Individualized-thresholding Shear Wave Elastography combined with clinical factors improves specificity in discriminating breast masses

Abstract

Purpose: To investigate the diagnostic role of new metrics, defined as individualized-thresholding of Shear Wave Elastography (SWE) parameters, in association with clinical factors (such as age, mammographic density, lesion size and depth) and the BI-RADS features in differentiating benign from malignant breast lesions.

Methods: Of 644 consecutive patients (median age, 55 years), prospectively referred for evaluation, 659 ultrasound detected breast lesions underwent SWE measurements. Multivariable logistic regression analysis was used to estimate the probability of malignancy. The area under the curve (AUC), optimal cutoff value, and the corresponding sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) were determined.

Results: 265 of 659 (40.2%) masses were malignant. Using two E_mean cutoffs, 69.6 kPa for large superficial lesions (size >10 mm, depth ≤5 mm) and 39.2 kPa for the rest, the overall specificity, sensitivity, PPV and NPV were 92.6%, 86.8%, 88.8% and 91.3%, respectively. Combining multiple factors, including E_mean with two cutoffs, age and BI-RADS, the new ROC curve based on the malignancy probability calculation showed the highest AUC (0.954, 95% CI: 0.938-0.969). Using the optimal probability threshold of 0.514, the corresponding specificity, sensitivity, PPV and NPV were 92.9%, 89.1%, 89.4% and 92.7%, respectively.

Conclusions: The false-positive rate can be significantly reduced when applying two E_mean cutoffs based on lesion size and depth. Moreover, the combination of age, E_mean with two cutoffs and BI-RADS can further reduce the false negatives and false positives. Overall, this multifactorial analysis improves the specificity of ultrasound while maintaining a high sensitivity.

Keywords: Breast cancer; Breast masses; Elastography; Shear wave elastography; Ultrasound.