Background: Curative potential of adjuvant nivolumab was compared with radical resection only among patients at high risk of recurrence following radical surgery of muscle-invasive urothelial carcinoma (MIUC).
Methods: Using patient-level disease-free survival (DFS) data from CheckMate 274 (n = 709, minimum follow-up, 31.6 months), we applied mixture cure models (MCMs) to the adjuvant nivolumab (NIVO) and placebo (PBO) arms of the intention-to-treat (ITT) population and tumor PD-L1 expression ≥1% subpopulation. DFS was derived for hypothetical "cured" and "uncured" subgroups. DFS for the cured subgroup was estimated using WHO background mortality rates matched to trial demographic characteristics. Uncured DFS was modeled using parametric distributions and characterized with cure fractions by maximum-likelihood methods. Model selection considered clinical plausibility, visual comparisons of model fit, and goodness-of-fit statistics.
Results: MCM analysis demonstrated that almost all uncured patients experience recurrence or death within 5 years. Clinically plausible models estimated higher cure fractions in tumor PD-L1 ≥1% subgroup for patients treated with NIVO (PD-L1 ≥1%: 59.1%-61.0% vs ITT: 43.1%-45.1%), and highly similar cure fractions for patients receiving PBO irrespective of their PD-L1 expression (PD-L1 ≥1%: 35.9%-36.4% vs ITT: 36.4%-37.0%). Projected 10-year mean DFS was 4.38 to 4.47 years for NIVO and 3.61 to 3.64 years for PBO in the ITT population, and 5.54 to 5.65 years for NIVO and 3.54 to 3.57 years for PBO in the PD-L1 ≥1% subpopulation.
Conclusions: Adjuvant NIVO for high-risk MIUC was associated with a higher cure fraction than PBO in the ITT and PD-L1 ≥1% populations. Results align with reported survival from the trial and highlight clinical outcomes of interest. CheckMate 274 ClinicalTrials.gov identifier, NCT02632409.
Keywords: Adjuvant therapy; Bladder cancer; Cure fraction; Immune checkpoint inhibitor; PD-L1 biomarker.
Copyright © 2025 Elsevier Inc. All rights reserved.