Contrast-enhanced spectral mammography vs. mammography and MRI – clinical performance in a multi-reader evaluation
- 901 Downloads
To compare the diagnostic performance of contrast-enhanced spectral mammography (CESM) to digital mammography (MG) and magnetic resonance imaging (MRI) in a prospective two-centre, multi-reader study.
One hundred seventy-eight women (mean age 53 years) with invasive breast cancer and/or DCIS were included after ethics board approval. MG, CESM and CESM + MG were evaluated by three blinded radiologists based on amended ACR BI-RADS criteria. MRI was assessed by another group of three readers. Receiver-operating characteristic (ROC) curves were compared. Size measurements for the 70 lesions detected by all readers in each modality were correlated with pathology.
Reading results for 604 lesions were available (273 malignant, 4 high-risk, 327 benign). The area under the ROC curve was significantly larger for CESM alone (0.84) and CESM + MG (0.83) compared to MG (0.76) (largest advantage in dense breasts) while it was not significantly different from MRI (0.85). Pearson correlation coefficients for size comparison were 0.61 for MG, 0.69 for CESM, 0.70 for CESM + MG and 0.79 for MRI.
This study showed that CESM, alone and in combination with MG, is as accurate as MRI but is superior to MG for lesion detection. Patients with dense breasts benefitted most from CESM with the smallest additional dose compared to MG.
• CESM has comparable diagnostic performance (ROC-AUC) to MRI for breast cancer diagnostics.
• CESM in combination with MG does not improve diagnostic performance.
• CESM has lower sensitivity but higher specificity than MRI.
• Sensitivity differences are more pronounced in dense and not significant in non-dense breasts.
• CESM and MRI are significantly superior to MG, particularly in dense breasts.
KeywordsBreast neoplasms Mammography Contrast media Magnetic resonance imaging Breast cancer
We are grateful to Helena Wiebe, Jana Förster, Michaela Krohn, MD, Jasmin-Maya Singh, MD, Nikola Bangemann, MD, Angela Reles, MD, Christiane Richter-Ehrenstein, MD, and Klaus-Jürgen Winzer, MD, for their contributions to patient recruitment and inclusion. We thank Luc Katz, MSc, Marc Dewey, PhD, and Lauren Mamer, MS, for reviewing the manuscript, Serge Muller, PhD, for scientific assistance and David Caumartin, MSc, for supporting the setup of this study.
The scientific guarantor of this publication is Prof. Ulrich Bick. The authors of this manuscript declare relationships with the following companies: GE Healthcare, Guerbet Healthcare, Siemens Healthcare and Bayer Healthcare; one author is a stockholder in all of the medical companies.
This study has received funding by a research grant from GE Healthcare and partly by a research grant from Guerbet, Roissy CdG, Cedex, France. The investigators had exclusive control of all data, manuscript drafting and submission of this study. One of the authors has significant statistical expertise. Institutional Review Board approval was obtained. Written informed consent was obtained from all subjects (patients) in this study.
Some study subjects or cohorts have been previously reported in the initial results of a single-site, single-reader clinical report evaluation of parts of this patient population (80 patients) have been previously published. Only the primary (index) lesion of each case was assessed .
A second publication analysed the clinical performance and size estimation of MG, CESM and the combination of MG and CESM with three readers in 107 index cancers only . Both were single-centre studies. Both studies could only focus on sensitivity as benign lesions were not included. Methodology: prospective, diagnostic or prognostic study, multicentre study.
- 2.Emaus MJ, Bakker MF, Peeters PH et al (2015) MR Imaging as an additional screening modality for the detection of breast cancer in women aged 50-75 years with extremely dense breasts: the DENSE trial study design. Radiology 0:141827Google Scholar
- 14.Agency IAE (2002) International action plan for the radiological protection of patients. GOV/2002/36-GC(46)/12 GOV/2002/36-GC(46)/12:1-9Google Scholar
- 16.Committee A (2014) ACR practice guideline for the performance of magnetic resonance imaging (MRI) of the breast.Google Scholar
- 17.Radiology ACo (2003) Breast Imaging Reporting and Data System (BI-RADS). VA: American College of Radiology 4 editionGoogle Scholar
- 20.Team RC (2014) R: a language and environment for statistical computing. R Foundation for Statistical Computing, ViennaGoogle Scholar
- 21.Dan Carr pbNL-K, Martin Maechler and contains copies of lattice function written by Deepayan, Sarkar (2014) hexbin: Hexagonal Binning Routines. R package version 1.27.0Google Scholar