Agreement Among Breast Cancer Screening Modalities in Breast Density Assessment and Cancer Risk Prediction

Breast density is increasingly recognized as a vital risk factor that affects early breast cancer detection. Therefore, this study was aimed at evaluating the agreement between different breast density measurements across multiple imaging modalities and identifying the best breast cancer predictor among these methods.

Data for women over 30 years old who underwent mammography, synthetic mammography of digital breast tomosynthesis (SM-DBT), ultrasound (US), and magnetic resonance imaging (MRI) were collected. Breast density was assessed by two breast radiologists using the American College of Radiology (ACR) categories and the visual analog scale (VAS) for percentage density (%PD). The agreement was calculated using the kappa coefficient (k) and Spearman correlation coefficient (ρ). Logistic regression odds ratios (ORs) were used to assess the best predictor of breast cancer based on breast density.

Among 77 women (mean age 47.34 years), 25 had breast cancer. Categorical breast density assessments showed the highest agreement between mammography and SM-DBT (k = 0.535) and moderate agreement between mammography and MRI (k = 0.452). VAS analysis revealed moderate positive correlations between mammography and SM-DBT (ρ = 0.49), mammography and MRI (ρ = 0.56), and SM-DBT and MRI (ρ = 0.56), p < 0.05. Ultrasound showed the lowest correlation with all breast imaging modalities. Breast cancer risk prediction based on breast density showed significant associations for mammography (OR = 3.09) and MRI (OR = 4.16), p < 0.05.

The results suggest notable variability in radiologists' breast density assessment across different imaging modalities. MRI showed a greater ability to identify dense breast tissue and demonstrated potential value in breast cancer risk prediction, although these findings should be interpreted cautiously given the limited sample size. Establishing standardized approaches to breast density assessment remains important to improve the accuracy of breast cancer screening and risk prediction, and further research with larger cohorts is warranted.