OBJECTIVES: The rapidly evolving treatment landscape for locally advanced or metastatic urothelial carcinoma (la/mUC), and challenges associated with trial design have limited direct comparison of outcomes with different therapies/sequences. Modeling can estimate the impact of therapeutic sequences ahead of real-world data (RWD) availability. Modeling was used to estimate median overall survival (mOS) among patients with la/mUC considered cisplatin-ineligible, treated with four treatment options/sequences.

METHODS: mOS was estimated for a simulated cohort of treatment naïve patients with la/mUC from initiation of four treatment options/sequences: 1) enfortumab vedotin (EV) + pembrolizumab; 2) gemcitabine+carboplatin (G/C)±maintenance avelumab followed by EV; 3) pembrolizumab followed by EV; 4) G/C followed by pembrolizumab, then EV. For each option/sequence, mOS was the composite curve based on best-fit OS curves for each treatment informed by published trials (EV-103 Dose Escalation/Cohort A, JAVELIN Bladder 100, KEYNOTE-045/-052/-361) using the previously published oncology simulation model framework. Progression from each treatment was based on PFS curves of published trials and assumed all patients received first-line treatment. Attrition during second- (38.3% treatment rate) and third-line (33.2%) was based on published RWD. Sensitivity analyses varied second- and third-line treatment rates.

RESULTS: The model estimated a mOS of 26.5 [95% credible interval [CI]:18.7-37.9] months for EV+pembrolizumab; 14.4 [95% CI:12.4-16.4] months for G/C±maintenance avelumab then EV; 11.9 [95% CI:10.3-13.3] months for pembrolizumab then EV; and 12.7 [95% CI:10.9-14.7] months for G/C then pembrolizumab then EV. Sensitivity analyses varying treatment rates across second- and third-line treatments (20%-60%) showed small variations in mOS for treatment options/sequences: 14.1-14.8 months for G/C±maintenance avelumab then EV; 11.2-12.7 months for pembrolizumab then EV; and 12.6-12.8 months for G/C then pembrolizumab then EV.

CONCLUSIONS: Model results, while limited by data inputs, estimated longer mOS with first-line EV+pembrolizumab than alternative options/sequences, indicating the clinical value of selecting effective therapies upfront and the importance for future studies of this combination.