OBJECTIVES:

This study describes the results of a cost-effectiveness analysis of nivolumab plus ipilimumab (NIVO+IPI) versus platinum doublet chemotherapy (Chemo) in previously untreated metastatic non small-cell lung cancer (NSCLC) using ‘mixture cure’ survival analysis, a statistical method to address circumstances where some patients experience long-term survival and negligible excess mortality.

METHODS:

A semi-Markov model was developed in Microsoft Excel from a US third-party payer perspective based on CheckMate 227 data (3 year database lock [DBL]); model selection was informed by 5 year DBL. The model consisted of four health states: progression-free, post-progression, progression-free with long-term response (LTR), and dead. A mixture cure survival modelling approach was used to explore long-term survival trends, with LTR patients exposed to general population mortality hazard. US-specific costs (2021 dollars) were derived from publicly available sources, utilities were informed by CheckMate 227 patient reported outcomes. Model outcomes included the incremental cost per life-year (LY) and quality-adjusted life-year (QALY) gained for NIVO+IPI versus Chemo based on a lifetime time horizon; uncertainty was evaluated in probabilistic sensitivity analysis (PSA).

RESULTS:

NIVO+IPI was associated with a mean gain of 1.69 LYs and 1.43 QALYs versus Chemo using a mixture-cure model of overall survival. 3.97 LYs were accrued on NIVO+IPI, of which 2.01 were in the LTR state. The incremental cost of NIVO+IPI versus Chemo was $134,792, with an incremental cost per QALY of $93,982; the 95% confidence interval in PSA was $129,705 - $157,079 for incremental costs and 0.86 – 2.04 for incremental QALYS.

CONCLUSIONS:

The results indicate that compared with Chemo, NIVO+IPI is a cost-effective first-line treatment at established norms for cost-effectiveness. The cost-effectiveness of NIVO+IPI was associated with a significant LTR fraction and gain in QALYs using a mixture-cure modelling framework. PSA supported the robustness of these conclusions. Further research based on future DBL is indicated to confirm these findings.