OBJECTIVES: The ANRS CO20-CUPIC study was designed to evaluate triple therapy (TT) efficacy and safety in HCV-cirrhosis treatment-experienced patients in the French Early Access Programme. 60-week interim analysis has been reported confirming better clinical outcomes than double therapy in a real clinical setting. However, independently of the protease inhibitor (PI) used, boceprevir (BOC) or telaprevir (TLV), treatment failure (TF) was reported in up to 60% of the patients. Little is known about these patients' treatment cost: that of successfully treated patients can be generally calculated from the full course of PI treatment, but the cost of those who failed highly depends on TF timing. Our objective is to estimate the average PI cost/patient who failed treatment based on CUPIC study reported data. METHODS: Using reported data on ITT virological response and TF, BOC and TLV on-treatment rates over time were estimated. Based on this curve, the average PI treatment duration and average PI cost/patient who failed treatment were calculated. When not enough information about time to treatment discontinuation was available, the same conservative approach was applied for both drugs, considering the midpoint of the treatment interval. Sensitivity analyses on the time to TF were performed to confirm the robustness of the results. RESULTS: 472 patients (72%) were included in the 60-week interim analysis. Independently of the IP used, about 60% the patients who start treatment did not achieve viral cure and their estimated average PI treatment duration was 26 wks in BOC patients and 11 wks in TLV patients. The average PI cost/patient who failed treatment with TLV (23.012 €) was 26% higher than that of treatment with BOC (18.253 €). The sensitivity analysis confirmed the robustness of the base case estimation. CONCLUSIONS: In a scenario of comparable efficacy between both PIs, the resources wasted on TF acquire a great importance in selecting the least costly of the two alternatives. Based on CUPIC study reported data, the average PI cost/patient who failed treatment was significantly higher in TLV patients than in BOC patients.