OBJECTIVES: To review the recent CEAs of approved chimeric antigen receptor -T cell (CAR-T) therapies.

METHODS: A systematic review was conducted in Embase, Medline and Medline in-process, in line with NICE guidelines. Eligible studies included economic evaluations of approved CAR-T therapies published between 2018-2022.

RESULTS: Twenty-one papers were elegible, assessing tisagenlecleucel (n=12), axicabtagene ciloleucel (n=7), and brexucabtagene autoleucel (n=3). Studies related to diffuse large B-cell lymphoma (DLBCL, n=11), acute lymphoblastic leukemia (ALL, n=9), and mantle cell lymphoma (MCL, n=3). Two studies compared two CAR-T therapies, and 19 assessed a CAR-T treatment vs. the standard of care (SoC). Most studies relied on partitional survival models (n=13), and seven were based on semi-Markov models (2 microsimulations, 5 cohort models). Survival extrapolation methods included standard parametric functions (n=8), mixture cure models (n=6), and spline models (n=2). Three studies used different extrapolation methods between the CAR-T intervention and the comparator. For studies comparing tisagenlecleucel vs. SoC (n=10), the incremental life-years (iLYs) and incremental quality-adjusted life years (iQALYs) ranged between 0.807–13.27 and 0.509–10.77, respectively; for axicabtagene ciloleucel vs. SoC (n=3), iLYs were 1.89-6.9 and iQALYs 1.52–6.54; and for brexucabtagene autoleucel vs. SoC (n=3), iLYs and iQALYs ranged between 4.52–9.56 and 3.64–7.03, respectively. Most differences were explained by the time horizon, but the extrapolation method also had a strong impact on the results. Drivers of the CAR-Ts comparison were the acquisition cost and pre-progression utility while other tested factors had a limited impact. For studies comparing CAR-Ts to SoC, the most influential parameters included discount rate (n=10), health states utilities (n=10), cost of CAR-T therapies (n=9) and subsequent treatments (n=4).

CONCLUSIONS: Over the past four years, CEAs have expanded to the comparison of CAR-T therapies and further work has been conducted to better understand the CEA drivers. However, real world evidence and long-term data is needed to confirm these assessments.