Clinical utility of interleukin-28B testing in patients with genotype 1


  • Potential conflict of interest: Dr. Afhdal consults for, advises, and received grants from Gilead, Merck, and Vertex.

See Editorial on Page 5

Over the last 2 years exciting novel pharmacogenetic data have emerged on the single nucleotide polymorphisms (SNPs) in the region of the interleukin-28B (IL28B) gene as a predictor of response to the treatment of chronic hepatitis C infection. The CC IL28B genotype has been found to be highly predictive of response to therapy with pegylated interferon (PEG-IFN) and ribavirin (RBV). We will address the predictivity and utility of IL28B testing in patients infected with chronic genotype 1 hepatitis C virus (HCV) as well as the variation in utility and implications with ethnicity in this population of patients. The clinical utility of IL28B testing in patients with genotype 1 needs to be placed in the context of the currently available treatments that include PEG-IFN and RBV.


cEVR, complete early virological response; EVR, early virological response; IL28B, interleukin-28B; PEG-IFN, pegylated interferon; RBV, ribavirin; RVR, rapid virological response; SNP, single nucleotide polymorphism; SVR, sustained virological response.

Baseline Predictors for Response to PEG-IFN and RBV

In most of the world where direct-acting antivirals are still not available, the current standard of care for chronic infection with genotype 1 HCV is 48 weeks of therapy with PEG-IFN and RBV. Response to therapy is variable and influenced by certain baseline viral and host characteristics. These baseline viral and host characteristics influence whether a patient can achieve a sustained virological response (SVR), defined as having undetectable serum HCV RNA at 24 weeks after cessation of treatment. The IDEAL trial was instrumental in delineating the baseline predictors of response to therapy1, 2 (Fig. 1). Because of the potential significant adverse effects of PEG-IFN and RBV, clinicians weigh the various viral and host characteristics for each patient before initiating treatment.

Figure 1.

Relative predictivity of patient and viral characteristics for sustained virological response. In multivariate logistic regression analysis from the IDEAL study.

Viral Predictors.

Baseline viral characteristics that influence response to treatment include genotype and viral load. Viral genotype is the most important baseline viral characteristic that predicts response. Patients infected with genotype 1 virus who are treated for 48 weeks with PEG-IFN and RBV have a 40%-50% likelihood of having an SVR, whereas patients with genotype 2 or 3 virus have an SVR rate of 70%-80% after only 24 weeks of PEG-IFN and RBV therapy.2-6 Compared to patients with a high viral load (>600,000 IU/mL), those with low viral load (≤600,000 IU/mL) had 3.1 times the likelihood of achieving an SVR.1

Host Predictors.

In addition to the viral genotype and viral load, there are several host characteristics that predict response to treatment, including race, gender, age, weight, serum glucose, hepatic steatosis, liver fibrosis stage, and adherence to therapy.3, 6-9 Independent of sociodemographic characteristics or compliance to treatment, African American patients with chronic HCV have an almost 50% reduction in SVR rates with PEG-IFN and RBV compared with non-Hispanic patients of European ancestry.8, 10 Younger age, female gender, lower weight, normal serum glucose, absence of hepatic steatosis, lower liver fibrosis stage, and adherence to therapy are all positive predictors of an SVR.1, 2

On-Treatment Predictors

Changes in serum levels of HCV RNA during the first 4 to 12 weeks of therapy with PEG-IFN and RBV can predict the likelihood of achieving SVR.3, 11-13 Rapid virological response (RVR), defined as undetectable HCV RNA at week 4 of treatment, is highly predictive of subsequent SVR.11, 12 For patients who achieve an RVR, SVR rates are as high as 80%-90%.11-15 RVR is a more robust predictor of SVR than IL28B status but because RVR is most frequently seen in CC genotype patients, we would consider that these two predictors are so closely intertwined that they drive each other to promote SVR. Conversely, the absence of an early virological response (EVR), defined as at least a 2 log10 drop in viral load from baseline at week 12 of treatment, is the best negative predictor for nonresponse to treatment. Complete early virological response (cEVR) is defined as undetectable HCV RNA at week 12 of treatment. Partial early virological response (pEVR) is defined as detectable HCV RNA at week 12 of treatment, but >2 log10 drop from baseline. Patients with genotype 1 chronic HCV infections who have a cEVR have a 68% chance of SVR. The probability of SVR in a patient with genotype 1 HCV infection who has pEVR is 29%.13 Again, the interplay between viral response and IL28B genotype is evidenced by the fact that in these non-RVR patients, IL28B genotype is an even more important predictor of SVR.1

