Comparative efficacy and safety of first-line systemic combinations for advanced urothelial carcinoma: A Bayesian network meta-analysis.

Abstract

184 Background: The phase 3 EV-302 trial established enfortumab vedotin plus pembrolizumab (EV+P) as the new first-line (1L) standard of care over platinum-based chemotherapy (PBCT) for advanced urothelial carcinoma (aUC). However, comparative efficacy and toxicity between EV+P and the broader landscape of modern frontline regimens including chemo-immunotherapy (IO), dual IO, and maintenance strategies remain unestablished. We conducted a Bayesian network meta-analysis (NMA) to resolve these comparative efficacy and safety profiles. Methods: A systematic literature search identified phase 3 randomized controlled trials evaluating 1L systemic therapies for aUC. Interventions included EV+P, IO+PBCT (nivolumab, atezolizumab, pembrolizumab), dual IO (durvalumab+tremelimumab), maintenance IO (avelumab), and targeted combinations (pembrolizumab+lenvatinib). Endpoints included OS, PFS, and Grade 3 treatment-related adverse events (TRAEs). A Bayesian NMA with a random-effects model was conducted utilizing JAGS. Efficacy was evaluated via Hazard Ratios (HR), safety via Odds Ratios (OR), and treatment rankings via the Surface Under the Cumulative Ranking curve (SUCRA). Results: Seven pivotal trials comprising >5,500 patients were included. For OS, EV+P demonstrated significant superiority over PBCT (HR 0.47, 95% Credible Interval [CrI] 0.38-0.58) and ranked most efficacious (SUCRA 99%). Sequential PBCT followed by avelumab maintenance ranked second for OS (HR 0.69; SUCRA 78%), followed by nivolumab+PBCT (HR 0.78; SUCRA 62%). For PFS, EV+P similarly achieved the highest ranking (HR 0.45, 95% CrI 0.38-0.54; SUCRA 98%), followed by avelumab maintenance (HR 0.62). Safety profiles exhibited stark heterogeneity. Analysis of Grade 3 TRAEs revealed that dual IO (durvalumab+tremelimumab) ranked as the safest regimen (27.0% Grade 3 TRAEs; OR vs PBCT < 1). EV+P demonstrated a favorable safety profile compared to chemotherapy alone (55.9% vs 69.5%; OR < 1). Conversely, all concurrent IO+PBCT triplets resulted in increased toxicity compared to PBCT alone (OR > 1), with Grade 3 TRAEs ranging from 61.8% to 81.0%. Conclusions: EV+P provides unprecedented OS and PFS benefits, firmly ranking as the most efficacious 1L strategy for aUC while demonstrating a manageable Grade 3 toxicity profile relative to traditional chemotherapy. Notably, concurrent chemo-immunotherapy triplets offer inferior survival benefits and higher toxicity burdens. The distinct safety profiles highlight a critical need for biomarker-driven patient selection and optimized post-progression sequencing.