OBJECTIVES : Pembrolizumab has been approved as a treatment for non-small cell lung cancer (NSCLC) across countries. Economic assessment provides crucial evidence for payers’ reimbursement decision-making. To date, multiple cost-effectiveness analyses (CEAs) of pembrolizumab for the treatment of NSCLC have been published. This study aims to review these publications and provide in-depth assessment of their methodologies.

METHODS : Fourteen published CEAs evaluating pembrolizumab or its chemotherapy combination as a treatment for NSCLC were identified through searches of the PubMed database. Their modeling approaches, survival and cost estimation, and utility analyses were compared and evaluated.

RESULTS : The fourteen studies covered all regulatory-approved pembrolizumab NSCLC indications, including Keynote-010 (one study), Keynote-024 (six), Keynote-042 (four), Keynote-189 (two), and Keynote-407 (one). All studies included progression free, progression, and death as the designated health states and counted in drug and adverse event costs. Differences were observed in modeling approaches (partitioned survival vs. Markov vs. Bayesian Markov), survival extrapolation/transition probability estimation, inclusion of additional costs (e.g., PD-L1 testing, subsequent treatment, disease management, etc.), treatment duration measures (time on treatment vs. treatment to progression), utility sources (trial data vs. literature), and utility analyses (time-to-death vs. health states). Certain aspects of variability across models were problematic, including deviation from observed utilization of treatments within trials without adjustments to efficacy, and long-term mortality risks for pembrolizumab that were higher than the historical real-world chemotherapy mortality. Consequently, results differed even among studies examining the same population and comparator within similar time intervals. For example, Keynote-024 ICERs varied from $49,000/QALY to $97,621/QALY from a US payer perspective.

CONCLUSIONS : Inappropriate CEA methodologies may bias results and alter study conclusions. Payers and policy makers should carefully examine study designs and assumptions when using CEAs for evidence-based decision-making.