Comparison of contrast-enhanced cone-beam breast CT, MRI, and mammography for breast cancer characterization.

Objectives: To compare the consistency of contrast-enhanced cone-beam breast CT (CE-CBBCT), MRI and mammography regarding characterization of breast cancer.

Methods: In this retrospective study, patients with breast cancer who underwent preoperative CE-CBBCT, MRI, and mammography between January 2017 and July 2022 were enrolled. Three experienced radiologists independently interpreted the characteristics of breast lesions on each imaging mode referring to BI-RADS, with a 4-week wash-out period. One of the three radiologists reviewed the CE-CBBCT images 4 weeks after the initial evaluation. Cross-modality consistency was calculated by Cohen's Kappa based on majority report. Inter-and intra-reader agreement were assessed using Fleiss and Cohen's Kappa, respectively. The association between imaging factors and consistency levels was analyzed using chi-square and Mann-Whitney U test.

Results: A total of 214 malignant lesions identified in 207 patients were enrolled. CE-CBBCT showed almost perfect agreement with MRI on lesion type identification (Kappa = 0.865, 95% CI: 0.802-0.928), but fair agreement with mammography (Kappa = 0.287, 95% CI: 0.205-0.369). CE-CBBCT showed substantial agreement on characterization with MRI for both mass (Kappa = 0.752-0.824) and non-mass enhancement (NME) (Kappa = 0.702-0.729), and non-contrast-enhanced CBBCT (NCE-CBBCT) showed substantial agreement with mammography for calcification (Kappa = 0.717-0.777). Inter- (Kappa = 0.611-0.738) and intra-reader (Kappa = 0.757-0.887) agreement were substantial on CE-CBBCT interpretation. There was no statistically significant difference in imaging factors between different consistency levels (all p > 0.05).

Conclusions: CE-CBBCT showed high consistency on mass and NME characterization with MRI, and on calcification with mammography, indicating that CE-CBBCT could combine morphology, hemodynamic and calcification features, and the corresponding descriptors have the feasibility to describe CE-CBBCT characteristics of breast cancer.

Key points: Question Contrast-enhanced cone-beam breast CT (CE-CBBCT) is widely used in the diagnosis and assessment of breast cancer, but there is no standardized lexicon for image interpretation. Findings CE-CBBCT showed high consistency and comparable reproducibility with MRI and mammography for characterizing breast cancer lesions. Clinical relevance The findings prove that CE-CBBCT could combine morphology, hemodynamic, and calcification features and provide support for the feasibility of applying the BI-RADS descriptors of MRI and mammography to interpret contrast-enhanced cone-beam breast CT images in clinic.