CLINICAL ADVANCES IN LIVER, PANCREAS, AND BILIARY TRACT Interleukin-28B Polymorphism Improves Viral Kinetics and Is the Strongest Pretreatment Predictor of Sustained Virologic Response in Genotype 1 Hepatitis C Virus ALEXANDER J. THOMPSON,* ANDREW J. MUIR,*,��� MARK S. SULKOWSKI,�� DONGLIANG GE,�������� JACQUES FELLAY,�������� KEVIN V. SHIANNA,�������� THOMAS URBAN,�������� NEZAM H. AFDHAL, IRA M. JACOBSON,�� RAFAEL ESTEBAN,# FRED POORDAD,** ERIC J. LAWITZ,������ JONATHAN MCCONE,���� MITCHELL L. SHIFFMAN, GREG W. GALLER,���� WILLIAM M. LEE,## ROBERT REINDOLLAR,*** JOHN W. KING,��������� PAUL Y. KWO,������ REEM H. GHALIB, BRADLEY FREILICH,������ LISA M. NYBERG,### STEFAN ZEUZEM,**** THIERRY POYNARD,������������ DAVID M. VOCK,* KAREN S. PIEPER,* KEYUR PATEL,*,��� HANS L. TILLMANN,*,��� STEPHANIE NOVIELLO, KENNETH KOURY, LISA D. PEDICONE, CLIFFORD A. BRASS, JANICE K. ALBRECHT, DAVID B. GOLDSTEIN,�������� and JOHN G. MCHUTCHISON*,��� *Duke Clinical Research Institute and ���Duke University Medical Center, Durham, North Carolina ��Johns Hopkins University School of Medicine, Baltimore, Maryland Beth Israel Deaconess Medical Centre, Boston, Massachusetts ��Weill Cornell Medical College, New York, New York #Hospital General Universitario Valle de Hebron, Barcelona, Spain **Cedars-Sinai Medical Center, Los Angeles, California ������Alamo Medical Research, San Antonio, Texas ����Mt. Vernon Endoscopy Center, Alexandria, Virginia Liver Institute of Virginia, Newport News, Virginia ����Kelsey Research Foundation, Houston, Texas ##University of Texas Southwestern Medical Center, Dallas, Texas ***Piedmont Healthcare, Statesville, North Carolina ���������Louisiana State University, Shreveport, Louisiana ������Indiana University School of Medicine, Indianapolis, Indiana The Liver Institute at Methodist Dallas Medical Center, Dallas, Texas ������Kansas City Gastroenterology and Hepatology, Kansas City, Missouri ###Kaiser Permanente, San Diego, California ****J.W. Goethe-University Hospital, Frankfurt, Germany ������������Groupe Hospitalier Pitie-Salpetriere, Paris, France ��������Center for Human Genome Variation, Duke University, Durham, North Carolina and Schering-Plough Research Institute, Kenilworth, NJ BACKGROUND & AIMS: We recently identified a poly- morphism upstream of interleukin (IL)-28B to be associ- ated with a 2-fold difference in sustained virologic re- sponse (SVR) rates to pegylated interferon-alfa and ribavirin therapy in a large cohort of treatment-naive, adherent patients with chronic hepatitis C virus genotype 1 (HCV-1) infection. We sought to confirm the polymor- phism���s clinical relevance by intention-to-treat analysis evaluating on-treatment virologic response and SVR. METHODS: HCV-1 patients were genotyped as CC, CT, or TT at the polymorphic site, rs12979860. Viral kinetics and rates of rapid virologic response (RVR, week 4), complete early virologic response (week 12), and SVR were compared by IL-28B type in 3 self-reported ethnic groups: Caucasians (n 1171), African Americans (n 300), and Hispanics (n 116). RESULTS: In Caucasians, the CC IL-28B type was associated with improved early viral kinetics and greater likelihood of RVR (28% vs 5% and 5% P .0001), complete early virologic response (87% vs 38% and 28% P .0001), and SVR (69% vs 33% and 27% P .0001) compared with CT and TT. A similar association occurred within African Americans and His- panics. In a multivariable regression model, CC IL-28B type was the strongest pretreatment predictor of SVR (odds ratio, 5.2 95% confidence interval, 4.1���6.7). RVR was a strong predictor of SVR regardless of IL-28B type. In non-RVR patients, the CC IL-28B type was associated with a higher rate of SVR (Caucasians, 66% vs 31% and 24% P .0001). CONCLUSIONS: In treatment-naive HCV-1 patients treated with pegylated interferon and