Concordance of Digital Breast Tomosynthesis (DBT) and Magnetic Resonance Imaging (MRI) Measured Tumor Size with the Pathological Gold Standard: A Retrospective Cohort Analysis by Molecular Subtype and Breast Density

Abstract

Background/Objectives: This study aimed to evaluate the concordance of preoperative digital breast tomosynthesis (DBT)- and magnetic resonance imaging (MRI)-measured tumor size with the pathological gold standard in invasive breast cancer and how this varies by molecular subtype and breast density (BI-RADS). Methods: This single-center, retrospective cohort study included 55 patients with unifocal invasive breast cancer who underwent preoperative DBT and MRI (1.5 T) followed by surgery within ≤60 days. Agreement with pathological size was assessed using Pearson’s correlation, Lin’s CCC, and Bland–Altman analysis; measurements within ±20% of pathological size were classified as concordant, with DBT/MRI rates compared using McNemar’s test. Subgroup analyses were performed by molecular subtype and breast density. Results: The mean pathological tumor size was 21.7 ± 10.7 mm. Pearson’s r was 0.784 for DBT and 0.785 for MRI (Lin’s CCCs were 0.775 and 0.783). Bland–Altman bias was −1.16 mm for DBT and +0.10 mm for MRI, with wider limits of agreement for MRI. Concordance (±20% threshold) was 54.5% for DBT and 45.5% for MRI, a non-significant difference (McNemar p = 0.38). Subtype- and density-stratified findings were directionally consistent but based on small subgroups and are hypothesis-generating only. Conclusions: In this cohort, DBT showed correlation and agreement with pathology comparable to MRI, without a statistically significant difference in concordance rate. Subgroup heterogeneity suggests DBT may hold value as a complementary sizing tool where MRI is unavailable, pending confirmation in larger, prospective, multicenter studies with independent reader validation.