Breast ultrasound elastography--results of 193 breast lesions in a prospective study with histopathologic correlation

Purpose: To evaluate the diagnostic performance of ultrasound elastography in breast masses.

Material and methods: 193 lesions (129 benign, 64 malignant) were analyzed with the EUB 8500 Logos-ultrasonic-unit (Hitachi Medical, Japan) and a linear-array-transducer of 7.5-13-MHz. Standard of reference was cytology (FNAfine needle aspiration) or histology (core biopsy). The elastic-score was classified according to a 6-point colour-scale (Ueno classification; 1-3 = benign, 4-5 = malignant). Conventional B-mode ultrasound (US) findings were classified according to the BI-RADS classification. Statistical analysis included sensitivity, specificity, ROC-analysis and kappa-values for intra-/interobserver reliability.

Results: The mean score for elasticity was 4.1 ± 0.9 for malignant lesions, and 2.1 ± 1.0 for benign lesions (p < 0.001). With a best cut-off point between elasticity scores 3 and 4, sensitivity was 96.9%, and specificity 76%. Setting a best cut-off point for conventional US between BI-RADS 4 and 5, sensitivity was 57.8%, and specificity 96.1%. Elastography provided higher sensitivity and lower specificity than conventional US, but two lesions with elasticity score 1 were false negative, whereas no lesion scored BI-RADS 1-3 were false negative. ROC-curve was 0.884 for elastography, and 0.820 for conventional US (p < 0.001). Weighted kappa-values for intra-/interobserver reliability were 0.784/0.634 for BI-RADS classification, and 0.720/0.561 for elasticity scores.

Conclusion: In our study setting, elastography does not have the potential to replace conventional B-mode US for the detection of breast cancer, but may complement conventional US to improve the diagnostic performance.