ribavirin, a polymorphism upstream of IL-28B is as- sociated with increased on-treatment and sustained virologic response and effectively predicts treatment outcome. Keywords: Genetics IL-28B Interferon-Lambda Peg- Interferon-Alfa. View this article���s video abstract at www.gastrojournal. org. Oare ne hundred and eighty million individuals worldwide chronically infected with hepatitis C virus (HCV)1 and at risk for related morbidity and mortality from cirrho- sis and hepatocellular carcinoma. Curative antiviral therapy may prevent these complications. The current standard of Abbreviations used in this paper: ALT, alanine aminotransferase BMI, body mass index cEVR, complete early virologic response CI, confidence interval EVR, early virologic response HCV, hepatitis C virus HCV-1, hepatitis C virus genotype 1 IL, interleukin ITT, intention- to-treat pegIFN, pegylated-interferon RBV, ribavirin RVR, rapid viro- logic response SNP, single nucleotide polymorphism SVR, sustained virologic response. �� 2010 by the AGA Institute 0016-5085/$36.00 doi:10.1053/j.gastro.2010.04.013 CLINICAL ADVANCES IN LIVER, PANCREAS, AND BILIARY TRACT GASTROENTEROLOGY 2010 139:120���129

care is pegylated-interferon-alfa (pegIFN-alfa) and ribavirin (RBV) combination therapy. However, of patients infected with genotype 1 HCV (HCV-1), the most common HCV genotype in North America, Europe, and Japan, only ap- proximately 40% are cured by standard therapy.2���6 Further- more, therapy may be associated with considerable toxicity. Therefore, the ability to prospectively identify individual patients who are likely to respond to treatment would be clinically valuable. A number of pretreatment host and viral factors have been associated with treatment outcome in HCV-1.6 These include baseline viral load, age, sex, body mass index (BMI), insulin resistance, hepatic steatosis, and hepatic fibrosis. African American ancestry is a powerful negative predictive factor for sustained virologic response (SVR).7,8 The rate of plasma HCV-RNA decline during treatment is predictive of treatment outcome, and virologic responses at week 4 (rapid virologic response [RVR]) and week 12 (early virologic response [EVR]) are additional key therapeutic milestones. However, our understanding of the genetic determinants of treatment outcome has been limited. We recently performed a genome-wide association study to identify genetic determinants of treatment re- sponse in HCV-1 patients treated with pegIFN plus RBV.9 We identified a single nucleotide polymorphism (SNP) upstream of the gene IL-28B on chromosome 19, coding for IFN- -3, which was associated with an approximately 2-fold difference in SVR rates in patients of European, African American, or Hispanic ancestry.9 The analysis was restricted to 1137 of 1671 patients, in which nonre- sponders were required to have been more than 80% adherent to both pegIFN and RBV dosing, and ethnicity was defined by genetic ancestry.9 The importance of this genetic region as a determinant of treatment response has now been confirmed by 2 independent genome-wide association studies.10,11 Interleukin (IL)-28B polymor- phism also has been shown to be associated with spon- taneous clearance after HCV infection.12,13 In this intention-to-treat (ITT) analysis of the discovery cohort, we sought to interpret the IL-28B polymorphism in a more detailed clinical context to determine how knowl- edge of this genetic information might impact physician practice. We describe how the genotype of the IL-28B poly- morphism influences on-treatment virologic responses, as well as relapse rates, and consider in detail the effect of the polymorphism in the context of other variables predictive of antiviral therapy outcome. Our analyses included all pa- tients, regardless of their level of adherence to therapy, and ethnicity was