Objective: The objective of our study was to test dynamic MRI in evaluating mammographically detected suspicious microcalcifications.
Materials and methods: One hundred twelve patients with mammographically detected microcalcifications with BI-RADS category 5 (n = 78) or 4 (n = 34) lesions were studied at 17 centers a using 3D gradient-echo dynamic coronal technique (< or = 3 mm thickness) and 0.1 mmol/kg of gadoteridol. A pathologic sample was obtained in all cases. Agreement between the major diameter measured on mammography, MRI, or both and the major diameter measured at pathologic examination was calculated in 62 cases.
Results: Of the 112 lesions, pathologic examination revealed 37 benign lesions, 33 ductal carcinoma in situ (DCIS), and 42 invasive carcinomas. The specificity of MRI for benign lesions was 68%. Considering the subgroups of calcifications alone and calcifications associated with masses, the specificity values became 79% and 33%, respectively. The sensitivity of MRI for DCIS was 79%. Analysis of the two subgroups showed sensitivity values of 68% for calcifications alone and of 1% for calcifications associated with masses. The sensitivity for invasive carcinomas was 93%. Analysis of the two subgroups showed sensitivity values to be 92% for calcifications alone and 94% for calcifications associated with masses. Considering the overall results, the sensitivity of MRI was 87%; specificity, 68%; positive predictive value, 84%; negative predictive value, 71%; and accuracy, 80%. Considering the subgroups of calcifications alone and calcifications associated with masses, the sensitivity values became 80% and 97%; the positive predictive values, 86% and 82%; the negative predictive values, 71% and 75% (95% confidence interval [CI], 0.19-0.99); and the accuracy values, 80% and 82% (95% CI, 0.66-0.92), respectively. An odds ratio (OR) of 13.54 (95% CI, 5.20-35.28) showed a raised risk of malignant breast tumor in subjects with positive MR examination of mammographically detected suspicious clusters of microcalcifications. The statistical analysis on each subgroup showed an OR of 15.07 (95% CI, 4.73-48.08) for calcifications alone and an OR of 14.00 (95% CI, 1.23-158.84) for calcifications associated with masses. Any significant improvement in the predictive ability of dynamic MRI depending on the extent of calcifications on mammography was not proved. Considering the 62 cases of proved malignancy with measured maximal diameter at pathologic examination, both mammography and MR examination seem to overestimate tumor extent.
Conclusion: The not-perfect sensitivity of MRI (87%), when applying our interpretation criteria and imaging sequences, is a crucial point that prevents us from clinical use of MRI in the diagnosis of mammographically detected microcalcifications.