Comparison of 3.0 T magnetic resonance imaging and X-ray mammography in the measurement of ductal carcinoma in situ: A comparison with histopathology

Purpose: To determine if MRI data obtained at 3.0 T can more accurately report the size of DCIS as compared to radiographic mammography, as a whole cohort and when subdivided by lesion characteristics. Methods: Thirty-nine participants underwent X-ray mammography and MRI prior to breast surgery for DCIS. Longest diameter (LD) measurements were recorded for each imaging modality and compared to histopathological LD via a logarithmic transformed Bland-Altman agreement plot methodology resulting in dimensionless mean difference and 95% limits of agreement (LoA). Results: Data from 39 patients with a median age of 55 years (range 38-78 years) underwent analysis. Mastectomy was undertaken in 21 cases, while breast conserving surgery was performed in 18 subjects. Histopathological analysis revealed one low grade, nine intermediate grade, and 21 high grade lesions. The mean +/- standard deviation LD measurements for histopathology, X-ray mammography and MRI were 50.6 +/- 34.2 mm, 30.7 +/- 23.1 mm and 49.6 +/- 26.8 mm respectively. Bland-Altman agreement plot analysis for the whole cohort revealed not only a smaller logarithmic mean difference between MRI and histopathology (0.086), but also narrower 95% LoA (-0.941 to 1.113) compared with X-ray mammography and histopathology (mean difference 0.658, 95% LoA 3.503 to 2.187). When the level of agreement was assessed between clinically relevant subgroups additional significant differences were noted based on grade, hormonal receptor status, invasion, necrosis, mircocalcifications and growth pattern. Conclusion: MRI provides a more accurate estimation of DCIS size than X-ray mammography. MRI's superior ability was not only noted in general, but also for clinical relevant subdivisions such as grade and the presence or absence of necrosis. (C) 2015 Elsevier Ireland Ltd. All rights reserved.