An audit of mammographic screen detected lesions of uncertain malignant potential (B3) diagnosed on initial image guided needle biopsy: how has our practice changed over 10 years?

ABSTRACT

AIM: To review all cases of B3 lesion diagnosed at initial image-guided needle biopsy over two 5-year cohorts to identify upgrade rates to malignancy and the effect of changing guidance on the management of such lesions. MATERIALS AND METHODS: Data was collected retrospectively. Mammographic features, biopsy type and management were recorded for each lesion. Upgrade rates for each B3 histological category were quantified. Statistical analysis was performed using SPSS. RESULTS: There were 224 cases in 2005-2010 and 240 cases in 2010-2015. Mammographically 211 lesions were microcalcifications, 182 masses, 65 distortions and six asymmetric densities with no difference in the mammographic features in the two cohorts. Two hundred and eight 14 G core biopsies and 256 initial vacuum-assisted biopsies were performed. There was a statistically significant reduction in benign surgical biopsies and an increase in second-line vacuum biopsy/excision in the latter cohort, with no significant change in the upgrade rate. There was an overall 6% upgrade to invasive malignancy and 13% upgrade to ductal carcinoma in situ (DCIS). The upgrade rates for the following histological categories were atypical intraductal epithelial proliferation (AIDEP) 33.2% (21/63); classical (not pleomorphic) in situ lobular neoplasia (ISLN) 18.2% (6/33); flat epithelial hyperplasia (FEA) 21.7% (20/92); papilloma with atypia 53.8% (7/13), without atypia 12.1% (8/66); and radial scar/complex sclerosing lesion with atypia 16.7% (2/12), and without atypia 7.9% (6/76). CONCLUSION: Upgrade rates remain high for some histological categories even with first-line use of vacuum biopsy. Management of borderline lesions should be considered carefully in a multidisciplinary meeting. In many cases, the need for diagnostic surgical excision has been replaced by image-guided vacuum sampling.