Histopathological correlation of non-mass enhancement detected in breast magnetıc resonance ımagıng

HİSTOPATHOLOGİCAL CORRELATİON OF NON-MASS ENHANCEMENT DETECTED İN BREAST MAGNETIC RESONANCE IMAGING Purpose: The aim of this retrospective study is to evaluate the distribution patterns, morphologic properties, internal enhancement features and enhancement kinetics with a correlation of histopathologic diagnoses of non-mass enhancement (NME) in breast MRI. Additionally, in cases diagnosed with malignant tumors, we investigated whether internal enhancement features and distribution patterns of NME are correlated with molecular subtypes of breast cancer. Materials and Methods: We included 104 patients diagnosed with non-mass enhancement (NME) in breast MRI that were scanned in Ege University School of Medicine, Department of Radiology between January 2013-November 2015. Patients without histopathologic diagnoses (25 cases) and the ones with mass (18 cases) were excluded. Lesions were classified based on their distribution patterns, morphologic properties, internal enhancement features, enhancement kinetics, background parenchymal enhancement and presence of additional findings. The histopathological results of the lesions (n=62) were malignant (n=39, %62,9), benign (n=18, %29) and high risk lesions (n=5, %8.1). Imaging was performed by Siemens Symphony Vision, 1,5T MRI, and Siemens Verio 3T MRI devices. Conventional sequences and contrast-enhanced dynamic images were obtained from these studies. NME distribution patterns, morphologic properties, internal enhancement features and enhancement kinetics and their evaluations were based on the lexicon described in ACR BI-RADS Atlas (5th edition, 2013). Statistical analysis of the data obtained from the study was done with IBM SPSS (Statistical package for the social sciences) V21, by Department of Statistics at Ege University. Cross-tabulation and chi-square test were used since the components of data consisted of classifications, mainly Pearson Chi-Square and Fisher's Exact test. Confidence intervals were chosen as %95 (with α error of %0.05). Results: Distribution patterns of NME's were distributed as focal (%25.8), segmental (%22.6), multiregional (%22.6), regional (%14.5), linear (%8.1) and diffuse (%6.5). In our study, similar to the other studies in the literature, focal distribution was associated with benign histology, while segmental or regional patterns were associated with malignant. Lesion enhancement with contrast was most commonly seen as heterogeneous type (%50), while homogeneous (%19.4), clustered ring (%16.1) and clumped (%14.5) internal enhancement patterns were of similar rates. Both in malignant and benign cases, the heterogeneous internal enhancement was the most common type of contrast enhancement. No significant association between internal enhancement characteristics and histopathologic results was found. Among distribution patterns, the most specific one was the diffuse type (%100), followed by multiregional (%95,2) and segmental (%90,5). Specifity among internal enhancement patterns were highest for the clustered ring (%95,2) and clumped (%90,5). Positive predictive values for the clustered ring and clumped were %90 and %77, respectively. Background parenchymal enhancement was most commonly seen as minimal type (%50), followed by significant (%19,3), moderate (%17,8) and mild (%12,9). Association between background parenchymal enhancement and malignancy was also established, especially for 'significant' and 'asymmetric' types. Other findings that may accompany NME were nipple retraction, thickening of skin and skin invasion, edema, structural distortion, axillar lymphadenopathy, chest wall invasion, which were found significant for prediction of malignancy. Type 2 and 3 enhancement kinetics were the most common type for both malignant and benign lesions. While type 3 was seen more in malignant lesions, no significant associations were found for the distinction between malignant or benign lesions. The results were similar to the former studies in literature, parameters of enhancement kinetics did not play a role in predicting histopathological nature of tissues sampled from NME. Breast carcinoma classification based on immunohistochemical studies were based on ER, PR, HER-2, Ki-67 positivities, and types were luminal A, luminal B, HER-2 and triple negative. No associations between distribution patterns, internal enhancement patterns, and molecular subtypes were found. Ki-67 indices of cases were categorized as <%10, %10-30, >%30. All of the non-mass malignant lesions with focal distribution had Ki-67 index <%10. For other distribution patterns, indices were %10-30 or >%30. We believe this finding shows mitotic indices might have effects on distribution patterns. Conclusion: Breast carcinomas are heterogeneous tumors which might show different clinical, histopathological and molecular properties. MRI of the breast is being used increasingly because of its ability to spot lesions and solve problems encountered. For evaluating NME in MRI, distribution patterns, internal enhancement features, background parenchymal enhancement and additional features are important, since they can give insight into benign or malignant nature of the lesion. While molecular phenotypes might effect tumor behavior, morphologic or imaging features, histopathologic studies are the gold standard for the distinction of subtypes. In an age when image-guided biopsies are routinely performed, it might be wise to use enhancement and distribution properties of NME's in MRI for prediction of histopathologic results and for making correlations and for deciding on this basis whether to repeat the biopsy or to make changes necessary in follow-up protocols when in need.