Assessment of extent of breast cancer: Comparison between digital breast tomosynthesis and full-field digital mammography

Abstract

To compare the accuracy of digital breast tomosynthesis (DBT) and full-field digital mammography (FFDM) in preoperative assessment of local extent of breast cancer. Lesion sizes of breast cancers on DBT and FFDM images were independently evaluated by breast radiologists. Each lesion was flagged as either mis-sized or not depending on whether the assessment of size at imaging was within 1 cm of the lesion size at surgery. Additional analyses were made by mammographic parenchymal density and by lesion size, using 2 cm as the boundary to separate the two subgroups. