Diagnostic performance of quantitative shear-wave elastography for reclassifying BI-RADS category 4 breast lesions: a retrospective single-center study

Objective To evaluate four quantitative shear-wave elastography (SWE) parameters (Emax, Emean, Emin, elasticity ratio) for differentiating benign and malignant BI-RADS 4 breast lesions and reclassifying lesions that may potentially avoid immediate biopsy. Methods This single-center retrospective analysis enrolled 454 pathologically confirmed BI-RADS 4 lesions (one per patient) examined with conventional ultrasound and SWE (2022–2025). BI-RADS subcategories (4A/4B/4C) were determined prior to SWE acquisition, and Emax, Emean, Emin, Eratio were recorded. A multivariable logistic regression model was developed. Receiver-operating characteristic (ROC) analysis and paired DeLong test evaluated discrimination, with bootstrap resampling for internal validation. Results Of the 454 lesions, 277 (61.01%) were benign, and 177 (38.99%) were malignant. All four SWE parameters were higher in malignant than in benign lesions (all p < 0.001) and remained independent predictors of malignancy [Emax: odds ratio (OR), 1.106 (95% confidence interval (CI), 1.073–1.140); Emean: OR, 1.115 (95% CI, 1.080–1.152); Emin: OR, 1.101 (95% CI, 1.053–1.151); and Eratio: OR, 2.438 (95% CI, 1.875–3.170), per 1-kPa/1-unit increment; all p < 0.001]. The combined model achieved an AUC of 0.948 (95% CI, 0.926–0.970), with a sensitivity of 84.75% and a specificity of 93.86%, and outperformed each individual parameter (all p < 0.001); the bootstrap optimism-corrected AUC was 0.945. When the model probability was used as a standalone post-hoc exploratory rule (cut-off p = 0.32), 255 lesions (56.17%) were classified as low-risk; this group nevertheless contained 17 malignant lesions (apparent negative predictive value, 93.33%; missed-cancer rate, 9.60%; bootstrap-corrected, 92.96% and 10.14%, respectively). The rule therefore did not demonstrate that immediate biopsy could be safely avoided. Conclusion Quantitative SWE parameters differentiated benign and malignant BI-RADS 4 lesions, with the four-parameter combined model achieving better discrimination than single parameters. The reclassification strategy may potentially reduce benign biopsies if validated prospectively. Given the absence of external validation, these preliminary findings require prospective multicenter confirmation before clinical application.