Exploratory Cost-Effectiveness Analysis of Response-Guided Neoadjuvant Chemotherapy for Hormone Positive Breast Cancer Patients

Abstract

Purpose: Guiding response to neoadjuvant chemotherapy (guided-NACT) allows for an adaptative treatment approach likely to improve breast cancer survival. In this study, our primary aim is to explore the expected cost-effectiveness of guided-NACT using as a case study the first randomized controlled trial that demonstrated effectiveness (GeparTrio trial).

Materials and methods: As effectiveness was shown in hormone-receptor positive (HR+) early breast cancers (EBC), our decision model compared the health-economic outcomes of treating a cohort of such women with guided-NACT to conventional-NACT using clinical input data from the GeparTrio trial. The expected cost-effectiveness and the uncertainty around this estimate were estimated via probabilistic cost-effectiveness analysis (CEA), from a Dutch societal perspective over a 5-year time-horizon.

Results: Our exploratory CEA predicted that guided-NACT as proposed by the GeparTrio, costs additional €110, but results in 0.014 QALYs gained per patient. This scenario of guided-NACT was considered cost-effective at any willingness to pay per additional QALY. At the prevailing Dutch willingness to pay threshold (€80.000/QALY) cost-effectiveness was expected with 78% certainty.

Conclusion: This exploratory CEA indicated that guided-NACT (as proposed by the GeparTrio trial) is likely cost-effective in treating HR+ EBC women. While prospective validation of the GeparTrio findings is advisable from a clinical perspective, early CEAs can be used to prioritize further research from a broader health economic perspective, by identifying which parameters contribute most to current decision uncertainty. Furthermore, their use can be extended to explore the expected cost-effectiveness of alternative guided-NACT scenarios that combine the use of promising imaging techniques together with personalized treatments.

Fig 1. Decision tree and Markov model.

Decision nodes (black squares) are points at which the patient or health provider makes a choice. Chance nodes (black circles) are points at which more than one event is possible but is not decided by neither the patient or health provider. During the 1^st model cycle, patients receive the intervention; response-guided neoadjuvant chemotherapy (NACT), starting with 2xTAC followed by 4xNX (unfavorable at monitoring) or by 6xTAC (favorable at monitoring), or conventional-NACT, with equal treatment of 6xTAC to all patients, followed by surgery. In the following 4-year cycles, the Markov model simulates the clinical evolution of the patients. TAC docetaxel, doxorubicin, and cyclophosphamide, NX vinorelbine and capecitabine

Fig 2. Cost-effectiveness acceptability curves.

They show the probability of response-guided neoadjuvant chemotherapy (NACT) and conventional-NACT of being cost-effective at different levels of willingness-to-pay threshold (WTP). At WTP thresholds below €80.000/QALY, response-guided NACT had a higher probability of being cost-effective, ranging from 60% at €10.000/QALY to 78% at the Dutch WTP threshold for severe diseases of €80.000/QALY

Fig 3. One-way sensitivity analysis to all model parameters.

We explored how varying model parameter values could affect the net monetary benefit (NMB). If this became negative, it means that response guided neoadjuvant chemotherapy became cost-ineffective. The NMB remained cost-effective at all parameters values tested, except at specificity of 55% and sensitivity of 100%, were the NMB became negative. Furthermore, an increase in the proportion of relapses and deaths in the conventional-NACT strategy, an increase in the costs of the R health-state and a decrease in the costs of NX markedly increased cost-effectiveness.