OBJECTIVES: Cost-effectiveness evaluations in oncology based on data available at the time of initial approval can be challenging. Available clinical data are often limited to interim immature survival data; inclusion criteria may limit the external validity of results; evidence limited to one head-to-head comparison limits network meta-analysis (NMA) feasibility. Re-evaluation using real world evidence can solve these issues. However, several shortfalls need to be considered when estimating comparative effectiveness. This study explores potential solutions in the context of real-world evidence on metastatic Castration Resistant Prostate Cancer (mCRPC) in patients pre-treated by docetaxel.

METHODS: A three-state survival model was constructed to estimate the cost-effectiveness of post-docetaxel mCRPC treatments in France and populated with data from the Janssen European Prostate Cancer Registry (NCT02236637). Baseline characteristics, progression-free survival and overall survival were obtained for abiraterone acetate plus prednisone (AAP) (n=199), enzalutamide (n=98) and cabazitaxel (n=145). Survival was extrapolated based on the NICE guidelines. Alternative methodologies were tested to take into consideration the differences in patients’characteristics between each treatment arm, including a cox model, adjusted HR, evidence from published NMA based on arm equivalence assumptions, adjusted HR based on propensity score matching and restricting analysis to a subgroup of patients presenting the same characteristics.

RESULTS: Variation in the results was observed with the different methodologies tested. However, AAP and enzalutamide were always the two optimal treatments; cabazitaxel was dominated (the most expensive and less efficient treatment) on the efficiency frontier. In some cases, AAP was more expensive and efficient than enzalutamide with Incremental Cost-Effectiveness Ratios (ICERs) between 2,400 and 46,000 €/QALY. In other cases, enzalutamide was either dominated or more expensive and efficient than AAP with ICERs between 60,000 and 130,000 €/QALY.

CONCLUSIONS: Despite the associated shortfalls, using registry data to estimate comparative effectiveness in CE evaluation is feasible and yielded coherent results despite the different methodologies used.