Accuracy of Breast Magnetic Resonance Imaging in Predicting Pathologic Response in Patients Treated With Neoadjuvant Chemotherapy

Abstract

Background Prior studies of the ability of magnetic resonance imaging (MRI) to predict pathologic response to neoadjuvant chemotherapy have shown conflicting results that vary depending on baseline molecular characteristics. This study examines the ability of MRI to predict pathologic complete response (pCR) and explores the influence of tumor molecular profiles on MRI sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). Methods Eighty-one patients with invasive breast cancer treated with neoadjuvantsystemic therapy between 2002 and 2009 who were imaged with breast MRI pre- and post-treatment were reviewed. Patient, tumor, and treatment characteristics were recorded. Comparisons of molecular subsets and their influence on MRI sensitivity, specificity, PPV, and NPV were made using χ 2contingency tables. Results The sensitivity, specificity, PPV, and NPV of MRI for predicting pCR for the total group were 92%, 50%, 74%, and 80%, respectively. Patients had the following molecular subtypes: 33/81 (41%) HR+Her2−, 23/81 (28%) HR+/-Her2 +, and 25/81(31%) triple receptor negative (TN). Molecular subtype did not demonstrate a significant correlation of radiographic and pathologic response, although MRI NPV was highest in the TN subset (100%) followed by those with HR+/-Her2+ disease (87.5%). Multivariate analysis did not show that tumor characteristics (estrogen receptor status, progesterone receptor status, HER2 status) or neoadjuvant treatment (doxorubicin, cyclophosphamide, paclitaxel versus other or trastuzumab) had any effect on MRI sensitivity or specificity. Conclusions In patients receiving neoadjuvant systemic therapy for invasive breast cancer, molecular subtype and systemic regimen administered did not significantly influence the sensitivity, specificity, PPV, or NPV of MRI in predicting pathologic response.