STRUCTURAL UNCERTAINTY OF MARKOV MODELS FOR ADVANCED BREAST CANCER

Sunday, October 19, 2014
Poster Board # PS1-54

Quang A. Le, PharmD, PhD, Western University of Health Sciences, Pomona, CA
Purpose: To examine the impact of structural uncertainty of Markov models in modeling cost-effectiveness for treatment of advanced breast cancer (ABC). 

Methods: Four common Markov models for ABC were examined. Markov models 1 and 2 have four health states (stable, responding-to-therapy, disease-progressing, and dead), and Markov models 3 and 4 only have three health states (stable, disease-progressing, and dead). In models 1 and 3, the possibility of death can occur in any health states; while in models 2 and 4, the chance of dying can only occur in disease-progressing health state. Relationships of transition probabilities among the Markov models were derived based on the key conditions: (1) the average overall survival times are the same in all models, and (2) no significant difference in clinical outcome in terms of disease progression between stable and non-progressive patients and those who respond to therapy for Markov models 3 and 4. A simulation was conducted to examine the impact of using these Markov models on the cost-effectiveness results in a context of the combination therapy of lapatinib and capecitabine for treatment of HER2-positive ABC. The overall cost-effectiveness result of the combination therapy was also estimated by parameterizing both parameter and structural uncertainties directly in the model.

Results: Our simulation results yielded a range of incremental cost-effectiveness ratios (ICER) between $300,000 per additional quality-adjusted life year (QALY) and $350,000/QALY with the 95% confidence intervals ranged from -$35,220/QALY to $892,609/QALY among four Markov models. When parameterizing both parameter and structural uncertainties directly in the model, the overall ICER produced $318,327/QALY for the combination therapy with lapatinib and capecitabine as opposed to the monotherapy with capecitabine (95% CI: $10,592/QALY - $733,069/QALY).

Conclusions: Our study suggests that modeling ABC with different Markov model structures may produce different cost-effectiveness results. When applying in the context of treatment of HER2-positive ABC, the combination therapy with lapatinib is not cost-effective regardless which Markov model was used and whether uncertainties were accounted for.