Feng-Yang Zheng1, Qing Lu2, Bei-Jian Huang3, Han-Sheng Xia4, Li-Xia Yan5, Xi Wang6, Wei Yuan7, Wen-Ping Wang8. 1. Department of Ultrasound, Zhongshan Hospital, Fudan University, Shanghai 200032, China; Shanghai Institute of Medical Imaging, Shanghai 200032, China. Electronic address: fyzheng16@fudan.edu.cn. 2. Department of Ultrasound, Zhongshan Hospital, Fudan University, Shanghai 200032, China. Electronic address: lu.qing@zs-hospital.sh.cn. 3. Department of Ultrasound, Zhongshan Hospital, Fudan University, Shanghai 200032, China; Shanghai Institute of Medical Imaging, Shanghai 200032, China. Electronic address: huang.beijian@zs-hospital.sh.cn. 4. Department of Ultrasound, Zhongshan Hospital, Fudan University, Shanghai 200032, China. Electronic address: zs12036@126.com. 5. Department of Ultrasound, Zhongshan Hospital, Fudan University, Shanghai 200032, China. Electronic address: dndyanlixia@163.com. 6. Department of Ultrasound, Zhongshan Hospital, Fudan University, Shanghai 200032, China; Shanghai Institute of Medical Imaging, Shanghai 200032, China. Electronic address: wang.xi@zs-hospital.sh.cn. 7. Department of Pathology, Zhongshan Hospital, Fudan University, Shanghai 200032, China. Electronic address: yuan.wei@zs-hospital.sh.cn. 8. Department of Ultrasound, Zhongshan Hospital, Fudan University, Shanghai 200032, China; Shanghai Institute of Medical Imaging, Shanghai 200032, China. Electronic address: wang.wenping@zs-hospital.sh.cn.
Abstract
OBJECTIVES: To investigate the correlation between the imaging features obtained by an automated breast volume scanner (ABVS) and molecular subtypes of breast cancer. METHODS: We examined 303 malignant breast tumours by ABVS for specific imaging features and by immunohistochemical analysis to determine the molecular subtype. ABVS imaging features, including retraction phenomenon, shape, margins, echogenicity, post-acoustic features, echogenic halo, and calcifications were analysed by univariate and multivariate logistic regression analyses to determine the significant predictive factors of the molecular subtypes. RESULTS: By univariate logistic regression analysis, the predictive factors of the Luminal-A subtype (n=128) were retraction phenomenon (odds ratio [OR]=10.188), post-acoustic shadowing (OR=5.112), and echogenic halo (OR=3.263, P<0.001). The predictive factors of the Human-epidermal-growth-factor-receptor-2-amplified subtype (n=39) were calcifications (OR=6.210), absence of retraction phenomenon (OR=4.375), non-mass lesions (OR=4.286, P<0.001), absence of echogenic halo (OR=3.851, P=0.035), and post-acoustic enhancement (OR=3.641, P=0.008). The predictors for the Triple-Negative subtype (n=47) were absence of retraction phenomenon (OR=5.884), post-acoustic enhancement (OR=5.255, P<0.001), absence of echogenic halo (OR=4.138, P=0.002), and absence of calcifications (OR=3.363, P=0.001). Predictors for the Luminal-B subtype (n=89) had a relatively lower association (OR≤2.328). By multivariate logistic regression analysis, retraction phenomenon was the strongest independent predictor for the Luminal-A subtype (OR=9.063, P<0.001) when present and for the Triple-Negative subtype (OR=4.875, P<0.001) when absent. CONCLUSIONS: ABVS imaging features, especially retraction phenomenon, have a strong correlation with the molecular subtypes, expanding the scope of ultrasound in identifying breast cancer subtypes with confidence.
OBJECTIVES: To investigate the correlation between the imaging features obtained by an automated breast volume scanner (ABVS) and molecular subtypes of breast cancer. METHODS: We examined 303 malignant breast tumours by ABVS for specific imaging features and by immunohistochemical analysis to determine the molecular subtype. ABVS imaging features, including retraction phenomenon, shape, margins, echogenicity, post-acoustic features, echogenic halo, and calcifications were analysed by univariate and multivariate logistic regression analyses to determine the significant predictive factors of the molecular subtypes. RESULTS: By univariate logistic regression analysis, the predictive factors of the Luminal-A subtype (n=128) were retraction phenomenon (odds ratio [OR]=10.188), post-acoustic shadowing (OR=5.112), and echogenic halo (OR=3.263, P<0.001). The predictive factors of the Human-epidermal-growth-factor-receptor-2-amplified subtype (n=39) were calcifications (OR=6.210), absence of retraction phenomenon (OR=4.375), non-mass lesions (OR=4.286, P<0.001), absence of echogenic halo (OR=3.851, P=0.035), and post-acoustic enhancement (OR=3.641, P=0.008). The predictors for the Triple-Negative subtype (n=47) were absence of retraction phenomenon (OR=5.884), post-acoustic enhancement (OR=5.255, P<0.001), absence of echogenic halo (OR=4.138, P=0.002), and absence of calcifications (OR=3.363, P=0.001). Predictors for the Luminal-B subtype (n=89) had a relatively lower association (OR≤2.328). By multivariate logistic regression analysis, retraction phenomenon was the strongest independent predictor for the Luminal-A subtype (OR=9.063, P<0.001) when present and for the Triple-Negative subtype (OR=4.875, P<0.001) when absent. CONCLUSIONS: ABVS imaging features, especially retraction phenomenon, have a strong correlation with the molecular subtypes, expanding the scope of ultrasound in identifying breast cancer subtypes with confidence.