Malignancy risk stratification prediction of BI-RADS 4B calcifications based on contrast-enhanced mammographic features: a multicenter study

• Published in: Breast cancer research and treatment 2024-11
• Main Authors: Long, Rong, Luo, Yao, Cao, Min, Cao, Kun, Li, Xiao-Ting, Mao, Ning, Yang, Guang, Sun, Ying-Shi
• Format: Article
• Language: English
• ISSN: 0167-6806; 1573-7217; 1573-7217
• DOI: 10.1007/s10549-024-07546-w

This study aims to investigate the factors influencing the malignant risk of BI-RADS 4B calcification-only lesions detected on Contrast-Enhanced Mammography (CEM) and to develop a predictive model for stratifying malignant risk. A retrospective analysis was conducted on 131 calcification-only lesions of BI-RADS 4B identified on low-energy (LE) images of CEM from 125 females between March 2017 and April 2023 at three institutions. The patients were grouped as training (95 lesions) and external validation sets (36 lesions). On LE images, morphological features of the calcifications, including morphology, distribution and size, were evaluated. On recombined images, the presence and types of enhancement were assessed as qualitative variables, and the grey values from lesion areas and background were measured as quantitative variables. Multivariate logistic regression analysis was used to construct a predictive model. The discrimination of the model was assessed by the receiver operating characteristic (ROC) curve and confirmed by the external validation set. Of the 131 lesions, 43 were malignant. The morphology, distribution, the presence and types of enhancement and the grey values of calcifications showed significant differences between benign and malignant lesions. The nomogram was developed based on morphology and the presence of enhancement, with areas under the ROC curve of 0.859 (95% confidence interval [CI]: 0.769, 0.949) and 0.856 (95% CI: 0.729, 0.983) in the training and external validation sets, respectively. On CEM, the presence of enhancement and morphology were identified as independent predictors of malignant calcifications of BI-RADS 4B. The predictive model demonstrated favorable performance.