OBJECTIVES

CAR-Ts are entering the market in populations with high unmet need and limited treatment options. CAR-Ts have also been assessed in single-arm trials. Thus, identifying and including relevant comparators is problematic. We investigated included comparators, methods for comparative efficacy and extrapolation and HTA feedback.

METHODS

We performed a targeted review of grey literature sources for HTA submissions from 2016–2019 for CAR-T therapies and extracted information on the comparator used by the company, the approach and data used and HTA feedback.

RESULTS

We identified nine HTA submissions (one ICER, three CADTH, three NICE and two SMC). Indications were for large B-cell lymphoma (LBCL)(n=7) and acute lymphoblastic leukaemia (ALL)(n=2) in pre-treated relapsed/refractory patients.

All submissions presented salvage chemotherapy as a comparator, informed from real-world data. For LBCL, RWE sources included SCHOLAR-1(n=5), a multicohort, retrospective study which pools two phase III trials and two observational cohorts; the HMRN database(n=1), a population-based UK cohort of haematological malignancy patients; and an open-label extension to the CORAL trial combined with HMRN data(n=1). For ALL, RWE sources included a single-arm trial of clofarabine(n=1) and a retrospective cohort of post-hoc analysis of three trials(n=1). Comparative efficacy was via naïve indirect comparison(n=4) or with adjustments to patient-level data based on ECOG score or proportion receiving transplant(n=5). Approaches for extrapolating OS for salvage chemotherapy included using a spline(n=2), standard parametric functions(n=3), combined standard parametric and spline functions for lines of therapy(n=1) and weighted standard parametric and splines(n=2). No data were available for PFS/EFS, so a relationship between OS and PFS/EFS was assumed.

CONCLUSIONS

Determining appropriate methods/data for comparative efficacy was a major focus for HTAs and was often cited as a reason for resubmission, with HTAs preferring alternative sources to those used by companies. There is need for further work to identify approaches that are acceptable to HTAs.