IL28B as a Predictor of SVR

In addition to the baseline and on-treatment predictors discussed above, data have shown IL28B polymorphism to be a strong pretreatment predictor of SVR in treatment-naive HCV genotype 1 patients treated with PEG-IFN and RBV, across different racial and ethnic populations (Caucasian, African American, Hispanic, and Asian).1, 16, 17 When compared to non-CC IL28B genotypes, the CC genotype is associated with double the SVR rate, significantly higher rates of viral clearance at weeks 2, 4, 12, and 48 of treatment and lower virological relapse rate.1 In making the decision of whether or not to treat the a patient with genotype 1 chronic HCV infection, IL28B genotype should be weighed along with the other baseline viral and host predictors of SVR.

IL28B and Viral Kinetics

The CC IL28B genotype was found to be associated with improved SVR rates by enhancing early viral kinetics, increasing the rates of week 4, week 12, and week 48 viral clearance, and decreasing the rate of posttreatment relapse.1 Patients with the CC IL28B genotype were found to have a higher rate of RVR. However, for patients who did not achieve RVR, the CC IL28B genotype also strongly predicted SVR. In contrast, for those subjects attaining RVR, IL28B type was not associated with SVR (CC versus non-CC genotype). In multivariate regression modeling to predict SVR, the predictive odds ratio for RVR versus (non-RVR + non-CC genotype) was 9.1 (P < 0.001). In patients who did not achieve RVR, the odds ratio for CC genotype versus non-CC genotype) was 5.2 (P < 0.001) (Table 1). These findings suggest that IL28B testing during therapy for patients who do not achieve RVR may be helpful to make decisions about whether or not to continue therapy (Fig. 2).

Table 1. On-Treatment Predictors of Sustained Virologic Response
PredictorsOdds Ratio95% Confidence IntervalP-value
  1. Adapted from Thompson et al., Gastroenterology. 139:120-129.

RVR vs. (non-RVR + non CC genotype)9.15.8-14.0<0.0001
(non RVR + CC) vs. (non-RVR + non-CC)5.23.9-6.9<0.001
Metavir F0-2 vs. F3-42.71.7-4.1<0.001
HCV-RNA ≤600,000 vs. >600,000 IU/mL2.41.7-3.4<0.001
Caucasian vs. AA ethnicity2.31.6-3.3<0.001
Hispanic vs. AA ethnicity1.81.04-3.10.0361
Fasting blood sugar level <5.6 vs. ≥5.6 mol/L1.71.3-2.30.0 001
Figure 2.

Potential algorithm for managing treatment of patients with peginterferon-alfa and ribavirin ± DAAs incorporating IL28B genotyping. High VL, high viral load (>600,000 IU/mL); Low VL, low viral load (≤600,000 IU/mL); RVR, rapid virological response (undetectable viral load at week 4 of treatment); cEVR, complete early virological response (undetectable viral load at week 12); pEVR, partial early virological response (>2log reduction in viral load at week 12); EVR, early virological response (cEVR + pEVR); DAA, direct-acting antiviral; PEG-IFN, pegylated interferon; RBV, ribavirin.

IL28B and Racial Factors

Other variables that were found to predict SVR in this multivariate regression model included ethnic background, fibrosis stage, baseline viral load, and fasting blood sugar. In the IDEAL trial, the CC IL28B genotype predicted SVR across the racial and ethnic populations of Caucasian, African American, and Hispanic.1 The frequency of the CC genotype was highest in Caucasians (37%), followed by Hispanics (29%) and lowest in the African Americans (14%). However, the IL28B did not account for all of the racial and ethnic differences in response to treatment. The rate of SVR in Caucasians with the CC IL28B genotype (69%) was higher than in either Hispanics (56%) or African Americans (48%). Multivariate regression analysis found both the IL28B CC genotype and ethnic background to independently predict SVR. In a prediction model derived from this multivariate analysis using only baseline variables, IL28B genotype had the greatest odds ratio (CC versus non-CC genotype: odds ratio 5.2; P < 0.0001), compared to the ethnic background (Caucasian versus African American ethnicity: odds ratio 2.8, P < 0.0001; Hispanic versus African American ethnicity: odds ratio 2.1, P = 0.0041) (Table 2). In making decisions on treatment, both IL28B and race/ethnicity need to be considered.

Table 2. Baseline Predictors of Sustained Virologic Response
PredictorsOdds Ratio95% Confidence IntervalP-value
  1. Adapted from Thompson et al., Gastroenterology. 139:120-129.

