Characterization of the enhancing lesions on dynamiccontrast enhanced magnetic resonance imaging in patientswith interstitial mammoplasty

Purpose: The purpose of this study was to categorize the morphologic and kinetic features of enhancinglesions in breasts with interstitial mammoplasty using dynamic contrast-enhanced magnetic resonanceimaging and to assess factors predictive of breast cancer.Materials and methods: We retrospectively reviewed the clinical and radiological data of 21 enhancinglesions in 19 patients with interstitial mammoplasty, who underwent breast magnetic resonance imagingand biopsy or an operation in our hospital from September 2008 to July 2012. These lesions were sortedby morphological and kinetic features and final assessment category according to the BI-RADS lexicon.Results: Nine cases were confirmed to be ductal carcinoma in situ (n = 2) and invasive ductal carcinoma(n = 7), and the remaining 12 cases were fibrocystic disease (n = 2), fibroadenoma (n = 2), fat necrosis(n = 1), foreign body granuloma (n = 3) and silicone mastitis (n = 1).Common features of malignancy included irregular shape (50.0%), spiculated margins (75.0%), hetero-geneous enhancement (50.0%) and type III kinetic pattern (87.5%). The correlations of margins and kineticcurve pattern with benignity and malignancy approached statistical significance (p = 0.02, respectively).We found no correlation for shape (p = 0.33) or internal enhancement (p = 0.42) between lesion types.The malignancy rate of enhancing lesions was 42.8% (9/21). The sensitivity and specificity of dynamiccontrast-enhanced magnetic resonance imaging were 100% and 16.67%, respectively. The positive pre-dictive value, negative predictive value and accuracy of magnetic resonance imaging were 47.38%, 100%and 52.38%. Overall inter-observer agreement for the kinetic curve pattern was good ( = 0.67). Moder-ate agreement was seen in describing the shape, margin, enhancement and assessing the final category( = 0.59, 0.46, 0.58 and 0.49, respectively).Conclusion: Dynamic contrast-enhanced magnetic resonance imaging had a high sensitivity, negativepredictive value for the prediction of breast cancer but a low specificity due to features of foreign body-related lesions that mimicked malignant lesions. The significant predictive factors for malignancy weremargins, kinetic curve pattern and final assessment category. Overall inter-observer agreement for thekinetic curve pattern was good.