According to univariate analysis, mucinous carcinoma was frequently detected in older age-group, was larger than fibroadenoma on presentation and had more frequent irregular borders and non-circumscribed margins, and all mucinous carcinoma had high T2 signal. Fibroadenomas had more frequent homogenous enhancement than mucinous carcinoma, none of the fibroadenomas had rim enhancement or enhancing internal septations, while mucinous carcinoma had frequent rim enhancement and enhancing internal septations. Fibroadenoma had frequent dark non-enhancing internal septations than mucinous carcinoma. On ADC map and MRS, mucinous carcinoma had higher ADC values and higher CHO.SNR than fibroadenoma. On multivariate analysis, mucinous carcinoma had a significant combination of being common in older age, larger in size, with irregular shape, and non-circumscribed margins.
According to our results, the presence of type 3 (washout curve), dark internal septations, non-circumscribed margins and irregular shape of the lesion showed the highest sensitivity and accuracy for differentiation of mucinous carcinomas and fibroadenomas (100, 78.6%), (89.4, 82%), (86.5, 80.3%) and (81, 71.4%), respectively.
The previous study performed by Igarashi et al. [2] showed that combination of irregular margin and delayed heterogeneous enhancement showed the highest sensitivity and accuracy (96.3%, 87.8%) for differentiation of mucinous carcinomas and fibroadenomas.
Regarding lesion shape and margins: Mucinous carcinomas had more frequent irregular shape and non-circumscribed margins (13/19, 68.4%) than fibroadenomas). In agreement with our results, Igarashi et al. [2] reported that some mucinous carcinomas showed non-circumscribed margins in the univariate analysis (11/27, 41%). Margin irregularity is more characteristic of mixed mucinous carcinomas and is attributed to fibrotic and infiltrative nature of their non-mucinous component.
Regarding lesion enhancement: In our study, 34 fibroadenoma cases (91.9%) had homogenous enhancement and only three cases (8.1%) showed heterogeneous enhancement which may be attributed to cystic changes. Patterns of mucinous tumor enhancement in dynamic study depend on the amount of mucin versus the solid tumor tissue. Mucinous carcinoma showed homogenous enhancement in only two cases (10.5%): predominant solid part, heterogeneous enhancement in four cases (21%) corresponding to islands of neoplastic epithelial cells floating in mucin, and rim enhancement (Fig. 2) in six cases (31.5%) corresponding to a peripheral distribution of tumor cells and central location of mucin, that was in consistent with a previous study [8] which reported that hypocellular pure mucinous tumors have typical pattern of gradual enhancement and hypercellular pure tumors have strong early enhancement. Gradual enhancement pattern in mucinous carcinomas is due to large amount of mucin, and diffusion of contrast throughout the stroma and around the epithelial component takes time [9].
Internal septations: Internal septations within the lesion may be either dark non-enhanced or enhanced septa. Dark non-enhanced internal septations were detected in 29 fibroadenoma cases (78.4%) in our study; lack of enhancement in these septa is due to the presence of hypovascular collagenous bands within the lesion. While enhanced internal septations were detected in seven mucinous carcinoma cases (36.8%) in our study with 100% specificity and 77.4% accuracy, which is attributed to the vascular nature of the internal fibrous bands in contrast to the non-vascular bands in fibroadenomas, this have been suggested to differentiate mucinous carcinoma from fibroadenoma in consistent with previous studies [2] which reported that enhancing septations on dynamic CE-MRI was a characteristic finding of mucinous carcinomas (Fig. 3).
Dynamic curve
In our study, type 1 and type 2 curves were detected in both fibroadenoma and mucinous carcinoma; type 1 curve was of significance only for fibroadenoma cases; it was detected in 29 (78.3%) fibroadenoma cases with high specificity and accuracy (96.6, 83.9%). Type 3 curve was detected only in mucinous carcinoma (7 cases, 36.8%) with high specificity and accuracy (100% and 78.6%). This was not in agreement with Igarashi et al. [2] who reported that there were no significant differences in the kinetic patterns between mucinous carcinoma and fibroadenoma.
In our study, morphological features assessment was predictive for differentiation between mucinous carcinoma and fibroadenoma after exclusion of type 3 curves in dynamic study as none of fibroadenomas showed type 3 curve in our results. Previous studies were in agreement; they reported that assessment of the morphological features was more predictive in differentiation between mucinous carcinoma and fibroadenoma and in diagnosis of breast cancer than was characterization of the kinetic curves [10, 11].
T2 signal: In our study, the lesion signal was not specific for any point of differentiation as all cases of mucinous carcinomas and 27 fibroadenoma cases had high signal on T2 WI, while for fibrous or calcified fibroadenomas (10 cases) their T2 signal was not high.
Functional assessment (Fig. 4, and Fig. 5)
Our study showed a significant difference in the mean ADC value between fibroadenoma and mucinous carcinoma; fibroadenomas mean ADC value was 1.19 which was significantly lower than mucinous carcinoma (1.39). This was in agreement with other previous studies [12, 13], reporting that among malignant breast tumors, mucinous carcinoma displayed the highest ADC values; they had a mean ADC of 1.8 × 10−3mm2/s, statistically significantly greater than that of benign lesions (1.3 × 10−3mm2/s) and other malignant tumors (0.9 × 10−3mm2/s), due to the large amount of the extracellular protein.
Others reported that mean ADC values of mucinous tumors in areas rich in tumor cells were lower than areas containing large amounts of mucin and small amounts of tumor cells and were markedly higher than solid component in conventional invasive duct carcinoma (IDC) [9].
This was not in agreement with Igarashi et al. [2] as they reported no significant difference in ADC values between fibroadenoma and mucinous carcinoma attributing this finding to T2 shine-through effect.
In our study, MRS showed significant CHO.SNR in mucinous carcinomas which was 3.1, while in fibroadenomas it was 0.8 (Fig. 6). This was in agreement with previous studies [14, 15] which assessed diagnostic accuracy of MRS in differentiating benign and malignant breast lesions considering the lesion of malignant nature if the signal-to-noise ratio of choline resonance peak was equal to or greater than 2.
Limitations
There are few limitations to this study. First was the small sample size as mucinous breast carcinoma being not a common pathological diagnosis, also multiple analysis items were included. Second, cases with typical fibroadenomas were diagnosed by US; MRI was not mandatory for diagnosis nor histopathological sampling, so a larger number of fibroadenomas are necessary to evaluate both morphologic and kinetic features for confident diagnosis. Third, there may be bias from the manual selection of region of interest. Application of automated approach would be preferred. Fourth, our study was retrospective, and some selection bias may have been present which may have limited the accuracy of results.