P2-08-04: Computer-Derived Breast MRI Features Have Complementary Value for Preoperative Selection of Systemic Drug Therapy in Node-Negative Stage-I/II Breast Cancer Patients.

Abstract

Background: Typical guidelines for adjuvant systemic drug therapy require assessment of lymph node status, tumor extent and grade. Increasing interest exists, however, for preoperative therapy (neoadjuvant drug therapy and preoperative partial breast irradiation). Here, representative excision specimens are not available to assess eligibility for systemic therapy. Moreover, tumor grade may be difficult to assess in core biopsies resulting in underestimation rates up to 40%. Contrast-enhanced MRI has been shown to correlate with underlying tumor biology, but it is unknown whether it can complement eligibility criteria for systemic drug therapy. Purpose: To assess the complementary value of MRI to determine eligibility for systemic drug therapy prior to breast-conserving surgery in node-negative stage-I/II breast cancer patients. Materials and methods: Patients with preoperative node-negative invasive breast cancer ≥1.0 cm and eligible for breast-conserving therapy after conventional breast imaging and MRI were consecutively included between February 2000 and March 2007. Informed consent was obtained from all patients. Eligibility to systemic therapy was determined on the basis of national guidelines using age, size, tumor grade, as well as axillary lymph node status at final pathology. Twenty temporal and morphological features of contrast uptake at MRI were automatically analyzed by a previously build CAD workstation1,2. Preoperative lymph node status was derived from ultrasound-guided fine-needle aspiration. Multivariate logistic regression and receiver-operating characteristics (ROC) analysis was used to assess associations between preoperative features (age, ER status, computer-extracted MRI features) and eligibility to systemic therapy based on post-operative features (tumor size, grade, lymph node status). Results: Three-hundred-and-eighty-six patients were included. The mean patient age was 58 years. The mean tumor size at pathology was 1.6 cm. Tumor grade was I, II and III in 128 (33%), 163 (42%), and 94 (24%) of tumors, respectively. Post-operative lymph node status was positive in 115/386 (30%) patients. In total, 262/386 (68%) patients were eligible for systemic drug therapy based on post-operative pathology. Multivariate analysis of preoperatively available features yielded ER status, and principle components 1, 6, 8 and 15 of the computer-extracted features. These components describe tumor volume and shape, heterogeneity of contrast uptake, late-enhancement kinetics, and probability of malignancy. The resulting model correlated with the Adjuvant! program (R=0.55, p<0.0001). The area under the ROC curve was 0.82 (p<0.0001). Without the model, agreement between preoperative and postoperative assessment of eligibility to drug therapy was 232/386 (60%) at 15/124 (12%) overestimation rate. With the MRI model, it increased to 258/386 (67%) (p=0.004) without increasing overestimation rate. Conclusion: Computer-analysis of contrast-enhanced MRI has potential to complement preoperative selection of stage I/II node-negative breast-cancer patients for systemic drug therapy. Ref: 1. Gilhuijs et al., Radiology 2002; 225: 907–916 2. Deurloo et al., Radiology 2005; 234: 693–701 Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr P2-08-04.