Reply to “Design–analysis mismatch in the analysis of longitudinal HER2 data in breast cancer”