Background: A discrepancy often exists between the post-neoadjuvant chemotherapy (NAC) breast tumor size on magnetic resonance imaging (MRI) and pathologic tumor size. We seek to quantify this MRI/pathology discrepancy and determine if the accuracy of MRI post NAC varies with tumor subtype.
Methods: The University of Pittsburgh Medical Center (UPMC) Cancer Registry and radiology database were searched for patients with breast cancer who underwent NAC and MRI staging between 2004 and 2009. We compared radiologic to pathologic staging and stratified differences based on tumor biology using univariate, multivariate, and receiver operating characteristic (ROC) analysis.
Results: Two hundred three of 592 patients undergoing surgery after NAC for breast cancer had MRI staging pre and post chemotherapy. All patients had intact tumors prior to the initiation of chemotherapy. Average tumor size by MRI was 4.0 cm pre chemotherapy and 1.2 cm post chemotherapy. The average pathologic tumor size was 1.7 cm (range 0-13 cm). The difference between MRI and pathologic tumor size was greatest in luminal (1.1 cm) and least in triple-negative (TN) and human epidermal growth factor receptor 2 (HER2)-positive tumors (<0.1 cm) (p = 0.015). MRI was a good discriminator for pathologic complete response (pCR) [area under the curve (AUC) 0.777]. Its predictive value for pCR was much greater in TN and estrogen receptor(ER)-/HER2+ than in luminal tumors (73.6 vs. 27.3%).
Conclusions: MRI is an effective tool for predicting response to NAC. The accuracy of MRI in estimating postchemotherapy tumor size varies with tumor subtype. It is highest in ER-/HER2+ and TN and lowest in luminal tumors. Knowledge of how tumor subtype affects MRI accuracy can guide recommendations for surgery following NAC.