An IL28B genotype-based model for personalized prediction of response to pegylatedinterferon-alfa and ribavirin in the treatment of chronic hepatitis C

Abstract

An IL28B Genotype-Based Model for Personalized Prediction of Response to Pegylated-Interferon-Alfa and Ribavirin in the Treatment of Chronic Hepatitis C Thomas R. O'Brien, James E. Everhart, Raymond T. Chung, Timothy R. Morgan, Herbert L. Bonkovsky, Yongwu Shao, Anna S. Lok, Mitchell L. Shiffman, John J. Sninsky, Ruth M. Pfeiffer and the HALT-C Trial Group Purpose: Personalized medicine may improve medical practice by tailoring treatment decisions to individual characteristics. Genetic variation in IL28B and other factors are associated with sustained virological response (SVR) after pegylated-interferon- alfa/ribavirin (peg-IFN/RBV) treatment for chronic hepatitis C (CHC). We developed a model for predicting an individual's pre-treatment probability of SVR based on IL28B genotype and commonly measured clinical variables. Methods: The HALT-C trial enrolled patients with advanced CHC who had failed previous interferon-based treatment. During the lead-in phase of HALT-C, subjects were re-treated with a standard peg- IFN/RBV regimen. SVR was defined as non-detectable HCV RNA 24 weeks after the end of therapy. A step-wise algorithm (criteria: entry, p=0.10; exit, p=0.05) was used to select variables for a logistic regression model that is used to predict an individual's probability of SVR. We compared fit between nested models by the likelihood ratio test and used area under the ROC curve (AUC) to measure the model's ability to discriminate subjects who achieved SVR from non-responders. To compare AUC between models, we computed a p-value based on a X2 test (1 df) that used a bootstrap variance estimate. Results: Among 646 European American patients infected with HCV genotype 1, the median age was 49 years, 75.4% were male, median pre-treatment HCV RNA was 6.53 log10 IU/ml; 37.3% had cirrhosis. The distribution of IL28B rs12979860 genotypes was: CC, 24.0%; CT, 56.8%; TT, 19.2%. The overall SVR rate was 14.2%. IL28B genotype was the strongest statistical predictor of SVR in the model. A model including IL28B genotype, pre-treatment HCV RNA level, AST/ALT ratio, Ishak fibrosis score and prior treatment with ribavirin (yes/no) yielded an AUC of 80.3%. Exclusion of IL28B genotype reduced model fit (p<0.001) and AUC (75.0%; p<0.001). The model provides an estimate for an individual's probability of achieving SVR, which ranged from 0.4% to 86.2% among these HALT-C subjects. For example, for patients with HCV RNA 6.50-6.74 log10 IU/ml, AST/ALT 0.50-0.74, fibrosis score 3 and prior non-response to a regimen that did not include ribavirin, the predicted IL28B genotype-specific probability of SVR is: IL28B-TT, 10.3%; IL28B-CT, 17.3%; IL28B-CC, 46.3%. Conclusions: A model based on IL28B genotype and other parameters may facilitate personalized decisions regarding peg-IFN/RBV treatment of CHC. Independent data are needed to validate the model and potentially improve its predictive ability. This approach can be extended to patients previously naive for peg-IFN/RBV and those receiving regimens that include additional agents.