Contrast-enhanced digital mammography and magnetic resonance imaging: reproducibility compared to pathologic anatomy

Abstract

Purpose:

To compare the reproducibility between contrast-enhanced digital mammography (CEDM) and magnetic resonance imaging (MRI) with the postsurgical pathologic examination. In addition, the applicability of the Breast Imaging–Reporting and Data System (BI-RADS) lexicon of MRI to CEDM was evaluated for mass lesions.

Methods:

A total of 62 patients with a histologically proven diagnosis of breast cancer were included in this study, for a total of 67 lesions. Fifty-nine patients underwent both methods. The reproducibility between MRI vs CEDM and the reference standard (postoperative pathology) was assessed by considering the lesion and breast size as pivotal variables. Reproducibility was evaluated by computing the concordance correlation coefficient (CCC). Bland-Altman plots were used to depict the observed pattern of agreement as well as to estimate the associated bias. Furthermore, the pattern of agreement between the investigated methods with regard to the breast lesion characterization (i.e. mass/nonmass; shape; margins; internal enhanced characteristics) was assessed by computing the Cohen kappa and its 95% confidence interval (CI).

Results:

The reproducibility between MRI and the reference standard and between CEDM and the reference standard showed substantial agreement, with a CCC value of 0.956 (95% CI, 0.931–0.972) and 0.950 (95% CI, 0.920–0.969), respectively. By looking at the Bland-Altman analysis, bias values of 2.344 and 1.875 mm were observed for MRI and CEDM vs reference evaluation, respectively. The agreement between MRI and CEDM is substantial with a CCC value of 0.969 (95% CI, 0.949–0.981). The Bland-Altman analysis showed bias values of −0.469 mm when comparing CEDM vs MRI. Following the Landis and Koch classification criteria, moderate agreement was observed between the two methods in describing BI-RADS descriptors of mass lesions.

Conclusion:

CEDM is able to measure and describe tumor masses comparably to MRI and can be used for surgical planning.