se of BI-RADS-MRI descriptors for differentiation between mucinous arcinoma and fibroadenoma

Objective: We evaluated the latest breast imaging reporting and data system (BI-RADS) magnetic resonance imaging (MRI) (5th edition) descriptors and non BI-RADS MRI factors that contribute to differentiation between mucinous carcinomas (MCs) and fibroadenomas (FAs). Materials and methods: This retrospective study included 27 patients with P-MCs or M-MCs similar to P-MCs and 22 patients with FAs who underwent breast MRI between October 2008 and July 2014 at our institution. Definitive histopathological diagnoses were made for all of the MCs and FAs. The latest BI-RADS MRI descriptors for abnormal enhancement, including maximum diameter, shape (irregular or round/oval), margin (irregular or circumscribed), rim enhancement (present or absent), dark internal septation (absent or present), delayed internal enhancement (heterogeneous or homogeneous), and the time–intensity curve pattern (not persistent or persistent) were evaluated. As additional non BIRADS MRI factors related to differentiation between MC and FA, age, signal intensity in the T2-weighted image (high or not high), extent of lobulation (strong or weak), enhancing internal septation (present or absent), and the apparent diffusion coefficient value were also evaluated. One radiologist retrospectively evaluated interpreted MR findings and analyzed the findings. Statistically significant findings were identified through univariate and multivariate analyses. Then, three blinded radiologists reviewed the MR images where MR findings had shown a significant association with outcomes during univariate analyses. Independently, the three blinded readers reviewed the MR images for the evaluation of inter-observer variability, and then arrived at a consensus for the evaluation of observer performance. Observer performance and inter-observer variability were determined via a receiver-operating-characteristic curve analysis and weighted k statistics. The sensitivity, specificity, and accuracy of each of the MR findings were calculated. Results: Univariate analyses showed that irregular margins were observed more frequently in MCs than in FAs (11/27, 41% vs. 1/22, 0.5%, p < 0.05). MCs also showed rim enhancement, delayed heterogeneous enhancement, and enhancing internal septation more frequently than FAs (p < 0.05). FAs showed circumscribed margins more frequently than MCs (21/22, 95% vs. 16/27, 59%, p < 0.05). FAs also showed dark internal septation more frequently than MCs (18/22, 82% vs. 3/27, 11%, p < 0.05). In multivariate analyses, the most significant feature in lesion characterization was delayed heterogeneous enhancement. In the blinded reading, a combination of irregular margin and delayed heterogeneous enhancement showed the highest sensitivity (96.3%) and accuracy (87.8%). Enhancing internal septation showed the highest specificity (90.9%). The values with confidence ratings for differentiation between MCs and FAs were 0.63–0.67, which showed substantial agreement among the three radiologists. Conclusions: The combination of irregular margin and delayed heterogeneous enhancement and enhancing internal septation were significant findings for differentiation between P-MC or M-MC similar to P-MC and FA. Abbreviations: BI-RADS, breast imaging reporting and data system; MRI, magnetic iffusion coefficient; MMG, mammography; US, ultrasonography; AUC, area under the cu ∗ Corresponding author. E-mail address: igarashi-t@jikei.ac.jp (T. Igarashi). ttp://dx.doi.org/10.1016/j.ejrad.2016.03.012 720-048X/© 2016 Elsevier Ireland Ltd. All rights reserved. © 2016 Elsevier Ireland Ltd. All rights reserved. resonance imaging; MC, mucinous carcinoma; FA, fibroadenoma; ADC, apparent