BI-RADS 4 and 5 lesions on mammography and ultrasonography: Positive predictive values and analysis of discordant cases

Abstract

Background: Breast cancer is the leading cause of morbidity and mortality among women worldwide, including Thailand. Early detection and treatment significantly decrease mortality. The accuracy and effectiveness of mammograms are important in early breast cancer detection. In the present study, the positive predictive values of BI-RADS categories were compared against the suggested values and the histopathological results for the purposes of auditing, quality assurance, and benchmarking. 

Objective: To determine the positive predictive values of different BI-RADS categories and compare the values against the suggested values based on the BI-RADS 2013 Standard, and to describe the discordant cases (cases with a mismatch between imaging and BI-RADS categorization) to determine how the misdiagnoses might happen.

Materials and Methods: This retrospective diagnostic study was conducted at Srinagarind Hospital, Thailand, between 1 January 2017 and 31 December 2021. It included 1,666 breast lesions in BIRADS 4 and 5 that had undergone image-guided biopsy (ultrasound or tomosynthesis-guided biopsy) to assess the positive rates of breast cancer among these lesions.   

Results: The 1,666 breast lesions were in the BI-RADS subtype categories 4A (n=899), 4B (n=492), 4C (n=149), and 5 (n=126), with positive rates for breast cancer of 5.01%, 19.51%, 57.05%, and 89.68%, respectively, with category 5 showing a lower positive rate than the standard. The discordant cases were mostly overdiagnoses of benign morphologies. 

Conclusions: The positive rates of lesions in the BI-RADS 4 subcategories were within the expected ranges, whereas the positive rate of lesions in the BI-RADS 5 category was lower than expected, indicating overcategorization.