Prospective trial of breast MRI versus 2D and 3D ultrasound for evaluation of response to neoadjuvant chemotherapy.

ABSTRACT

BACKGROUND: Preoperative imaging to assess response to neoadjuvant chemotherapy in breast cancer is routine but no single imaging modality is standard of practice. Our hypothesis is that ultrasound (US) is comparable to magnetic resonance imaging (MRI) in the prediction of residual disease. METHODS: A single-institution, Institutional Review Board-approved prospective trial of primary invasive ductal breast cancer patients receiving neoadjuvant chemotherapy enrolled women from 2008 to 2012. Two-dimensional (2D) and three-dimensional (3D) US, as well as MRI images of pre- and post-neoadjuvant tumors were obtained. Skin involvement or inadequate images were excluded. Residual tumor on imaging was compared with surgical pathology. Differences of tumor volume on imaging and pathology were compared using the non-parametric Wilcoxon signed-rank test. US to MRI agreement was determined by the kappa coefficient. Tumor volumes in estrogen receptor (ER), progesterone receptor (PR), and Her2neu subgroups were compared using the Kruskal-Wallis test. ER/PR staining <5 % was considered negative; Her2neu status was determined by in situ hybridization. RESULTS: Forty-two patients were enrolled in the study; 39 had evaluable post-treatment data. Four patients were Her2neu positive, and 17 (46 %) patients had triple-negative tumors. Among 11 (28 %) patients with pathologic complete response (pCR), US correctly predicted pCR in six (54.5 %) patients compared with eight (72.7 %) patients when MRI was used. This is a substantial agreement between US and MRI in predicting pCR (kappa = 0.62). There was no difference between 2D and 3D US modalities. For the 39 patients, US and MRI had no significant difference in volume estimation of pathology, even stratified by receptor status. CONCLUSION: The estimation of residual breast tumor volume by US and MRI achieves similar results, including prediction of pCR.