Magnetic resonance imaging of fibroadenoma-like lesions and correlation with breast imaging-reporting and data system and kaiser scoring system

Abstract

Background: Multiple breast lesions resembling fibroadenomas are a common imaging finding in patients presenting to the mammography unit at Dr George Mukhari Academic Hospital in the North-West district of Tshwane, South Africa. Patients often present with multiple lesions, up to 20 lesions per breast. These lesions often have atypical features on ultrasound and/or a clinical history of growth is commonly given. Phyllodes tumours may be indistinguishable from fibroadenomas and breast cancers may on occasion present with benign features, which can lead to misdiagnosis. Breast magnetic resonance imaging (bMRI) evaluation of lesions resembling fibroadenomas may improve accurate assessment and identification of lesions requiring biopsy. Objectives: To assess the reliability of bMRI to characterise lesions resembling fibroadenomas on ultrasound, using the Breast Imaging-Reporting and Data System (BI-RADS) and Kaiser scoring systems with histopathological correlation. Method: A quantitative, prospective, investigative study was performed with a sample size of 100 breast lesions among a total of 35 patients at Dr George Mukhari Academic Hospital. Patients were recruited after a breast ultrasound investigation revealed lesions resembling fibroadenomas, but with an indication for ultrasound-guided biopsy, for example, very large size, atypical features on ultrasound or a history of recent growth. The bMRI was performed prior to the ultrasound-guided breast biopsies. Three investigators independently evaluated the bMRI and applied BI-RADS descriptors to each lesion. The Kaiser score was then calculated for each lesion. Statistics were calculated using Pearson’s and Spearman’s coefficients for inter-reader variability, kappa scores for BI-RADS and Kaiser score correlation with histology. Results: Evaluation with bMRI, BI-RADS and the Kaiser scoring system showed statistically significant correlation with each other and with histopathology results for each lesion. There was statistically significant agreement among the investigators regarding the interpretation of the lesions and allocation of appropriate BI-RADS scores. Conclusion: Multiple lesions resembling fibroadenomas can be evaluated with bMRI when multiple breast biopsies would not be feasible. With a good imaging protocol and technique, adequate interpretation skills by the radiologist and the use of the Kaiser scoring system, an accurate diagnosis can be achieved.