OBJECTIVES : Exemestane (EXE), exemestane + everolimus (EXE+EVE), toremifene (TOR), and fulvestrant (FUL) are second-line endocrine therapies for postmenopausal hormone receptor–positive/ human epidermal growth factor receptor 2–negative (HR+/HER2-) metastatic breast cancer (mBC) in Japan. Although recent studies demonstrated the efficacy and safety of these therapies, their cost-effectiveness has not yet been determined in Japan. This study aimed to examine the cost-effectiveness of EXE+EVE, TOR, FUL 250mg, and FUL 500mg compared with EXE.

METHODS : A Markov model was developed to analyze cost-effectiveness over a 15-year time horizon within a monthly-cycle model from a healthcare payer’s perspective. The efficacy, and utility were determined via a systematic search of literature. Only direct health care costs were used. A 2% discount rate applied for costs and outcomes. Subgroup analysis was performed for non-visceral metastasis. A series of one-way sensitivity, probabilistic sensitivity analyses (PSA), and threshold analyses were performed.

RESULTS : Base-case analyses estimated incremental cost-effectiveness ratios (ICERs) for EXE+EVE, TOR, and FUL 500mg of 18.7, 3.7, and 7.3 million yen/QALY, respectively. FUL 250mg was dominated. Cost-effectiveness acceptability curve showed that TOR is the most cost-effective when willingness-to-pay (WTP) is above 3.5 million yen/QALY. The subgroup analysis revealed a lower ICER compared to the base case results. The threshold analysis indicated that EVE and FUL prices should be reduced 75% and 40%, respectively. The overall survival (OS) highly influenced the ICER.

CONCLUSIONS : As a second-line therapy for postmenopausal women with HR+/HER2- mBC, only TOR is cost-effective and the most favorable choice at a WTP of 5 million yen/QALY that is frequently used in Japan. To improve a cost-effectiveness of EXE+EVE and FUL 500mg, a large price reduction would be necessary, while the incremental cost-effectiveness ratio is lower for a non-visceral metastatic subgroup. Policy makers should carefully interpret the results by taking into account some parameter uncertainties in the model.