The KEYNOTE-905/EV-303 study evaluated perioperative combination therapy with enfortumab vedotin (EV) plus pembrolizumab (Pem) in patients with muscle-invasive bladder cancer (MIBC) who were predominantly cisplatin-ineligible or who declined cisplatin. Compared with surgery alone with or without adjuvant nivolumab (referred to here as surgery ± adjuvant nivolumab), EV+Pem improved event-free survival (EFS), overall survival (OS), and pathological complete response in the trial population. The present analysis assessed the cost-effectiveness of perioperative EV+Pem versus surgery ± adjuvant nivolumab from a US third-party payer perspective using a cost-utility framework.
A state-transition Markov model with three health states was used: event-free (EF), progressed or recurrent disease (PRD), and death. The model estimated transitions from EF using parametric multistate survival models fitted to trial data and extrapolated beyond observed follow-up. Survival following PRD was modeled via a treatment-mix approach intended to reflect US clinical practice patterns. The analysis adopted a US third-party payer perspective for costs and assessed outcomes in quality-adjusted life-years (QALYs).
The target population comprised adults with MIBC who were cisplatin-ineligible or who declined cisplatin, reflecting the KEYNOTE-905 trial population. Transition probabilities and utilities were derived from KEYNOTE-905 trial data where available. Costs were obtained from US sources. The publication reports that EQ-5D-5L data from KEYNOTE-905 were used to estimate health-state utilities.
Transitions from the event-free state were estimated using parametric multistate survival models based on trial outcomes, with extrapolation beyond the trial follow-up period. For patients who entered PRD, survival was modeled using a treatment-mix approach that aimed to mirror US clinical practice after progression. Costs included initial treatment and subsequent care; the manuscript reports higher initial costs for EV+Pem but lower costs associated with PRD treatment relative to the comparator. Utilities for health states were taken from the KEYNOTE-905 EQ-5D-5L responses collected in the trial. Scenario analyses explored alternative extrapolation methods and structural modeling choices, while deterministic and probabilistic sensitivity analyses assessed parameter uncertainty.
In the base-case analysis, perioperative EV+Pem increased mean life-years by 3.08 compared with surgery ± adjuvant nivolumab (7.70 vs 4.62 life-years). The gain in QALYs was 2.56 (6.19 vs 3.62 QALYs). Total costs were higher with EV+Pem: $546,831 versus $413,171 for the comparator, an incremental cost increase of $133,661. The higher initial treatment costs for EV+Pem were partially offset by lower subsequent PRD treatment costs ($34,636 for EV+Pem vs $129,813 for the comparator). These inputs produced a base-case incremental cost-effectiveness ratio (ICER) of $52,117 per QALY gained.
Across explored scenarios that varied extrapolation approaches and structural assumptions, reported ICERs ranged from $50,363 to $69,703 per QALY gained. A probabilistic sensitivity analysis indicated that EV+Pem had a 99.1% probability of being cost-effective at a willingness-to-pay threshold of $100,000 per QALY, reflecting robustness to parameter uncertainty within the modeled assumptions.
The authors highlight key limitations, notably uncertainty related to long-term survival extrapolation beyond the trial follow-up and potential changes in post-progression treatment patterns over time. These factors may affect projected downstream costs, outcomes after PRD, and therefore the ICER. The report indicates that results should be interpreted in light of these uncertainties.
From a US third-party payer perspective and using the assumptions and inputs reported, perioperative enfortumab vedotin plus pembrolizumab is cost-effective compared with surgery ± adjuvant nivolumab for cisplatin-ineligible or cisplatin-declining patients with MIBC at a willingness-to-pay threshold of $100,000 per QALY. The analysis found meaningful gains in life-years and QALYs associated with EV+Pem, higher total costs driven by initial therapy, and lower costs for treatment of progressed or recurrent disease. The authors caution that long-term extrapolation and evolving treatment patterns after progression represent important sources of uncertainty.