Further signs in the evaluation of magnetic resonance mammography: a retrospective study

Abstract

Purpose: To increase accuracy and reliability of magnetic resonance breast imaging, a new evaluation method might be helpful. The recently suggested evaluation method (Fischer U, et al) resulted in a relevant number of equivocal cases (3 or 4 points). Additional morphologic and dynamic signs as an extension of this score were evaluated.

Method and materials: One hundred thirty-two histologically verified lesions were evaluated by 3 radiologists double-blinded using 2 evaluation methods: 1) method 1 (according to Fischer, et al): 2pt: initial signal increase >100%, washout, centripetal enhancement, 1pt: initial signal increase 50-100%, plateau phenomenon, centrifugal inhomogeneous enhancement, irregular borders, linear, stellar or dendritic structure; and 2) method 2 (according to Malich, et al): 3pt: hook sign (sign of pectoral invasion), 2pt: unifocal edema, blooming. 1pt: hypointensity in T2, lymph nodes >10 mm, skin thickening, adjacent vessels, a lesion's distorted inner architecture, disruption of the mamillary edge; -1pt: isointensity in T2, no edema, enhancing septations; -3pt: hyperintensity in T2, non enhancing septations. Method 1 judged a lesion to be malignant if 5 or more points were given and benign if 2 or less points were given, respectively. Method 2 (mean value of 3 radiologists) was tested in those cases in which a clear possible decision using method 1 was not sufficiently possible.

Results: Method 1 alone resulted in a negative predictive value of 96.8% and a positive predictive value of 90.8% (without carcinoma in situ), a sensitivity of 83.1%, a specificity of 58.8%, and revealed uncertain results (3 and 4 points) in 29 cases (out of 132; 22%). Adding the new scoring system in these 29 equivocal cases and an increase of 2 or more points by using method 2 is supposed to be a sign of malignancy; findings suggest a sensitivity of 90.9% and a specificity of 60% if an increase of maximum 1 is observed in benign lesions. In conclusion, our results show that Göttingen score alone has a sensitivity of 83.1%, a specificity of 58.8%; the second evaluation method reveals a sensitivity of 90.9% and a specificity of 60% in equivocal cases of Göttingen score. Göttingen score then reaches in all cases and second, adding the second evaluation method in equivocal cases, a sensitivity of 97% and a specificity of 76.5%.

Conclusion: The application of a second evaluation method in those cases remaining unclear in Göttingen score can lead to a decrease of uncertainty and a higher sensitivity and specificity of diagnosis in MR mammography. In this study, Göttingen score reaches a sensitivity of 83.1% and a specificity of 58.8%, increasing to a sensitivity of 97% and a specificity of 76.5% when being extended by a second evaluation method in unclear cases.