Are Standard Indirect Treatment Comparison Methods Suitable to Compare First-Line Vs Maintenance Therapies? An Assessment of Enfortumab Vedotin + Pembrolizumab Vs Avelumab in Locally Advanced/Metastatic Urothelial Carcinoma

OBJECTIVES: To assess the suitability of standard methods for an indirect treatment comparison (ITC) of first-line (1L) enfortumab vedotin + pembrolizumab (EV+P) versus gemcitabine + platinum-based chemotherapy (PBC) with avelumab maintenance in patients with locally advanced/metastatic urothelial carcinoma (la/mUC). Avelumab maintenance is recommended for non-progressors following 1L PBC and from a health technology assessment (HTA) perspective may be considered a relevant comparator for EV+P, which has been evaluated in a 1L la/mUC population.

METHODS: A feasibility assessment was conducted using publicly available aggregate-level data (AD) from EV-302 and JAVELIN Bladder 100 (JB-100) to explore between-trial differences in study/patient/treatment characteristics and outcomes. EV-302 evaluated 1L EV+P versus PBC in patients with la/mUC, while JB-100 evaluated avelumab maintenance versus best supportive care in patients with la/mUC with no disease progression on 1L PBC and a 4-10-week treatment-free interval before maintenance. Population-adjusted indirect comparison (PAIC) methods typically accepted in HTAs were explored (i.e., matching-adjusted indirect comparison and simulated treatment comparison).

RESULTS: Assessment of trial characteristics found study design differences that may preclude a standard PAIC, including differences in time of randomization which impacts endpoint assessment and timing of patient characteristic collection. As a PAIC adjusts for between-trial differences in baseline patient characteristics by aligning individual patient data from EV-302 (index trial) to AD from JB-100 (external trial), results would be presented in a population similar to JB-100 (i.e., non-progressors following 1L PBC). The JB-100 population is a subset of the 1L treated population and therefore not reflective of the EV-302 target population for decision makers.

CONCLUSIONS: Differences in trial designs and populations between EV-302 and JB-100 preclude a scientifically robust and meaningful comparison through standard ITC methods. Alternative non-standard ITC methods that allow for time-zero adjustment and alignment with the EV-302 population could be explored.