Evaluation of Suspicious Appearing Microcalcification Groups on Mammogram: Comparing BI-RADS 5th Edition Descriptors with the BI-RADS 4th Edition and Pathologic Association

Abstract

Background: Groups of microcalcifications are the most frequent recognized features of ductal carcinoma on mammograms. However, heterogeneity (in size, morphology and density) and number of microcalcification groups as well as presence of accompanied soft-tissue density are not included in breast imaging reporting and data system (BI-RADS) descriptors. Objectives: The study purposes to determine the malignancy risk of microcalcification groups regarding these characteristics and also compare the 4th and 5th versions of BI-RADS. Patients and Methods: In a cross sectional study, 88 patients with microcalcification groups (age range, 26 - 80 years; mean, 53.4 years) who had undergone mammographically guided biopsy between March 2013 and March 2015 were evaluated. The overall number of microcalcification groups in each patient, number of deposits within each group, group location and heterogeneity in size, density and morphology were assessed and subsequently BI-RADS descriptors for 4th and 5th editions were recorded separately. Finally, correlation with histopathology was performed. Results: Overall, positive predictive value (PPV) of suspicious microcalcifications was 22.4%. PPVs of morphology descriptors were as follows: amorphous, 7.9%; coarse heterogeneous, 17.8%; fine pleomorphic, 63.2%; fine linear/fine linear branching, 100%; (P < 0.001). Heterogenicity in size existed in 81 cases (92%), in density in 69 cases (86.4%) and in morphology in 86 cases (97.7%). Additionally, microcalcification groups that were accompanied with soft-tissue density had a higher percentage of malignancy (67.5% vs. 54.5%) but with no significant difference (P = 0.2). According to BI-RADS 4th edition, the risk of malignancy was 49.1%, 66.7% and 88.1% in 4b, 4c and 5, respectively. These figures were 30%, 82.9%, and 100% for BI-RADS 5th version, respectively. The area under the receiver (AUC) of 4th and 5th versions of BI-RADS was 0.76 and 0.74 (both P values < 0.001, 95% confidence intervals = 0.66-0.87 and 0.63 - 0.85 respectively). P value for comparison was insignificant. Conclusion: The risk of malignancy increased with the heterogeneity of microcalcifications, especially in the groups with heterogeneity in density, however with no statistically significant difference. BI-RADS 5th edition could predict the likelihood of malignancy as well as 4th version.