초음파 BI-RADS 범주 4와5 병변: 범주 4의 세분화와 각 범주의 양성 예측도

Abstract

Purpose: To retrospectively analyze the sonographic features and positive predictive values (PPVs) of the lesions classified as categories 4a, 4b, 4c and 5 by using Breast Imaging Reporting and Data System (BIRADS) ultrasonography(US) lexicon and to assess statistically significant BI-RADS US descriptors to predict malignancy in category 4 and 5 lesions. Materials and Methods: Between January 2004 and September 2005, two breast radiologists reviewed 1150 pathologically proven BI-RADS US category 4 and 5 lesions of 1010 patients retrospectively based on BIRADS US lexicon and reassigned one of the final assessment categories to each lesion. PPVs for category 4a, 4b, 4c and 5 were evaluated. Multiple logistic regression was used to assess combination of significant BI-RADS US descriptors. Results: PPVs for subcategory 4a, 4b and 4c and category 5 were as follows: category 4a, 80 (12%) of 646; category 4b, 76 (55%) of 138; category 4c, 144(87%) of 166; category 5, 196 (98%) of 200. The patient age, mammographic features, shape, margin, echogenicity, lesion boundary, calcifications, posterior echo pattern were statistically different between each subcategory (p < 0.05). The combination of statistically significant sonographic parameters that helped to discriminate malignancy from benignity of the lesions classified as category 4 and 5 were shape, margin, echogenicity, lesion boundary and calcifications. Lobular shape, spiculated margins, complex echo, echogenic halo, presence of microcalcifications were the features that showed relatively higher estimated odds ratio in each US parameter. Conclusion: Category 4 lesions were appropriately subacategorized by experienced breast radiologists based on BI-RADS US lexicon. Lobular shape, spiculated margins, complex echo, echogenic halo, and presence of microcalcifications showed relatively higher estimated odds ratio in each US parameter.