RONG Xiaocui, KANG Yihe, YIN Fenghua, WU Zhonglin, WU Yongchao, LI Yazhou
Objective To retrospectively investigate the value of breast MRI in evaluation of breast imaging reporting and data system (BI-RADS) 4-5 mammographic microcalcifications. Methods The clinicopathologic and radiological data of 118 patients (mean age, 46.6±9.3 years; range 27-66 years) who were confirmed by biopsy or surgical pathology were retrospectively analyzed. All mammography and MR images were reviewed with BI-RADS. The apparent diffusion coefficients (ADCs) of the lesions were measured on the ADC maps of diffusion-weighted imaging, and the tumor sizes were measured on mammography, MRI, and pathology, respectively. The comparison between the two groups was performed with chi-square test or t-test. The tumor sizes measured on mammography, MRI and pathology were compared using Wilcoxon signed ranks test. Using histopathological diagnosis as the golden standard, the sensitivity, specificity, accuracy, positive predictive value (PPV), and negative predictive value (NPV) were calculated on the basis of BI-RADS category. The diagnostic efficacy was evaluated with ROC curve. Z test was used to compare the areas under the ROC curves (AUC) of mammography and MRI in the diagnosis of microcalcification. The association and consistency of lesion sizes acquired by mammography, MRI, and pathology were assessed by Spearman rank correlation and Bland-Altman analysis. Results Of the 119 lesions, 46 were benign and 73 were malignant. There were significant differences in the morphology, distribution of microcalcifications, and ADC between benign and malignant lesions (P<0.05). The sensitivity, specificity, accuracy, PPV and NPV of mammography and MRI were 84.93%, 32.61%, 64.71%, 66.67% and 57.59%, and 95.89%, 71.74%, 86.55%, 84.34%, and 91.67%, respectively. The AUC of MRI assessment was larger than that of mammography (P<0.05). The tumor sizes measured on mammography and MRI were larger than that on pathology (P<0.05). The sizes of tumors measured on mammography and MRI showed moderate correlation with the measurements obtained on pathology (rs=0.602, rs=0.603, P<0.05). Bland-Altman analysis showed that compared with pathological results, mammography and MRI overestimated the tumor sizes, and the consistencies were not good enough. Conclusions In the evaluation of BI-RADS 4-5 mammographic microcalcifications, breast MR imaging provides additional information with high sensitivity and NPV. Both MRI and mammography overestimate tumor sizes.