OBJECTIVES: To assess the cost-effectiveness of pembrolizumab versus carboplatin plus gemcitabine in patients with locally advanced or metastatic urothelial cancer, who are not eligible for cisplatin-containing chemotherapy.

METHODS: A three-state partition survival model was developed with a 20-year time horizon and a weekly cycle. KEYNOTE-052, phase II single arm clinical trial informed the clinical efficacy and adverse events parameters of pembrolizumab in the model. Overall survival (OS) was extrapolated using a piece-wise approach, utilising Kaplan-Meier data until week 32 and using a log-normal distribution for the remainder of the time horizon. A network meta-analysis, enabled through the use of a simulated treatment comparison (STC), provided comparative effectiveness estimates of pembrolizumab versus carboplatin plus gemcitabine as time-varying hazard ratios (HRs). Quality-adjusted life years (QALYs) were estimated using utilities derived from EQ-5D data collected in KEYNOTE-052. Resource use and unit costs were applied from a NHS perspective. Probabilistic-, deterministic-, and scenario- analyses were performed to assess model uncertainty.

RESULTS: The model demonstrated that pembrolizumab increases the life expectancy of patients by 1.39 years, corresponding to 1.04 additional QALYs. The base case incremental cost-effectiveness ratio (ICER) is £35,016 (confidential discount included). Model results are sensitive to the following assumptions: comparative effectiveness HR for pembrolizumab vs. carboplatin plus gemcitabine, extrapolation of overall survival data, and the utility estimates used.

CONCLUSIONS: The results of this cost-effectiveness analysis demonstrated that pembrolizumab is a cost-effective choice compared to carboplatin plus gemcitabine when considering a willingness to pay threshold of £50,000 per QALY, as per NICE end-of-life criteria.