OBJECTIVES:

Novel oncology treatments, including chimeric antigen receptor T-cell therapies, immuno-oncology therapies and targeted treatments, offer potential for long-term survival and even cure. We investigated the use and acceptance of cure assumptions in cost-effectiveness analyses submitted in NICE oncology appraisals.

METHODS:

The NICE website was searched on 20 May 2021 for completed technology appraisals for oncology therapies. Information regarding the modelling approach was extracted from the ten most recent appraisals where the manufacturer’s cost-effectiveness analysis incorporated a cure assumption.

RESULTS:

Ten appraisals incorporating a cure assumption were identified after searching the forty-four most recent oncology appraisals. Four appraisals utilised mixture cure models (MCMs) and six modelled cure (three explicitly, three implicitly) where survival was informed by general population mortality (GPM) after a certain timepoint for a proportion of patients. Committees rejected the modelled cure assumption in the majority (N=7) of appraisals. Concerns with plausibility of cure included limited data follow-up (N=8), lack of an observed plateau (N=4) and reliance on surrogate outcomes (e.g. progression-free survival) (N=4). Generalisable external data with long-term follow-up, precedence from previous appraisals in the same indication and clinical expert opinion were considered more robust justifications for cure versus short-term observed survival profiles. For non-MCM approaches, committees cited methodological concerns with the exogenous choice of the cure fraction and timepoint beyond which patients were considered cured (N=2). Evidence review groups recommended the consideration of MCMs to model cure in the majority (4/6) of appraisals including GPM-based assumptions.

CONCLUSIONS:

Cure assumptions were included in a quarter of recent oncology appraisals but were rarely considered appropriate for decision-making. Absence of long-term trial data was a frequent limitation, but at times generalisable and robust external data or clinical expert opinion supporting the cure assumption have mitigated this. MCMs were considered more robust methods for modelling cure assumptions versus GPM-based models.