Several studies reported clustered ring enhancement and segmental distribution have the strongest relation with malignancy as compared with other NME features of 5th BIRADS [6, 10–12]. Clustered ring enhancement is considered higher risk of malignancy and more aggressive behavior as compared with clump enhancement, heterogeneous enhancement and homogeneous enhancement [5].
Tozaki et al [6] first provided the concept of clustered ring enhancement of ductal carcinoma in situ in MR image. The study showed segmental distribution, clustered ring enhancement have the highest positive predictive value (PPV), for malignancy, 100% and 96%, respectively. The specificity of clustered ring enhancement for malignant lesions is 63% [6]. Other studies conducted by Yang, Lunkiewicz and Chikarmane, respectively, all demonstrated CRE pattern and segmental distribution are significant indicators distinguishing malignant breast lesions.
Hypothesis of formation of CRE
Two hypothesis of clustered ring enhancement are contrast media accumulation in periductal stroma or ductal wall [13] or intraductal wash-in and washout appearance and the scan time is at washout phase [10].
Pathological diagnosis in NME
The most common pathological finding of NME is pure DCIS in our study. The result is compatible with previous studies [13]. On the other hands, most DCIS, about 60–81% of cases, was interpreted as NME on the MRI [14]. NME can also be seen in invasive breast cancer, benign lesions and even normal breast tissue [15].
Common pathological diagnosis in CRE lesions
Uematsu et al [16] reported that 77% CRE lesions were malignant. Of the malignant lesions, 55% were DCIS and 45% were invasive cancers. In the case series of Machida [13], 54% CRE lesions were invasive cancers and 46% CRE lesions were carcinoma in situ. Another study reviewed by two radiologists found that CRE was significantly associated with invasive cancer (p=0.001 and p<0.001, respectively) but absence of necrosis (both p<0.001). Interestingly, they mentioned that clumped enhancement was associated with DCIS (p=0.025 and 0.001, respectively) but presence of necrosis (p=0.003 and 0.001, respectively) [17].
CRE and biomarkers of ER, PR
ER and PR are two IHC staining markers frequently seen in the breast cancer(75-80%) [7]. Several studies showed that the higher percentage of ER, PR staining, the less aggressive behaviors of the cancer [4, 18]. They are also prognostic marker of response to treatment [4, 18, 19]. We found that clustered ring enhancement had significantly negative correlation with the two hormone biomarkers.
Breast cancers can be divided into luminal type(including luminal A and B) and non-luminal type(HER2-enriches and triple-negative) [7]. The relationship of non-luminal type breast cancer (with negative ER and PR staining) and clustered ring enhancement on MR images should be further investigated.
Furthermore, the percentage of ER staining has significant difference (p= 0.028) between groups with and without CRE (50.71± 45.39 v.s. 74.26 ± 33.59) clustered ring enhancement (50.71 ± 45.39) and non-clustered ring enhancement (74.26 ± 33.59). Lower mean value of the percentage of PR staining is noted in the group with CRE cases with clustered ring enhancement NME, but without significance (CRE: 26.61 ± 34.65 vs. non-CRE: 45.74 ± 41.87, p= 0.066).
CRE and biomarkers of Ki67
Ki-67 is proved to be proliferative marker and strong prognostic indicator for overall survival and disease free survival [18, 20].
Previous study by Lee. et al, has shown high Ki-67 expression may be correlated to CRE pattern (p=0.048 and 0.003, reviewed by two radiologists) but no correlation between HER2 overexpression and enhancing pattern. The results are similar with our finding [17].
Thus, CRE pattern should be considered as a more aggressive and invasive features of breast lesion.
Limitation
First, it is a retrospective, single-institution study, and the sample size is small.
Second, there is selection bias in our study: Only cases with lesions that were interpreted as NMEs by a single radiologist at the time of initial interpretation were included in the analyses.
For lesions with cluster ring NME without pathology result, if the lesion cannot be observed on sonography or mammography, their pathology result cannot be acquired due to lack of MR-guided biopsy in our institution.
Third, there was no pathology review for this study, and therefore there may be diagnostic heterogeneity among different pathologists.