Prediction of treatment week eight response & sustained virologic response in patients treated with boceprevir plus peginterferon alfa and ribavirin

Abstract

Aim: Sustained virologic response (SVR) can be attained with boceprevir plus peginterferon alfa and ribavirin (PR) in up to 68% of patients, and short duration therapy is possible if plasma HCV RNA levels are undetectable at treatment week 8 (TW8 response). We have developed predictive models for SVR, and TW8 response using data from boceprevir clinical trials. Methods: Regression models were built to predict TW8 response and SVR. Separate models were built for TW8 and SVR using baseline variables only, and compared to models with baseline variables plus HCV RNA change after 4 weeks of PR (TW4 delta). Predictive accuracy was assessed by c-statistics, calibration curves, and decision curve analyses. Nomograms were developed to create clinical decision support tools. Models were externally validated using independent data. Results: The models that included TW4 delta produced the best discrimination ability. The predictive factors for TW8 response (n = 856) were TW4 delta, race, platelet count and ALT. The predictive factors for SVR (n = 522) were TW4 delta, HCV-subtype, gender, BMI, RBV dose and platelet count. The discrimination abilities of these models were excellent (Cstatistics = 0.88, 0.80 respectively). Baseline models for TW8 response (n = 444) and SVR (n = 197) had weaker discrimination ability (C-statistic = 0.76, 0.69). External validation confirmed the predictive accuracy of the week 4 models. Conclusions: Models incorporating baseline and treatment week 4 data provide excellent prediction of TW8 response and SVR, and support the clinical utility of the lead-in phase of PR. The nomograms are suitable for point-of-care use to inform individual patient and physician decision-making. © 2014 Thompson et al.