Ultrasound variables and their prognostic value in a population of 1103 patients with 272 breast cancers.

In a prospective two-centre study targeted US was performed as an adjunct to mammography in a population of 1103 patients with 272 breast cancers, 517 benign lesions and no abnormalities in 314 patients. The purpose of the study was to analyse the distribution of the different US variables among the breast lesions and to determine the prognostic value of these variables with respect to the diagnosis of malignancy. The following variables were analysed: border; contour; orientation; structure; echogenicity; sound transmission; and size. These variables were correlated with the definitive diagnosis and univariate analysis was performed. A statistically significant association with breast cancer (p < 0.001) was present for irregular border; ill-defined contour; indeterminate or vertical orientation; homogeneous, complex or heterogeneous structure; hypoechogenicity; and unchanged or decreased sound transmission. Multivariate analysis showed a high independent prognostic value for malignancy for irregular border, followed by ill-defined contour and unchanged or decreased sound transmission. However, their discriminative power was not absolute: in 38 cancers none of the malignant US variables were present, and in 11 benign lesions all variables were present. There were hardly any benign-looking lesions that proved to be malignant.