OBJECTIVES: To assess the impact of method for modelling DR (transient or absorbing state) for MIUC patients with high risk of recurrence and tumor cell PD-L1 expression ≥1% on the CE of NIVO versus surveillance in France.
METHODS: A 4-state (disease-free, local recurrence, DR, death) semi-Markov model was developed to conduct an analysis from the French healthcare system perspective, over a 25-year time horizon with weekly cycles. Costs (in 2021 €s) and health outcomes were discounted by 2.5% annually. Patient characteristics, efficacy (for disease-free and local recurrence), safety and utilities (EQ-5D-3L, health-state specific calculated using the French value set) were derived from CheckMate 274 trial. First recurrence events were repartitioned using the trial data within the first year and the largest French prospective cohort with MIUC between years 2 and 5. Disease-free patients at year 5 were considered cured. Post-DR outcomes were assessed by either treating the DR state as a transient state prior to death via tunnel states or as an absorbing state with one-off costs and QALYs. Post-DR survival was informed by parametric survival models fitted to published data in first-line metastatic urothelial carcinoma. Outcomes of interest were life years (LYs), quality-adjusted LYs (QALYs), and incremental cost-utility ratio (ICUR). Probabilistic sensitivity analyses were run for both methods to generate CE acceptability curves.
RESULTS: NIVO was associated with higher total LYs, QALYs and costs (9.3, 5.9, and €126,510, respectively) vs surveillance (6.4, 4.0, and €75,845, respectively) with an estimated ICUR of €26,691/QALY gained with DR as tunnel state. Modelling DR as an absorbing state resulted in an ICUR of €26,083/QALY gained. At a willingness to pay threshold of €50,000/QALY gained, NIVO had 95% probability of being cost-effective for both DR modelling.
CONCLUSIONS: NIVO is estimated to be a life-extending and cost-effective adjuvant treatment in France with minimal sensitivity due to DR modelling approach.
