Sir,

We congratulate the authors on performing this multicentre trial with 164 patients (Heil et al, 2015). One important selection criterion is the initial size of the primary tumor (T size) on imaging – this information did not appear in the manuscript and is of interest. The probability of adequately sampling the initial region of disease by image guidance is likely related to the T size at diagnosis and the size of residual disease on imaging. Do the authors have information on the initial T stage of all patients and the size of residual imaging abnormality at the time of pre-surgical biopsy? This is not the first report regarding the concept of minimally invasive techniques for sampling to assess pathologic complete response in breast cancer (Clouth et al, 2007). The success of moving these studies into the realm of clinical omission of surgery is dependent on exquisite radiologic targeting of the lesion (37% did not have a clip placed at initial biopsy in the current study), adequate sampling with vacuum-assisted biopsy (utilized in the current study in ∼30%), and the number of biopsies performed (unknown 55% of cases). At MD Anderson, we have an actively accruing trial using image-guided vacuum-assisted biopsy with fine-needle aspiration and more stringent selection criteria including patients with a 50–70% chance of pathologic eradication of disease after neoadjuvant chemo- and targeted therapy (triple-negative and HER2-amplified breast cancers) (National Institute of Health, 2015).