Prediction of benignity or malignancy of a lesion using BI-RADS

To determine the Positive Predictive Value (PPV) of the Breast Imaging Reporting and Data System (BI-RADS) categories as confirmed by histopathology of localized lesion.

An observational study.

All patients undergoing mammographic needle localizations for breast lesions from January 2001 to December 2003 were included in the study. The study was conducted at the Radiology Department of Aga Khan University Hospital, Karachi.

The medical records of all patients undergoing needle localization during the study period were retrospectively reviewed. In a total of 111 patients, 114 mammographically-guided needle localizations were performed for breast lesions. In all these patients the mammographic finding and BI-RADS category were recorded. Statistical analysis was done on SPSS. Histopathological results of the localized sample were used as a gold standard to calculate the PPV. The PPV for each assessment category was calculated by dividing the number of patients positive for malignancy on histopathological examination by the total number of patients who underwent biopsy for each BI-RADS assessment category.

In 114 localizations the histopathologic finding showed 25 (22%) cancers (10 [40%] were ductal carcinoma in situ (DCIS); (15 [60 %] were invasive cancers) and 89 (78 %) benign lesions. All the mammographic lesions were categorized as, 7 (6%) category 0 lesions; 3 (2%) category 1 lesions, and 30 (26%) category 2 lesions; the PPV was 0% for category 0,1,2 (0% of 7,3,30 lesions respectively). There were 18 (16 %) category 3 lesions; the PPV was 5% (1 of 18 lesions). One case of cancer in this group was DCIS. There were 50 (44%) category 4 lesions; the PPV was 34% (17 of 50 lesions). There were 6 (5%) category 5 lesions; the PPV was 83% (5 of 6 lesions).

The number of biopsies performed for benign lesion can be reduced by using BI-RADS categories, as it has a good PPV.