Prediction of Low versus High Recurrence Scores in Estrogen Receptor-Positive, Lymph Node-Negative Invasive Breast Cancer on the Basis of Radiologic-Pathologic Features: Comparison with Oncotype DX Test Recurrence Scores.

Purpose To review mammographic, ultrasonographic (US), and magnetic resonance (MR) imaging features and pathologic characteristics of estrogen receptor (ER)-positive, lymph node-negative invasive breast cancer and to determine the relationship of these characteristics to Oncotype DX (Genomic Health, Redwood City, Calif) test recurrence scores (ODRS) for breast cancer recurrence. Materials and Methods This institutional review board-approved retrospective study was performed in a single large academic medical center. The study population included patients with ER-positive, lymph node-negative invasive breast cancer who underwent genomic testing from January 1, 2009, to December 31, 2013. Imaging features of the tumor were classified according to the Breast Imaging Reporting and Data System lexicon by breast imagers who were blinded to the ODRS. Mammography was performed in 86% of patients, US was performed in 84%, and MR imaging was performed in 33%, including morphologic and kinetic evaluation. Images from each imaging modality were evaluated. Each imaging finding, progesterone receptor (PR) and human epidermal growth factor receptor 2 (HER2) status, and tumor grade were then individually correlated with ODRS. Analysis of variance was used to determine differences for each imaging feature. Regression analysis was used to calculate prediction of recurrence on the basis of imaging features combined with histopathologic features. Results The 319 patients had a mean age ± standard deviation of 55 years ± 8.7 (range, 31-82 years). Imaging features with a positive correlation with ODRS included a well-circumscribed oval mass (P = .024) at mammography, vascularity (P = .047) and posterior enhancement (P = .004) at US, and lobulated mass (P = .002) at MR imaging. Recurrence scores were predicted by using these features in combination with PR and HER2 status and tumor grade by using the threshold of more than 30 as a high recurrence score. With a regression tree, there was correlation (r = 0.79) with 89% sensitivity and 83% specificity. Conclusion On the basis of preliminary data, information obtained routinely for breast cancer diagnosis can reliably be used to predict the ODRS with high sensitivity and specificity. (©) RSNA, 2016.