Diagnostic accuracy of mammography in characterizing breast masses using the 5th edition of BI-RADS: A retrospective study

Abstract

Background: Breast imaging-reporting and data system (BI-RADS) is intended for standardizing mammography reporting.Objectives: We aimed to evaluate the diagnostic precision of the BI-RADS assessment scoring system using histopathological findings as the reference standard. We also aimed to assess the positive predictive value (PPV) of different morphological descriptors for malignancy.Materials and Methods: This retrospective record-based analytical study was conducted in the Department of Radiodiagnosis of Acharya Harihar Post Graduate Institute of Cancer, Cuttack, Odisha, a tertiary cancer center in eastern India. We included patients attending the breast cancer unit with various breast complaints who were subjected to mammographic imaging and histopathological examination. The primary outcomes were the sensitivity, PPV, negative predictive value (NPV), and diagnostic accuracy (DA) of the BI-RADS scores for the pathological reports; secondary objective was the evaluation of the mammographic morphological characteristics. Mammography was interpreted using the BI-RADS 5th edition guidelines, without prior knowledge of the biopsy report. A BI-RADS final assessment score between 1 and 5 was assigned, where 1 indicated a normal study, 2 benign, 3 possibly benign requiring follow up, 4 suspicious requiring biopsy, and 5 indicating likely malignant requiring biopsy and further actions.Results: Between February 2020 and December 2020, we included 247 patients. All the category 5 lesions were malignant, while 76.5% of category 4 lesions were malignant. PPVs of BI-RADS categories 4a, 4b, and 4c were 38%, 90%, and 94%, respectively. Mammography had a sensitivity, specificity, PPV, NPV, and DA of 98.7%, 47.6%, 87.5%, 90.9%, and 87.9%, respectively. Morphological features that were significantly associated with malignancy were spiculated margins (P = 0.003, PPV = 100%), microlobulated margins (P = 0.005, PPV = 96.5%), irregular shape (P = 0.002, PPV = 89.6%), microcalcification (P = 0.005, PPV = 92.8%), skin thickening (P < 0.0001, PPV = 100%), and architectural distortion (P = 0.003, PPV = 96.7%).Conclusion: Digital mammography is a sensitive tool for the evaluation of breast lumps, but BI-RADS final assessment score is subjective as it depends on the interpreter's expertise.