Modelling the cost-effectiveness of adjuvant lapatinib for early-stage breast cancer

Background. It has been shown in the NeoALTTO trial that a neoadjuvant regimen containing paclitaxel, lapatinib and trastuzumab is superior to regimens which include only one of the HER2 antagonists with paclitaxel. In light of these results, we modelled the potential cost-effectiveness of adjuvant lapatinib for patients with HER2-positive early-stage breast cancer. Material and methods. We constructed a Markov state-transition model with three different health states: disease free, relapse, and death. We assumed an 18-week course of lapatinib was added to the TCH arm of the BCIRG 006 trial. Since no efficacy data are available for combining adjuvant lapatinib with trastuzumab, we ran the model assuming five different hypothetical hazard ratios for disease free survival when lapatinib is added to TCH (TCH was used as the control group). The hazard ratios were 0.9, 0.8, 0.7, 0.6, and 0.5. Outcomes are given in quality-adjusted life-years (QALYs). Both costs and QALYs were discounted at the 4% rate. We calculated the cost per QALY from the perspective of the Irish health care system. Probabilistic sensitivity analysis and one-way sensitivity were performed and confidence intervals were bootstrapped. Results. The incremental cost-effectiveness ratios (ICERs) for the five hazard ratios are €53 089/QALY, €27 893/QALY, €18 463/QALY, €13 527/QALY and €10 490/QALY, respectively. Using the €45 000/QALY threshold, an adjuvant lapatinib regimen is cost-effective at the 0.8 hazard ratio. Adjuvant lapatinib becomes cost-effective at the 0.879 hazard ratio where the ICER is €44 825/QALY. Conclusion. In the Irish setting, an adjuvant lapatinib regimen would be considered cost-effective for patients with HER2-positive early-stage breast cancer for four of the five hypothesised hazard ratios. Data from both adjuvant and neoadjuvant trials suggest that the hazard ratio required to achieve cost-effectiveness for adjuvant lapatinib is both possible and plausible.