Six-year absolute invasive disease-free survival benefit of adding adjuvant pertuzumab to trastuzumab and chemotherapy for patients with early HER2-positive breast cancer: A Subpopulation Treatment Effect Pattern Plot (STEPP) analysis of the APHINITY (BIG 4-11) trial

Abstract

AIM: The APHINITY trial showed that adding adjuvant pertuzumab (P) to trastuzumab and chemotherapy, compared with adding placebo (Pla), significantly improved invasive disease-free survival (IDFS) for patients with HER2+ early breast cancer both overall and for the node-positive (N+) cohort. We explored whether adding P could benefit some N- subpopulations and whether to consider de-escalation for some N+ subpopulations. METHODS: Subpopulation Treatment Effect Pattern Plot (STEPP) is an exploratory, graphical method that plots estimates of treatment effect for overlapping patient subpopulations defined by a covariate of interest. We used STEPP to estimate Kaplan-Meier differences in 6-year IDFS percentages (P minus Pla: Δ ± standard error [SE]), both overall and by nodal status, for overlapping subpopulations defined by (1) a clinical composite risk score, (2) tumour infiltrating lymphocytes (TILs) percentage, and (3) human epidermal growth factor receptor 2 (HER2) FISH copy number. Because of multiplicity, a Δ of at least three SE is required to warrant attention. RESULTS: The average absolute gains in 6-year IDFS percentages were 2.8 ± 0.9 overall; 4.5 ± 1.2 for N+ and 0.1 ± 1.1 for N-. Largest gains were for patients with intermediate clinical composite risk (5.3 ± 1.9 overall; 6.9 ± 2.3 N+; 4.0 ± 3.0 N-), highest TILs percentage (6.3 ± 1.7 overall; 7.4 ± 2.4 N+; 3.2 ± 1.7 N-), and intermediate HER2 copy number (2.8 ± 1.9 overall; 7.4 ± 2.5 N+; -1.3 ± 1.9 N-), but clear evidence indicating a pattern of differential subpopulation treatment effects was lacking. CONCLUSIONS: STEPP plots for N- did not identify subpopulations clearly benefiting from adding P, and those for N+ did not identify subpopulations warranting de-escalation. TILs percentage appeared to be more predictive of P treatment effect than clinical composite risk score. TRIAL REGISTRATION: clinicaltrials.gov Identifier NCT01358877.