determined by subject self-report, as it would be in a clinical practice setting. Materials and Methods Patients The study population included 1604 of 3070 pa- tients who were enrolled in the IDEAL study and con- sented to genetic testing (ClinicalTrials.gov number, NCT00081770).6 In addition, 67 patients were included from a second randomized controlled trial.7 For all 1671 patients, the protocol-specified treatment duration was 48 weeks, with an additional 24 weeks of follow-up evaluation. Clinical and laboratory data were collected as described previously.6,7 Ethnicity was defined by patient self-report, and not genetically inferred ancestry as in the analysis of Ge et al.9 A discrepancy between self-report and genetic ances- try was noted in 130 (8%) patients. All patients for whom the polymorphism of interest was genotyped successfully were included in this analysis, which therefore included 491 patients excluded from the analysis by Ge et al9 (336 [21%] on the basis of nonadherence). Genotyping A total of 1671 patients were genotyped using the Illumina Human610-quad BeadChip (Illumina, San Di- ego, CA) as previously described.9 We selected the discov- ery SNP, rs12979860, for this study. Genotype at the polymorphic site rs12979860 on chromosome 19 was suitable for analysis in 1628 patients. For simplicity, we refer to an IL-28B polymorphism throughout this article, noting that the association SNP actually lies 3 kilobases upstream of the IL-28B gene. Genotype was defined as CC, CT, or TT IL-28B type. Treatment Efficacy Assessments HCV-RNA levels were measured using sensitive reverse-transcription polymerase chain reaction assays. In the IDEAL study, the Cobas TaqMan assay (Roche Mo- lecular Diagnostics, Pleasanton, CA) was used, which has a lower limit of quantitation of 27 IU/mL.6 In the earlier study by Muir et al,7 the NGI SuperQuant assay was used (National Genetics Institute, Culver City, CA), which has a lower limit of quantitation of 39 IU/mL. Viral load was measured at baseline treatment weeks 2, 4, 12, 24, and 48 and follow-up evaluation weeks 4, 12, and 24 (pa- tients from the study by Muir et al7 did not have viral load measured at week 2 or week 4). On-treatment re- sponses were defined by undetectable plasma HCV-RNA levels at the following time points: ultrarapid virologic response at 2 weeks RVR at 4 weeks complete EVR (cEVR) at 12 weeks and end-of-treatment response at 48 weeks.14 SVR was defined by undetectable HCV-RNA levels at 24 weeks posttreatment (or 12 weeks posttreat- ment if 24-week follow-up data were not available n 40). Relapse was defined as detectable HCV-RNA levels during follow-up evaluation in patients who achieved end-of-treatment response. Statistical Analysis Comparisons between groups were performed us- ing a Wilcoxon test for the non-normal continuous vari- ables, and for categoric data the Pearson chi-square test/ Fisher exact test was used. Significance was defined at a P CLINICAL ADVANCES IN LIVER, PANCREAS, AND BILIARY TRACT July 2010 IL-28B POLYMORPHISM AND HCV TREATMENT RESPONSE 121

value of less than .05. Analysis of on-treatment response by IL-28B polymorphism was performed in 3 separate ethnic populations: Caucasians, African Americans, and Hispanics (on-treatment responses for the 41 patients of ���other��� ethnicity are not described). A linear mixed-ef- fects model that included subject-specific intercept and slope and accounted for the left censoring of the viral load measurements was built to analyze the association of IL-28B SNP genotype and race on the log10 viral load within the first 12 weeks of treatment.15 Multivariable logistic regression with backward elimination was used to identify baseline factors in the entire cohort associated with SVR. Separate models were not constructed for each ethnicity rather, ethnicity was