PURPOSE: To analyze the features of breast complex cystic lesions at ultrasonography (US) and to determine appropriate Breast Imaging Reporting and Data System (BI-RADS) categories and management recommendations for these lesions based on US findings with pathologic correlation. MATERIALS AND METHODS: From July 2001 to June 2007, 152 consecutive pathologically proven complex cystic lesions on US were retrospectively reviewed. All lesions at US were evaluated for size, lesion characteristics, margins, and presence of abnormal axillary nodes. US features of lesions were classified into four types, and positive predictive values (PPVs) were calculated for each type. Clinical, imaging, and histopathological findings were reviewed. RESULTS: Of the 152 lesions based on US appearance, 36 (24%) were classified as type I, 49 (32%) as type II, 28 (18%) as type III, and 39 (26%) as type IV. The PPVs for malignancy in each type were 14% for type I, 16% for type II, 14% for type III, and 41% for type IV. There was a significantly higher frequency of malignancy among lesions of type IV compared with the other three types (16/39=41% vs 5/36=14%, p=0.0089; 16/39=41% vs 8/49=16%, p=0.0098; and 16/39=41% vs 4/28=14%, p=0.018 [Chi-squared test]). Lesions with maximum diameter equal to or larger than 20mm, not circumscribed margins, or a mammographic finding of suspected malignancy had a high probability of malignancy (p<0.05 for each). CONCLUSION: US is useful in evaluating the complex cystic lesions and in clarifying the indication for biopsy of these lesions. The four types of US classifications used in our study establish accepted benchmarks for these breast abnormalities when stratified according to BI-RADS categories.
PURPOSE: To analyze the features of breast complex cystic lesions at ultrasonography (US) and to determine appropriate Breast Imaging Reporting and Data System (BI-RADS) categories and management recommendations for these lesions based on US findings with pathologic correlation. MATERIALS AND METHODS: From July 2001 to June 2007, 152 consecutive pathologically proven complex cystic lesions on US were retrospectively reviewed. All lesions at US were evaluated for size, lesion characteristics, margins, and presence of abnormal axillary nodes. US features of lesions were classified into four types, and positive predictive values (PPVs) were calculated for each type. Clinical, imaging, and histopathological findings were reviewed. RESULTS: Of the 152 lesions based on US appearance, 36 (24%) were classified as type I, 49 (32%) as type II, 28 (18%) as type III, and 39 (26%) as type IV. The PPVs for malignancy in each type were 14% for type I, 16% for type II, 14% for type III, and 41% for type IV. There was a significantly higher frequency of malignancy among lesions of type IV compared with the other three types (16/39=41% vs 5/36=14%, p=0.0089; 16/39=41% vs 8/49=16%, p=0.0098; and 16/39=41% vs 4/28=14%, p=0.018 [Chi-squared test]). Lesions with maximum diameter equal to or larger than 20mm, not circumscribed margins, or a mammographic finding of suspected malignancy had a high probability of malignancy (p<0.05 for each). CONCLUSION: US is useful in evaluating the complex cystic lesions and in clarifying the indication for biopsy of these lesions. The four types of US classifications used in our study establish accepted benchmarks for these breast abnormalities when stratified according to BI-RADS categories.