Real-world cost effectiveness of first-line pembrolizumab for advanced melanoma: A population-based study by the Canadian Real-world Evidence Value for Cancer Drugs (CanREValue) Collaboration.

Abstract

17 Background: Randomized controlled trials (RCTs) demonstrate large survival benefits with anti-PD-1 checkpoint inhibitors compared to anti-CTLA4 therapy for advanced melanoma. However, it remains unclear if patients in routine practice derive a similar survival benefit or if real-world health utilization differs from trials. Outcomes are needed to inform life-cycle health technology reassessment (HTA) with real-world cost-effectiveness analysis. Methods: This study compared patients with advanced melanoma treated with publicly funded first-line ipilimumab (September 2012 - December 2014) or pembrolizumab (June 2016 - March 2018) in Ontario, Canada. These periods were chosen to reflect distinct eras of access to treatment. Linked administrative databases were used to identify cases, covariates, health-utilization and all-cause death. Inverse probability of treatment weighting (IPTW) with stabilizing weights was used to adjust for covariates (including: age, sex, melanoma site, rurality, income, comorbidity, stage at diagnosis, cancer history, prior brain metastasis treatment). Using a three-year time horizon, individual patient-level censoring-adjusted costs in 2019 Canadian dollars with a 1.5% annual discount rate were determined from the public payer’s perspective. The outcome was quality-adjusted life-years (QALY) measured at the individual patient level. Health utilities were based on accepted Canadian values from the initial HTA. The incremental cost-effectiveness ratio (ICER) was determined with bootstrap confidence intervals (CI). Results: Ninety patients treated with first-line ipilimumab, and 300 with pembrolizumab were identified. Those receiving pembrolizumab were older (median 70 vs.63 years), more likely to have multiple comorbidities (12% vs. 8%), and no prior brain radiation (83% vs. 74%). Covariates were balanced after weighting. Pembrolizumab was associated with improved OS (43.1% vs. 22.1% with ipilimumab at 3 years, IPTW adjusted hazard ratio: 0.52, 95% CI: 0.39-0.70; p < 0.001). Mean costs for pembrolizumab and ipilimumab were $212,706 (95% CI 196,122 - 229,290) and $158,352 ($143,218 - 173,484), and mean survival 1.20 QALY (95%CI 1.11 – 1.29) and 0.66 QALY (0.52 – 0.80) respectively. The ICER was $101,183/QALY (65,375 – 139,298). The probability of cost-effectiveness was 0%, 48% and 99% at willingness-to-pay thresholds of $50k, $100k and $150k respectively. Conclusions: In real-world patients, first-line pembrolizumab for advanced melanoma was associated with improved OS compared to ipilimumab. The real-world cost-effectiveness estimate of $101,183/QALY is similar to the model-based cost-effectiveness estimates ($114,389/QALY to $151,369/QALY) used in the initial health technology assessment recommendation prior to reimbursement.