Analysis of sonographic features in the differentiation of fibroadenoma and invasive ductal carcinoma.

OBJECTIVE The purpose of this study was to determine the predictive power of sonographic tumor descriptors in the differentiation of fibroadenoma from invasive ductal carcinoma of the breast. SUBJECTS AND METHODS Three hundred thirty-six tumors (142 fibroadenomas and 194 invasive ductal carcinomas) of the breast diagnosed using sonography were prospectively recorded with respect to the shape, contour, echo texture, echogenicity, sound transmission, and surrounding tissue of the tumors. Evaluation included odds and odds ratios of single sonographic features as well as sensitivity, specificity, and positive and negative predictive values of combinations of features. Tumor descriptors were also evaluated using multiple logistic regression analysis after adjustment for age and clinical examination. RESULTS Irregular shape and contour, extensive hypoechogenicity, shadowing, echogenic halo, and distortion of surrounding tissue were the findings with the highest predictive value of malignancy. A thin echogenic pseudocapsule was the most important sonographic finding predictive of the benign nature of a solid mass. Echo texture was of little value in the differentiation of breast tumors. Age and clinical examination remained important predictors in a clinically referred patient population because a palpable mass in an elderly patient is most likely a carcinoma. We saw considerable overlap of most sonographic features in both benign and malignant tumors. However, using strict sonographic criteria and a combination of descriptors, we found a negative predictive value of 100% in palpable and 96% in impalpable tumors. CONCLUSION A combination of tumor descriptors gave negative predictive values approaching 100%, allowing downgrading of solid breast masses with a high degree of confidence. Extensive sonographic features analysis in patients with indeterminate clinical and mammographic findings has the potential for downgrading a tumor and possibly obviating the need for excision in a subgroup of patients. Further investigations may provide standardization of sonographic descriptor analysis and establishment of the combination of the most predictive features that would be useful in daily practice.