Breast MR biopsy: Pathological and radiological correlation.

Abstract

To identify pathological features for sample analysis of magnetic resonance imaging-guided vaccum-assisted breast biopsy (MRIgVaBB) to optimize radio pathological correlation and identify discordant benign result. Databases of two centres were queried to identify MRIgVaBB performed between January 2009 and February 2013. A cohort of 197 women (mean age: 54.5 years (24-77)) with 208 lesions was identified. We retrospectively analyzed all prebiopsy MRI examinations according to the new BI-RADS lexicon, and all biopsy samples to describe the lesion of interest, its interface with the surrounding breast tissue and other associated features. The malignancy rate was 26.0 % (54/208) with an underestimation rate of 15.67 % (5/32). A visible interface at pathology between a biopsied lesion and the surrounding breast tissue was more frequently identified in mass enhancement compared to NME or focus (p = 0.0003). Regional NME was correlated with a high degree of fibrosis (p = 0.001) and the presence of PASH (p = 0.0007). Linear or segmental NME was correlated with the presence of periductal mastitis (p = 0.0003). The description of a visible interface between the target lesion and the surrounding tissue is crucial to confirm the correct targeting of an MR mass or a NME. • Pathological interface correlated with magnetic resonance mass and focal non-mass enhancement (NME). • Linear or segmental NME correlated with mastitis or ductal carcinoma in situ. • Fibrosis and pseudoangiomatous stromal hyperplasia (PASH) are correlated with regional NME.