Positive Predictive Value of Malignancy in BI-RADS 4 and 5 Breast Lesions

Abstract

Objective: To determine the positive predictive value (PPV) of the breast imaging and data system (BI-RADS) category 4 and 5 in diagnosis of breast cancer. The associations of abnormal mammographic findings with BI-RADS 4 and BI-RADS 5 were also determined.

Methods: We identified the list of women who had mammography and ultrasonography in Department of Radiology from January 1st, 2009 to December 31st, 2011. Women who were diagnosed as BI-RAD 4 or BI-RADS 5 and had histopathologic diagnosis were included into the study. Data collected were age, abnormal mammographic and ultrasonographic findings, BI-RADS category and pathological results. The association between each imaging feature, BI-RADS category and pathologic results were analyzed.

Results: Total of 7,705 women had mammography in our institution during the study period. Three hundred and two women met inclusion criteria and were included in the study. Mean age was 51.6 ± 11.1 years. BI-RADS 4 diagnosis was made in 74.5% and BI-RADS 5 in 25.5%. Mass was the most common imaging abnormality found in 62.3%. We found that 54.3% of breast lesions was malignant while the other 45.7% was benign. The most common malignant lesion identified in 86.6% was invasive ductal carcinoma. Fibroadenoma and fibrocystic change were the most common benign lesions, found in 20.3% and 18.1% respectively. The positive predictive value of breast cancer in BI-RADS 4 and 5 were 40.4% and 94.8%, respectively.

Conclusion: The BI-RADS categorization is useful in predicting the chance of malignancy. BI-RADS categories 4 and 5 in this study had a probability of malignancy of 40.4% and 94.8%, respectively. Preoperative biopsy should be performed before performing definite treatment in all these lesions.