Palbociclib in advanced breast cancer: A cost-utility analysis

Abstract

Background: The addition of palbociclib to letrozole improves progression free survival (PFS) and response rates compared to letrozole alone in the 1st line treatment of hormone receptor positive advanced breast cancer (ABC). This study assesses the costutility of palbociclib from the Canadian healthcare payer perspective. Methods: To evaluate the cost-utility of palbociclib, a probabilistic discrete event simulation model was developed. The model was parameterized with data from the phase 2 and 3 PALOMA 1 and 2 trials and other sources. The incremental cost per qualityadjusted life-month (QALM) gained for palbociclib was calculated. A time horizon of 15 years was used in the base case with costs and effectiveness discounted 5% annually. The time to progression and death were derived from Weibull and exponential distributions, respectively. Expected costs were based on Ontario fees and other sources. Probabilistic sensitivity analyses were conducted to account for parameter uncertainty. Results: Compared to letrozole alone, the addition of palbociclib provided an additional 14.7 QALMat an incremental cost of $161,508. The resulting incremental costeffectiveness ratio was $10,999/QALM gained. Assuming a willingness to pay (WTP) of $4167 per QALM, the addition of palbociclib was not cost-effective and the probability of palbociclib to be cost-effective was 0%. Cost-effectiveness acceptability curves derived from a probabilistic sensitivity analysis showed that at a WTP of $11,667/ QALM gained, the probability of palbociclib to be cost-effective was 50%. Conclusions: Compared with letrozole alone, the addition of palbociclib is unlikely to be cost-effective for the treatment of ABC from a Canadian healthcare perspective with its current price. While ABC patients derive a meaningful clinical benefit from palbociclib, considerations should be given to increase the WTP threshold and reduce the drug pricing, to render this strategy more affordable. Model validation and calibration are needed to confirm those results.