 0.06||0.70 (0.50-0.98); 0.03||0.53 (0.34-0.84); 0.007||0.44 (0.27-0.74); 0.002||0.53 (0.33-0.84); 0.008||0.31 (0.15-0.62); 0.001|
|HBV genotype D||1.26 (0.29-5.38); 0.74||0.82 (0.15-4.38); 0.81||0.77 (0.14-4.12); 0.76|
|IFN standard||0.53 (0.22-1.28); 0.16||1.76 (0.58-5.30); 0.31||3.41 (0.92-12.5); 0.06|
|IFN treatment, months||1.09 (1.01-1.17); 0.02||1.09 (1.00-1.18); 0.03||1.07 (0.99-1.17); 0.08||1.16 (1.04-1.30); 0.008||1.20 (1.04-1.39); 0.01|
|Post-treatment follow-up, years||1.00 (0.93-1.07); 0.90||1.10 (1.00-1.21); 0.03||1.11 (1.01-1.22); 0.02|
|CC genotype IL28B||2.86 (1.26-6.49); 0.01||2.99 (1.24-7.21); 0.01||2.98 (1.09-8.13); 0.03||3.72 (1.19-11.5); 0.02||2.70 (0.98-7.42); 0.04||3.63 (1.05-12.5); 0.04|
|Patient Features||CC (%)||CT/TT (%)|
|HBV DNA low and ALT high||60||38|
|HBV DNA low and ALT low||28||11|
|HBV DNA high and ALT high||23||0|
|HBV DNA high and ALT low||0||0|
Optimization of IFN treatment of HBeAg-negative patients with CHB, based on an accurate pretreatment patient selection by response predictors, might help in overcoming such barriers to treatment as the need for parenteral administration, limited effectiveness, and poor tolerability. Our findings of an association between IL28B polymorphisms and likelihood of an SVR to IFN in the difficult-to-cure patients with CHB partially meet these expectations. Indeed, we demonstrated increased rates of SVR and HBsAg seroclearance in HBeAg-negative patients with CHB who carry the CC genotype of rs12979860 IL28B, compared to those carrying either the CT or TT genotype. Thus, our findings suggest that IL28B polymorphism adds to the already known list of pretreatment predictors of IFN therapy outcome that might be used for optimizing the management of the difficult-to-cure HBeAg-negative patients with chronic infection with the genotype D of HBV. This genetic pretreatment predictor, in fact, has major advantages over classical constitutional or virus-related predictors of treatment outcome that, as with age and gender, are either underrepresented in our geographical area or, as with viremia and ALT, may vary during the natural course of hepatitis, thus imposing a close patient monitoring to detect the appropriate timing of IFN therapy.14, 16, 25 Because in our geographical regions the less IFN-sensitive genotype D of HBV prevails among HBeAg-negative CHB, pretreatment patient stratification by a trustable, user-friendly treatment-outcome predictor might improve the cost-effectiveness ratio of IFN therapy. This might be the case of IL28B polymorphisms in the prediction of IFN-associated HBsAg seroclearance in HBeAg-negative patients with CHB, because rates of seroclearance were 3.9-fold higher in CC patients than in T-allele carriers (TT and CT). Our findings that baseline ALT and HBV DNA levels predicted an SVR to IFN independently on IL28B not only confirmed previous studies in patients with CHB, but also clearly indicates that our study was not biased by the retrospective selection of patients, thus well representing the complex pattern of the practice of HBV in our region. Interestingly, the combination of baseline values of ALT, HBV DNA, and IL28B polymorphisms allowed us to identify a set of patients with a high chance (60%) of losing serum HBsAg after IFN therapy. Although we acknowledge that the identification of a predictor of IFN treatment outcome in such a difficult-to-cure set of patients as HBeAg-seronegative patients infected by genotype D of HBV is clinically relevant, we also recognize that the prediction power of this combined set of variables is somehow diluted in an 11-year off-therapy period. On the other hand, no other predictor of IFN therapy outcome of comparable power is available to optimize the management of HBeAg-negative patients with CHB, including the on-therapy kinetics of serum HBsAg, which better performs as a negative predictor of treatment response.26, 27
The predictive value of IL28B has been recently documented also in the different scenario of HBeAg-positive patients with CHB from North Europe and Asia, where patients with other than D genotypes of HBV predominate.28 In that study, IL28B polymorphism was, in fact, positively associated with both HBeAg seroconversion and HBsAg seroclearance, supporting the existence of a genetic control of anti-HBV response to IFN, operating across a number of clinical variables, including the phase of infection (i.e., HBeAg versus anti-HBe) and disease severity, as shown by the high rate (42%) of patients with cirrhosis present in our study. One possible explanation for the association between IL28B and IFN response is the role of IL28B in the activation of the antiviral cascade by Janus kinase/signal transducer and activator of transcription, a very well-characterized event in the HCV scenario.29 In difficult-to-cure patients with genotype 1 of HCV, the IL28B CC genotype was associated with an early decline of serum HCV-RNA, heralding an SVR in most of the patients.21 Along the same line, it is worth noting that CC patients with the difficult-to-cure HBeAg-negative CHB have a faster, early decline of serum HBV DNA after IFN therapy, compared to T-allele carriers, a phenomenon that could herald faster seroclearance of HBsAg in CC patients (Prof. Alberti, A., Department of Molecular Medicine, University of Padova, personal communication).
We acknowledge that the strengths of our study are counterbalanced by a number of limitations. The homogeneous criteria for patient selection, the extended duration of IFN therapy, the 11 years of median post-treatment follow-up, and the hard endpoint of HBsAg seroclearance strongly support our conclusions about the predictive value of IL28B polymorphisms. On the other hand, our findings need to be externally validated in a larger cohort of patients prospectively followed up in whom seroclearance of HBsAg can be tested by a time-dependent multivariate analysis, including those receiving Peg-IFN therapy. The predictive power of IL28B polymorphism needs also to be investigated in HBeAg-negative patients with other than genotype D infection. Indeed, in the HCV scenario, the predictive value of IL28B polymorphisms appeared to be attenuated in patients with easy-to-cure genotypes HCV 2 and 3, compared to the more difficult-to-cure HCV 1 and 4, and in patients receiving potentiated IFN therapy by combination with direct antiviral agents.22, 30, 31
All in all, our findings indicate that the IL28B polymorphism is a reliable predictor of IFN therapy outcome in HBV, which might be used for pretreatment stratification aimed at optimizing the treatment of such a difficult-to-cure patient population as HBeAg-negative carriers of genotype D of HBV.
The authors thank Anna Molinari for her expert secretarial assistance.