P2-08-02: Magnetic Resonance Imaging as a Predictor of Pathologic Response in Patients Treated with Neoadjuvant Systemic Treatment for Operable Breast Cancer (TBCRC 017).

Abstract

Abstract Background: Increased pathologic complete response (pCR) rates observed with neoadjuvant chemotherapy for invasive breast cancer has prompted interest in whether patients with pCR can be identified preoperatively and potentially spared the morbidity of surgery. This multicenter retrospective study was performed to determine the accuracy of preoperative MRI in predicting pCR in the breast and whether MRI performance differs by molecular subtype, histology, and treatment regimen. Methods: 770 women from 8 institutions were retrospectively identified as having received neoadjuvant systemic therapy with MRI obtained at baseline and after completion of systemic treatment. Tumor phenotypes were defined on the basis of estrogen and progesterone receptor (ER/PR or HR) and HER2 receptor status. Univariate and multivariate analyses of factors influencing radiographic complete response (rCR) and pCR were recorded, with rCR defined as resolution of any abnormal enhancement, mass, or distortion on MRI, and pCR defined as resolution of both invasive disease and DCIS. Results: rCR and pCR for the total group were 182/746 (24%) and 179/746 (24%), respectively, with the highest rate of pCR seen among the triple-negative (TN; 57/155; 37%) and HR-/HER2+ (38/101; 38%) subtypes. Covariates significantly associated with rCR included T stage (p=0.0002), tumor grade (p=0.005), IHC phenotype (p=0.005), and chemotherapy regimen (p<0.0001). On multivariate analysis, only tumor phenotype was independently associated with likelihood of rCR, with both TN (OR = 2.00, 95% CI 1.20−3.33) and HR-HER2+ (OR=2.30, 95% CI 1.09–4.83) more likely to achieve rCR than HR+HER- (reference group). Overall accuracy of MRI for prediction of pCR was 74%. Sensitivity, NPV, PPV, and accuracy differed significantly among tumor subtypes, with the greatest NPV in the HR-/HER2+ and TN subtypes (table1). Among patients with rCR, ER- status (OR=6.4, 95% CI 1.1 to 35.6), PR- status (OR=3.8, 95% CI 1.2 to 11.4), and tumor grade of 3 vs 1 or 2 (OR=2.49, 95% CI 1.22−5.07) were independently associated with likelihood of pCR. Discussion: MRI performance for predicting pCR in patients with invasive breast cancer receiving neoadjuvant systemic therapy differed significantly among breast cancer subtypes; however this difference is likely due to subtype differences in frequency of pCR and not to intrinsically better or worse MRI detection. The relatively low NPV of MRI following neoadjuvant systemic therapy does not support using MRI rCR alone to accurately identify those patients that can safely avoid surgery. Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr P2-08-02.