Non-mass breast lesions on ultrasound: final outcomes and predictors of malignancy.

Background Breast cancer can present as non-mass lesions (NMLs) on ultrasound. However, knowledge of and understanding about NMLs are scarce. Purpose To retrospectively investigate the final outcomes of sonographic breast NMLs and determine the clinical and radiologic variables associated with malignancy Material and Methods In our radiologic database of breast ultrasound examinations between 2011 and 2014, we found 119 women with 121 NMLs with available histopathologic or sonographic follow-up (over 2 years) data. We collected the clinical variables (patient's age, symptoms, and mammographic density) and histopathologic data as well as radiologic variables (mammographic and ultrasound findings) after retrospective review by two radiologists, the authors of the current paper, in consensus. We classified the ultrasound findings according to distribution (focal, linear or segmental, and regional) and associated features (calcification, architectural distortion, and ductal changes) and analyzed the associations between variables and malignancy using the t test and χ test. Results Of the 121 NMLs, 88 (72.7%) were benign and 33 (27.3%) were malignant. Ductal carcinoma in situ (DCIS) (17/33, 51.5%) and invasive ductal cancer with or without DCIS (13/33, 39.4%) comprised the main malignancies, and malignancy was significantly associated with palpability ( P = 0.000). Mammographic findings and sonographic distribution and associated features were significantly different between benign and malignant lesions ( P = 0.000, P = 0.004, and P = 0.001, respectively). Malignant lesions showed more frequent calcifications combined with asymmetry ( P = 0.000) on mammography and linear-segmental distributions ( P = 0.001) and associated calcifications ( P = 0.019) or architectural distortions ( P = 0.015) on ultrasound. Conclusion Breast NMLs on ultrasound showed high risk of malignancy. Symptoms and mammographic and ultrasound findings can be possible predictors of malignancy in NMLs.