Benchmarking maintenance therapy survival in first-line advanced urothelial carcinoma using disease modeling.

4575 Background: First-line (1L) maintenance avelumab (maintA) in patients with advanced urothelial carcinoma (aUC) prolonged overall survival (OS) in the JAVELIN Bladder 100 trial. JAVELIN measured OS from the initiation of maintA among a subgroup of the overall aUC 1L treated population, including only patients who did not progress with 1L platinum-based therapy (PBT) and remained progression free during a 4-10 week treatment-free interval following completion of 1L PBT. As such, patients who progressed on or immediately following PBT were not included in the trial. To address this, we used disease modeling to estimate maintA OS measured from initiation of 1L PBT to align maintA OS with the common initiation point to benchmark with other 1L clinical trials. Methods: We developed a simulated cohort to estimate OS from initiation of 1L PBT in all aUC patients including those who received maintA. Eligibility for maintA (PBT progressed vs not progressed) was assessed at 5.6 months post-initiation of 1L PBT, accounting for 6 PBT cycles and a nominal 6-week treatment-free interval. Median OS of 14.3 months from the 1L PBT arm of a recent 1L trial, KEYNOTE-361, was used to represent the OS of the simulated cohort of 1L aUC patients. The simulated cohort was stratified by eligibility for maintA. Among the 1L treated population, 57% were projected to be progression-free and eligible for maintA based on progression-free survival (PFS) from KEYNOTE-361. Among those eligible, 85% were assumed to receive maintA based on expert clinical input. OS of maintA eligible patients was 21.4 months from JAVELIN. Our disease model estimated OS among a maintA ineligible simulated cohort, which was combined with the maintA eligible cohort to yield an estimated OS in the overall aUC population measured from initiation of 1L PBT. Results: Approximately 50% of the 1L modeled population receive maintA (i.e., 57% eligible x 85% treated). Our approach estimated a median OS of 9.9 months for the maintA ineligible cohort and 27.0 months for the maintA eligible cohort. Combined, the estimated median OS with maintA, measured from initiation of 1L PBT, in all 1L-treated aUC patients was 15.8 months (vs 14.3 months with PBT only in 1L). Conclusions: MaintA improves OS for eligible patients, however our simulation model suggests a significant proportion of patients will not receive MaintA. The estimated OS of 15.8 months for the population-level impact of maintA, accounting for patients ineligible for 1L maintA, demonstrates remaining unmet need in the overall 1L aUC population.