Level of suspicion of a mammographic lesion: use of features defined by BI-RADS lexicon and correlation with large-core breast biopsy.
OBJECTIVE: To determine whether the categories defined in the Breast Imaging Reporting and Data System (BI-RADS) are useful predictors of malignancy and to assess their positive predictive value. PATIENTS AND METHODS: A retrospective study was undertaken from July 1, 1994, to June 30, 1995. Of 568 abnormal mammograms for which largecore biopsy was recommended, 495 were available for review. The mammographic features of the lesions, as defined by the BI-RADS lexicon, were used to assess the level of suspicion. The lesions were classified as "probably benign," "suspicious" or "highly suggestive of malignancy." These diagnostic impressions were then correlated with the histologic diagnosis made after core biopsy. RESULTS: The diagnostic impressions before core biopsy were as followed: 16 (3%) benign lesions, 397 (80%) suspicious lesions and 82 (17%) lesions highly suggestive of malignancy. All benign lesions remained in the same category after core biopsy, whereas, of the suspicious lesions, 91% were diagnosed as benign, only 4% as malignant, and the other 5% as atypical hyperplasia. Among the lesions highly suggestive of malignancy according to the BI-RADS lexicon, 54% were found to be malignant at core biopsy. CONCLUSION: The BI-RADS lexicon is helpful in discriminating between lesions that are probably benign and probably malignant from the mammographic features. However, the rate of malignancy in the "suspicious" category is low.