OBJECTIVES: Treatment options for patients with advanced NSCLC (aNSCLC) are rapidly expanding. The iTEN model was developed to support medical decision-making by predicting aNSCLC patient survival and associated costs, from a Canadian healthcare system perspective.

METHODS: A discrete event simulation was developed to estimate survival and costs, dependent upon patient characteristics (histology, biomarker, performance and smoker status). Treatment sequencing and eligibility were derived from a modified Delphi process with Canadian clinical experts. Published Kaplan–Meier progression free and overall survival data were fit with parametric functions to estimate treatment efficacy and applied based on statistical and visual fit. Progression and death events were randomly assigned for each patient, using the estimated parametric functions. Treatment history was assumed to have no impact on the efficacy of subsequent therapies. Costs included were: drug acquisition and administration, monitoring, imaging, physician visits, end-of-life, best supportive care and adverse events. Analyses were based on 100,000 simulated patients. Model survival predictions were validated against published real-world estimates from the Ontario Cancer and Austrian (TYROL) registries and a retrospective analysis of US medical records (Nadler et al., RESULTS: The Ontario Cancer registry, TYROL registry and US medical records study reported overall survival for patients receiving two-lines of chemotherapy, up to five-lines of treatment with chemotherapy/targeted therapies or up to three-lines of treatment with PD-1 inhibitors/chemotherapy/targeted therapies, respectively. One-year survival rates for these studies were 67%, 38% and 49%. iTEN estimated one-year survival, after restricting the treatment pattern to match each study, was: 61%, 39% and 60%. Based on the Canadian (2017) treatment algorithm, the estimated one-year survival and per-patient cost of treating aNSCLC were 46% and $89,899.

CONCLUSIONS: Survival estimated by the iTEN model approximates published real-world data. Further validation is planned; however, the model may reasonably estimate the clinical and cost consequences of treating aNSCLC.