Based on the results of the KHBO1401-MITSUBA trial, GCS therapy has become the standard primary chemotherapy for advanced BTC. Our study is the first to examine the cost-effectiveness of GCS therapy for advanced BTC, and the results from this study validated the cost-effectiveness of GCS therapy. According to the one-way sensitivity analysis, changes in utility and cost did not increase the ICER value beyond the WTP threshold. All parameters affecting the OS and PFS curves influenced the ICER. Two of these parameters led to an increase in the ICER beyond the WTP threshold (variation in the OS scale parameter for GCS and variation in the OS mean log parameter for GC). In contrast, other parameters caused variations in the ICER within the WTP threshold. Probabilistic sensitivity analysis, which included analysis of these curve variations, revealed a robust and cost-effective trend for GCS therapy. Therefore, we believe that GCS therapy is efficacious for BTC patients and ensures efficient use of health insurance resources.
Despite its rarity, advanced BTC is characterised by a high rate of recurrence and metastasis. The information on the cost-effectiveness of the available first-line treatment options for advanced BTC is currently limited. In China, chemotherapy using capecitabine plus oxaliplatin (XELOX) is more cost-effective than chemotherapy using gemcitabine and oxaliplatin [27]. As the GCS and XELOX regimens were established independently, no clinical trials have been conducted to compare the efficiency of these regimens, further underscoring the need to investigate these differences. The cost-effectiveness of GC therapy was investigated in two earlier studies. Roth et al. reported the cost-effectiveness of gemcitabine and cisplatin combination therapy (ICER of 59,480 USD/QALY) in comparison to cisplatin monotherapy [28]. Tsukiyama et al. reported that the ICER of gemcitabine and cisplatin combination therapy in Japan was 14 million JPY/QALY compared with that using gemcitabine monotherapy [29]. Our study investigated the cost-effectiveness of GCS therapy. Collectively, the GCS regimen was a cost-effective first-line chemotherapeutic option for advanced BTC in the Japanese healthcare system.
Our study has several limitations. First, the utility estimates were based on data from a clinical trial that studied the efficacy of gemcitabine/pazopanib therapy in Greek patients and were not based on data from Japanese BTC patients, which may have influenced the results. One-way sensitivity analysis revealed that the ICER did increase beyond the WTP threshold, indicating the minimal influence of the ICER. Future advances in QOL research can be used to upgrade the model developed in this study. Second, several assumptions were made to minimise the impact of parameter uncertainty in the model, including the assumption that S-1 is the treatment of choice for a significant percentage of patients post-disease progression, which may not be accurate for all patients. Currently, the recommended treatment options for advanced BTC after first-line chemotherapy are not standardised. In addition to S1 monotherapy, regimens combining gemcitabine, CDDP, and S-1 may be recommended for post-disease progression. Although the impact of the drug choice post-disease progression was considered in this study, one-way sensitivity analysis results indicated that the impact of the cost of follow-up treatment post-disease progression was insignificant. Third, palliative care costs were not included in this study because they would be equivalent in both groups. The impact of palliative care costs is likely to be insignificant, similar to the impact of the cost of follow-up treatment post-disease progression described earlier. Nevertheless, the model can be updated to include palliative care costs based on the availability of palliative care estimates for advanced BTC. Fourth, although the long-term OS of patients with advanced BTC was modelled based on parametric distribution, its uncertainty is an unavoidable limitation of this study. Although the uncertainty of curve extrapolation was considered in the probabilistic sensitivity analysis, it did not influence the results, indicating that the model was robust. Updates to the present model based on the inclusion of long-term patient survival data would provide more accurate estimates.