OBJECTIVES: To conduct a systematic literature review (SLR) of economic models for R/R FL submitted to HTA agencies and to summarize issues raised by these groups when reviewing FL models.

METHODS: The SLR followed Cochrane methods to identify economic models in HTA submissions assessing the treatment of adults (aged ≥18 years) diagnosed with R/R FL. Searches were conducted for English-language publications from 2017 to September 2022 using HTA websites from Canada, United Kingdom, Germany, and Australia.

RESULTS: The search identified 10 economic evaluations on second-line or later treatment in patients with FL. All 10 HTA submissions applied either the Markov model or partitioned survival model to simulate cost utility of treatments, including rituximab (NICE), idelalisib (NICE, CADTH), obinutuzumab + bendamustine (SMC and CADTH [1 submission each]; NICE and PBAC [2 submissions each]), and lenalidomide + rituximab (NICE, SMC). Key comments from HTA assessment groups commonly focused on comparator of choice (n=8 submissions); model structure, inputs, and time horizon (n=6 submissions); scenarios selected (n=6 submissions); and target population (n=4 submissions). Concerns were raised about treatment comparators in the third-line or later setting, with a lack of treatment options for R/R disease and a lack of consensus in treatment guidelines. In 4 submissions, clinical inputs did not fully align with the modeled population, or extrapolation methods were inappropriate. The lack of head-to-head comparisons necessitated the estimation of clinical inputs using indirect methods, and data were often immature (n=4).

CONCLUSIONS: Comparator selection and clinical inputs were commonly criticized by HTA groups. New emerging treatment options may offer benefits to patients with FL in later lines of therapy, but economic evidence is still limited. The lack of consensus on treatment algorithms leads to challenges for future HTA submissions, including the selection of appropriate comparators and the resolution of uncertainty surrounding clinical inputs and utility/disutility values.