To develop a predictive scoring system for ultrasound-detected B3 lesions at ultrasound-guided core needle biopsy (US-CNB). A total of 2724 consecutive US-CNBs performed in our Institution (January 2011 to December 2014) were retrospectively reviewed. Inclusion criteria were as follows: (a) histopathological examination of the entire lesion or (b) availability of radiologic follow-up (FUP) ≥24 months. Patient- and lesion-related variables-patients' age, lesion consistency, lesion size, vascularization, BI-RADS category, and US-CNB result-were analyzed. Positive predictive values (PPVs) for malignancy were calculated correlating US-CNB results with excision histology or FUP. A scoring system for underlying malignancy was developed using risk factors weighting. A total of 102 B3 lesions were included: 27 atypical ductal hyperplasia (26.5%), 5 lobular intraepithelial neoplasia (4.9%), 32 radial scar (31.4%), 37 papillary lesions (36.3%), and 1 fibroepithelial lesion (0.9%). Surgery was performed on 71/102 (69.6%) lesions, and 22/71 were malignant; the remaining 31/102 lesions (30.4%) were unchanged at FUP. The overall PPV for malignancy was 21.6%. Patients' age (odds ratio [OR] = 3.63, P = 0.008), lesion consistency (OR = 5.96, P = 0.001), BI-RADS category (OR = 17.52, P < 0.001), and CNB result (OR = 3.6, P = 0.008) were associated with a higher risk of malignancy underestimation and selected as risk factors in the score definition. Two risk groups were identified: low (0-2 points) and high risk (3-5 points), with significantly different risk of malignancy underestimation (8.0% vs 59.3%, P < 0.001). The proposed score helps to predict the risk of malignancy underestimation and choose the management of B3 lesions at US-CNB.
Keywords: B3 lesions; borderline breast lesions; breast ultrasonography; core needle biopsy; malignancy underestimation.
© 2018 Wiley Periodicals, Inc.