included as a covariate in the model. Additional covariates considered for inclusion in the model included baseline viral load (log10 IU/mL), fasting blood sugar level, liver fibrosis stage, age, BMI, serum alanine aminotransferase (ALT) level, hepatic ste- atosis grade, ribavirin starting dose, sex, pegIFN (dose/ type), IL-28B type, and IL-28B type by ethnicity interac- tion. IL-28B polymorphism was evaluated according to CC versus non-CC IL-28B type for the regression model- ing. A significance level of 0.05 was used for removal from the model. A second model was built to consider the effect of IL-28B polymorphism for predicting SVR after adjusting for RVR, which included all subjects with measured covari- ates and virologic data at week 4 (1422 subjects). In addition to the covariates described earlier, we grouped week 4 re- sponse and IL-28B polymorphism as a 3-level variable: week 4 responders (RVR) week 4 nonresponders, CC genotype and week 4 nonresponders, non-CC genotype there were too few patients without the CC genotype who were also week 4 responders to subset the week 4 responders by genotype. All analyses were performed using R statistical software (R Foundation for Statistical Computing, http:// www.R-project.org) and SAS version 9.1 (SAS Institute, Cary, NC). Results Characteristics of the Study Patients A majority of the patients were male (61%) and older than 40 years of age (Table 1). Most patients were Caucasian (72%) African Americans comprised 18% of patients, and Hispanics comprised 7%. Compared with Caucasians, African Americans were older, more likely to have a BMI of 30 kg/m2 or greater, and an increased baseline fasting glucose level, and less likely to have an abnormal serum ALT level. Allocation of pegIFN type was balanced between and within each ethnic group. African American patients were less likely to have been assigned an RBV dose greater than 13 mg/kg/day. The frequency of the IL-28B SNP genotype differed between ethnic groups (P .0001) (Table 1), as previously de- scribed.9 The CC genotype was observed most frequently in Caucasians (37%), followed by Hispanics (29%) and African Americans (14%). The TT genotype was more common in African Americans (37%) than Hispanics (22%) or Caucasians (12%). Viral Kinetics As previously reported, a small but statistically sig- nificant difference in median viral load at baseline was noted according to IL-28B type, with higher levels present in CC patients (Caucasians, 6.6 (6.1���6.9) vs 6.4 (6.0���6.7) vs 6.3 (5.9���6.6) log10 IU/mL for CC, CT, and TT patients, respec- tively, Supplementary Table 1).9 However, when viral load was considered according to the threshold of 600,000 IU/ mL, the proportion of patients with high baseline viral load did not differ by IL-28B type. On-treatment, differences in viral load reduction be- tween genotypes were detectable as early as week 2, the earliest time point evaluated (Figure 1 Supplementary Table 2). Among Caucasians, median reductions of viral load at week 2 were as follows: 2.6, 0.9, and 0.6 log10 IU/mL for patients with the CC, CT, and TT IL-28B types, respectively (P .0005). Despite ongoing viral decline, the difference was of similar magnitude at weeks 4 and 12, corresponding to increased rates of RVR and cEVR in patients with the CC genotype (Figure 2 and Tables 2 and 3). The rate of viral load reduction in African American and Hispanic patients also was more rapid in those with the CC IL-28B type. However, among African American CC patients, the magnitude of viral decline was less than that observed in Caucasian CC patients at all times (weeks 2, 4, and 12 P .0020 Figure 1, Supplementary Table 2). Linear mixed-effects modeling confirmed that viral load declined more for patients with the CC versus non-CC IL-28B type (delta, 0.6190 95% confidence interval [CI], 