CC IL28B genotype vs. non-CC5.934.57-7.69<0.0001
HCV-RNA ≤600,000 vs. >600,000 IU/mL3.12.3-4.1<0.0001
Caucasian vs. AA ethnicity2.771.96-3.92<0.0001
Hispanic vs. AA ethnicity2.031.20-3.430.0041
Metavir F0-2 vs. F3-42.71.8-4.0<0.0001
Fasting blood sugar level <5.6 vs. ≥5.6 mol/L1.71.3-2.3<0.0001

IL28B and Fibrosis Stage

Although advanced fibrosis stage has been associated with a decreased rate of SVR,1, 2 IL28B genotype has a higher predictive odds ratio than fibrosis stage. The multivariate regression modeling of predictors of SVR performed in the IDEAL trial demonstrated that the IL28B genotype had a predictive odds ratio of 5.2 (CC IL28B versus non-CC genotype: P < 0.0001) compared to the odds ratio of 2.7 for fibrosis stage (Metavir 0-2 versus F3-4: P < 0.0001) (Table 2).

CC IL28B genotype was the strongest pretreatment predictor of SVR in genotype 1 HCV in the multivariate regression modeling. It is a stronger predictor than all the other significant predictors of baseline viral load, ethnic background, fibrosis stage, and fasting blood sugar. In patients with minimal fibrosis who have other baseline characteristics that predict a poor response (e.g., African-Americans, high baseline viral load, and/or high fasting blood sugar), IL28B genotype is clinically useful in the treatment decision. In patients who have other baseline characteristics that predict a poor response, having a CC IL28B genotype would confer a reasonable chance of SVR and treatment with PEG-IGN and RBV should be considered. However, in the patient with minimal fibrosis, non-CC genotype, and other unfavorable baseline characteristics, it is reasonable to wait for the availability of direct antivirals to be combined with PEG-IFN and RBV.

IL28B in Combination With Other Predictors of SVR

Significant work is being performed on other predictors of SVR to determine whether their predictive value is additive to the predictive value of IL28B genotype. Promising predictors are serum interferon-γ-inducible protein-10 (IP-10) and vitamin D levels, both of which have been found in improve the predictive value of IL28B genotype.

Interferon-γ-Inducible Protein-10 (IP-10).

Elevated levels of interferon-γ-inducible protein-10 (IP-10), a T-cell-specific chemokine, have been found to correlate with nonresponse to treatment of chronic genotype 1 HCV patients with PEG and RBV.18-24 Darling et al.19 assessed the combination of IL28B and pretreatment serum IP-10 as predictors of response to PEG and RBV in the VIRAHEP-C study cohort. This cohort of 272 patients consisted of treatment-naïve African-American (AA) and Caucasian-American (CA) patients with chronic genotype 1 HCV infection. They confirmed the previous findings of significantly lower pretreatment serum IP-10 in responders versus nonresponders (437 ± 31 pg/mL versus 704 ± 44 pg/mL, P < 0.001). This difference in serum IP-10 levels between responders and nonresponders was seen in both AA and CA patients. The authors set 600 pg/mL as a cutoff value based on receiver operating characteristic (ROC) curve analysis (i.e., low IP-10 level <600 pg/mL, high IP-10 level >600 pg/mL). Although pretreatment serum IP10-levels were not correlated with IL28B genotype, both IP-10 levels and IL28B genotype independently predicted SVR. Combining pretreatment serum IP-10 levels with IL28B genotype improved the predictive value of IL28B genotype in predicting SVR, especially in non-CC genotype. The combination of these two predictors increased the area under the ROC curve (AUC) to 0.80 from 0.70 for IL28B genotype alone.

Vitamin D.

Vitamin D plays an important role in innate immunity response and cell differentiation.25 Vitamin D deficiency has been linked to severe fibrosis and lower rates of SVR in patients with genotype 1 chronic HCV.26 In a study conducted by Petta et al.26 in Italy, patients with genotype 1 chronic HCV were found to have lower serum 25-hydroxyvitamin D (25[OH]D) levels compared to controls (25.1 ± 9.9 mg/L versus 43.1 ± 10.2 mg/L). In the multivariate regression analysis, low levels of 25[OH]D were found to predict higher stage of fibrosis and lower chance of SVR. There are also data showing that vitamin D supplementation can improve the rates of SVR.27, 28 In a retrospective study of 211 treatment-naïve chronic HCV Caucasian patients in Italy, Bitetto et al.29 observed that pretreatment serum vitamin D levels improved the predictive power of IL28B genotype in predicting SVR. Ninety-five of these patients were genotype 1 patients. Their definition of vitamin D deficiency was <20 mg/L, which is lower than the definition set by Petta et al. (<30 mg/L). Although they did not find an association between vitamin D levels and fibrosis stage, they confirmed that vitamin D levels independently predicted SVR and improved the predictive power of IL28B genotype (Table 3). This predictive power was observed across all viral genotypes, especially in the difficult-to-treat genotypes.

