OBJECTIVES : Trastuzumab deruxtecan (T-DXd) monotherapy and tucatinib in combination with lapatinib & capecitabine are two regimens recently approved by the FDA for treatment of human epidermal growth factor receptor 2-positive (HER2+) metastatic breast cancer, based on results from their Phase II trials. To indirectly compare the efficacy of T-DXd and tucatinib+lapatinib+capecitabine a matching-adjusted indirect comparison (MAIC) was conducted to adjust for differences in patient populations in the two trials and reduce the bias due to confounding variables.

METHODS : Patient-level data were available for T-DXd (DESTINY-Breast01; N=184), while only published summary data were available for tucatinib+lapatinib+capecitabine (HER2CLIMB- Murthy et al, 2019). Weights were assigned to DESTINY-Breast01 (DB01) patient data such that mean baseline characteristics matched those reported for the HER2CLIMB population; the resulting matched cohort was used in comparisons. Progression-free survival (PFS) was the primary outcome of the comparison (pending mature overall survival data from DB01). MAIC was conducted to calculate adjusted hazard ratios (HRs- 95% confidence intervals) for PFS, for covariates expected to be prognostic to the outcome).

RESULTS : Population was matched on age, hormone receptor expression, Eastern Cooperative Oncology Group (ECOG) status, brain metastases, number of prior therapy lines, and prior pertuzumab. In naïve comparison the unadjusted PFS HR was 0.46 (95% CI; 0.34-0.61) for T-DXd monotherapy versus tucatinib+lapatinib+capecitabine. MAIC including all covariates resulted in a HR of 0.13 (0.02, 0.42) for PFS with a low effective sample size (ESS) of 6.9. During sensitivity analyses single matching covariates brain metastases, prior lines and prior pertuzumab were taken out leading to HRs or 0.22, 0.48 and 0.46 respectively.

CONCLUSIONS : There are substantial differences in the patient populations included in DB01 and HER2CLIMB, resulting in limited ESS when adjusting for covariates of interest. Despite this, results favored T-DXd in terms of PFS for all different sets of covariates matched.