Multiparametric breast MRI to problem-solve mammographically detected suspicious calcifications

Abstract

Objective: To evaluate the performance of multiparametric breast MRI to problem-solve mammographicallydetected suspicious calcifications. Materials and methods: Participants with mammographically-detected suspicious calcifications were prospectively enrolled between August 2017 to May 2023. Pre-biopsy multiparametric MRI (standard and high-temporal resolution dynamic contrast enhanced [DCE]-MRI acquisitions and diffusion-weighted imaging [DWI]) was performed. The associations of MRI features with outcomes - (1) any malignancy and (2) invasive or high-grade ductal carcinoma in situ [DCIS] only - were analyzed using univariable logistic regression. Multivariable models, sequentially incorporating clinical/mammographic, qualitative, and quantitative features, were developed using penalized logistic regression with the least absolute shrinkage and selection operator. Area under the receiver operating characteristic curves (AUC) were estimated via cross-validation and compared between models using bootstrap methods. Results: 81 women (mean age, 55 years +/- 10 [standard deviation]) with 86 calcifications were included; 29 % (25/86) were malignant. Malignancy rates for non-enhancing mammographic Breast Imaging Reporting and Data System (BI-RADS) category 4a and 4b calcifications were 3.8 % (1/26) and 8.7 % (2/23), respectively, with no invasive cancer or high-grade DCIS. Mammographic BI-RADS category, MRI BI-RADS category, visibility on DWI, peak percent enhancement, functional tumor volume, and Ktrans values were associated with both outcomes (all p < 0.05). Multivariable models, including qualitative DCE-MRI assessments, showed higher AUCs (malignancy: 0.71-0.76; invasive cancer or high-grade DCIS: 0.78-0.91) than when including only clinical and mammographic features (malignancy: 0.57; invasive cancer or high-grade DCIS: 0.61, all p < 0.05). Further incorporation of DWI or quantitative MRI features did not improve AUCs (all Delta AUC <= 0). Conclusion: Breast DCE-MRI aids in evaluating mammographic BI-RADS category 4a/4b calcifications without biopsy. DWI or quantitative MRI features may not further improve diagnostic performance.