0.5562���0.6817 log10 IU/mL/wk Supplementary Table 3). This effect was independent of ethnic background, which also was associated with the rate of viral decline. There was no significant difference in the rate of decline be- tween patients with the CT and TT genotypes (P .1468). Viral Clearance���On-Treatment and SVR Within each ethnic group, the CC IL-28B type was associated with higher on-treatment response rates at all time points (4, 12, and 48 weeks) (Figure 2 and Table 2). In Caucasians who were CC, 87% attained a cEVR, 10% achieved a pEVR, and only 3% did not achieve a 2-log10 IU/mL reduction in viral load at week 12 of treatment. Within all populations, the CC IL-28B type was associated with a greater than 2-fold increase in SVR compared with the TT IL-28B type. The rate of SVR observed in Caucasians with the CC IL-28B type (69%) was higher than in either African Americans (48%) or Hispanics (56%) (P .0079). The CT IL-28B type consistently was associated with numer- ically higher virologic responses than TT however, the dif- ferences were small and not statistically significant (Figure 2 and Table 2). A detailed description of the SVR rates for CLINICAL ADVANCES IN LIVER, PANCREAS, AND BILIARY TRACT 122 THOMPSON ET AL GASTROENTEROLOGY Vol. 139, No. 1

each genotype of the IL-28B polymorphism on the basis of individual and combinations of baseline characteristics and week 4 and week 12 on-treatment responses is presented in Supplementary Table 4. SVR Rates According to Week 4 and Week 12 Responses The CC IL-28B type increased the proportion of patients who attained RVR in those who achieved this key therapeutic milestone, SVR rates were high, indepen- dent of IL-28B SNP genotype (Table 3). In contrast, in patients who did not achieve RVR, the effect of IL-28B SNP genotype was strikingly different���SVR rates were significantly higher in patients with the CC IL-28B type in all populations (Caucasian non-RVR:SVR 66% for CC vs 31% for CT vs 24% for TT P .0001). In patients who were CC at the polymorphic site, the rate of cEVR was high in all populations (Table 2). Rates of SVR were higher post-cEVR than in patients attaining only pEVR, but the predictive utility of the IL-28B polymorphism was not strong once week 12 virologic response was available (Table 3). Test Characteristics for IL-28B SNP Genotype Compared With RVR The performance of the IL-28B SNP genotype (CC vs non-CC) as a binary predictor for SVR was evaluated in the 3 major population groups (Table 4). In Caucasian patients, having the CC IL-28B type was more sensitive and had a higher negative predictive value for SVR than RVR however, RVR had superior positive predictive value and specificity for SVR. Importantly, the CC IL-28B type was present in 37% of the Caucasian population, whereas Table 1. Baseline Characteristics of the Clinical Cohort Baseline characteristics Caucasians African Americans Hispanics Othera P valueb N 1171 300 116 41 Age, y 48 (43���52) 51 (47���54) 45 (39���51) 48 (42���53) .0001 Age, 40 y 997 (85%) 283 (94%) 80 (69%) 33 (80%) .0001 Male sex 713 (61%) 172 (57%) 77 (66%) 24 (59%) .2226 BMI 27.4 (24.7���30.4) 29.4 (26.7���32.6) 28.8 (26.0���32.3) 25.5 (23.4���28.8) .0001 BMI 30 kg/m2 328 (28%) 138 (46%) 44 (38%) 9 (22%) .0001 HCV���RNA level, log10 IU/mL 6.5 (6.0���6.8) 6.3 (5.9���6.7) 6.2 (5.7���6.6) 6.6 (6.2���6.9) .0007 HCV���RNA level, 600,000 IU/mL 979 (84%) 244 (81%) 83 (72%) 35 (85%) .0046 ALT level ULN (range) 1.7 (1.2���2.6) 1.4 (1.0���2.0) 2.0 (1.3���3.5) 1.7 (1.2���2.8) .0001 ALT level ULN 978 (84%) 223 (74%) 103 (88%) 36 (85%) .0002 Fasting glucose level, mmol/L 5.1 (4.8���5.6) 5.2 (4.7���5.9) 5.1 (4.8���5.7) 5.0 (4.6���5.4) .0903 Fasting glucose level, 5.6 mmol/L 336 (29%) 112 (37%) 31 (26%) 10 (24%) .0102 Steatosisc Grade 0 443 (40%) 98 (35%) 29 (25%) 13 (35%) .0006 Grade 1 516 (46%) 155 (55%) 55 (48%) 18 (49%) Grade 2 135 (12%) 26 (9%) 27 (23%) 6 (16%) Grade 3 23 (2%) 3 (1%) 3 (3%) 0 (0%) Grade 4 4 (4%) 0 (0%) 1 (1%) 0 (0%) Steatosis grade 0 678 (60%) 184 (65%) 86 (74%) 24 (65%) .0059 METAVIR fibrosis stagec F0 18 (2%) 2 (1%) 3 (3%) 1 (3%) .2091 F1 795 (71%) 192 (68%) 81 (70%) 30 (81%) F2 175 (16%) 59 (21%) 15 (13%) 2 (5%) F3 60 (5%) 8 (3%) 7 (6%) 2 (5%) F4 73 (7%) 21 (7%) 9 (8%) 2 (5%) METAVIR F3���F4 133 (12%) 29 (10%) 16 (14%) 4 (11%) .5715 PegIFN-alfa 2b 1.0 ug/kg/wk 376 (32%) 88 (29%) 36 (31%) 16 (42%) .7612 2b 1.5 ug/kg/wk 417 (36%) 118 (39%) 45 (38%) 4 (11%) 2a 180 ug/wk 378 (32%) 94 (31%) 35 (30%) 18 (47%) RBV, mg/kg 13.2 (12.4���14.2) 12.8 (12.0���13.7) 13.5 (12.5���14.7) 14.3 (12.6���15.7) .0001 RBV 13 mg/kg 649 (55%) 123 (41%) 70 (61%) 29 (71%) .0001 rs12979860 genotype frequency CC 436 (37%) 42 (14%) 34 (29%) 26 (63%) .0001 CT 596 (51%) 146 (49%) 56 (48%) 13 (32%) TT 139 (12%) 112 (37%) 26 (22%) 2 (5%) NOTE. Data are presented as either median (25th-75th percentile), or n (%). ULN, upper limit of normal. aEthnicities were as follows: Asian American (n 19), American Indian (n 7), and other (n 15). bComparison across Caucasian, African American, and Hispanic patients (continuous data, Kruskal-Wallis Test categoric data, chi-square test). cMissing data: histology 50 cases (Caucasian) 18 cases (African American) 1 case (Hispanic) 4 cases (other). CLINICAL ADVANCES IN LIVER, PANCREAS, AND BILIARY TRACT July 2010 IL-28B POLYMORPHISM AND HCV TREATMENT RESPONSE 123

only 14% attained an RVR. A similar pattern was observed in African American and Hispanic patients. Multivariable Models Regression modeling was used to identify pre- treatment factors that were associated independently with SVR. Data from 1550 patients with a complete dataset of the covariates of interest were included in the model. We first modeled SVR considering all predictors as dichotomous variables (continuous and ordinal vari- ables were dichotomized according to clinically relevant thresholds6). Multivariable logistic regression using back- ward selection identified IL-28B type, ethnic background, baseline viral load, hepatic fibrosis stage, and fasting glucose level as being associated independently with SVR (Table 5). IL-28B type had the greatest odds ratio favoring SVR in this model (CC vs non-CC: odds ratio, 5.2 95% CI, 4.1���6.7 P .0001). A second multivariate logistic regres- sion model was built in which continuous and ordinal variables were not dichotomized, allowing us to use pseudo R-squared values to estimate the contribution of each variable to the variability observed in SVR. IL-28B type (CC vs non-CC) was estimated to explain 14.8% of the variability in treatment response in the cohort, after adjustment for the other independent predictors (Sup- plementary Table 5). Other independent predictors of SVR in this more powerful model included ethnic back- ground, baseline viral load, hepatic fibrosis stage, fasting glucose level, BMI, and RBV starting dose (mg/kg). No other predictor explained more than 5% of the variability in SVR, and the IL-28B type therefore was the strongest pretreatment predictor of SVR. Figure 2. Virologic responses on treatment on the basis of IL-28B type and ethnicity. (A) Caucasian, (B) African American, and (C) Hispanic patients. EOTR, end-of-treatment response. Statistical comparisons are presented in Table 2. Figure 1. Median reductions in viral load from baseline on the basis of IL-28B type. (A) Caucasian, (B) African American, and (C) Hispanic patients. Bars represent 25th and 75th percentiles. P .001 for all pairwise comparisons of median viral load for CC vs CT or TT using the Wilcoxon 2-sample test (see Supplementary Table 1 and 2). CLINICAL ADVANCES IN LIVER, PANCREAS, AND BILIARY TRACT 124 THOMPSON ET AL GASTROENTEROLOGY Vol. 139, No. 1