Table 3. Predictors of SVR in HCV Genotypes 1, 4, and 5
 OR95% CIP-value
  1. Adapted from Bitetto et al., Hepatology 53:1118-1126.

IL28B and vitamin D  <0.001
 -CC genotype and 25[OH]D >20 μg/L1.000- 
 -CC genotype and 25[OH]D ≤20 μg/L0.1050.014-0.792 
 -T allele and 25[OH]D >20 μg/L0.0550.010-0.316 
 -T allele and 25[OH]D ≤20 μg/L0.0280.005-0.162 
HCV RNA >600,000 IU/mL0.2090.069-0.6330.003

Pending prospective validation using the combination of IL28B genotype and pretreatment serum IP-10 and vitamin D levels, there may be a role for both serum IP-10 and vitamin D measurements in future treatment algorithms.

Putting It All Together: How Should the Clinician Use IL28B Genotyping?

There are as yet no guidelines for the use of IL28B and this section is a best practice guideline based on the opinions of the authors with the currently available data. Given the lack of data, there is a need for clinical trials to determine the optimal way to incorporate IL28B into clinical practice. The simplest approach would be to measure IL28b genotype in all genotype 1 HCV patients who are naïve to treatment and the current commercial price is around USD300 for the test. Because this is the best baseline predictor, this test could potentially decrease the need for the more expensive use of liver biopsy as a routine. Liver biopsy, however, still has an important role in patients in whom the stage of disease might affect the treatment decision or if there is a need to rule out other concomitant liver diseases. Liver biopsy could also be combined with a noninvasive serum marker to categorize the patient disease stage.

A second approach would be to reserve the use of IL28B to patients in whom the treatment decision was unclear and who had other poor predictors of response. In this case, the IL28B genotype would help define whether treatment would be with standard PEG-IFN and RBV or whether one should wait for or use a direct acting antiviral (DAA) in combination (Fig. 3). Cost-effectiveness analyses are needed to determine the optimal strategy but both could be used, depending on resources.

Figure 3.

Potential algorithm for utilization of IL28B genotyping in initial treatment decision making.

For treatment failure patients the paradigm for utilization of IL28B is not so clear. The majority of these patients will be non-CC genotypes and no real data exists on the role of IL28B in retreatment with standard therapy in failure patients with PEG-IFN and RBV. Most of these patients are waiting for DAA treatments and there is some data on IL28B in treatment failures with DAA, discussed elsewhere in this issue of the journal. No definite recommendation can be made for IL28B testing in this population for PEG-IFN and RBV therapy alone.

The second important question is how should the clinician use the IL28B genotype information during treatment as viral kinetics start to have a stronger predictive role for SVR. Here we have suggested utilizing the combination of IL28B and disease stage. For mild disease, we propose an algorithm, which uses the easy to treat characteristics of CC genotype, mild disease, and RVR to shorten duration and recommend treatment, while suggesting that those with more unfavorable baseline predictors await/undergo DAA-based treatment. Patients with stage 2 to 4 remain treatment candidates but we suggest utilizing a combination of IL28B genotype and viral load reduction at week 4 for decision making. In particular, patients with a T allele and a less than 1 log reduction in HCV RNA at week 4 are candidates for stopping therapy and considering DAA-based regimens. In addition, patients with only pEVR at week 12 who have an overall SVR of 21% and an SVR of 20% with a T allele might also be considered for treatment cessation and retreatment with a DAA.

These suggested algorithms apply to PEG-IFN and RBV therapy only and will likely change as new DAAs become available and the need for IFN is potentially reduced. What is critical now is some prospective studies to see how IL28B can effect treatment duration with PEG-IFN and RBV, particularly whether CC genotype treatment duration can be reduced even further than 24 weeks in patients with favorable predictors and also which strategy is the most effective. Combination of IL28B with other novel predictors such as IP-10 and vitamin D levels may also further define the approach to non-CC genotype patients.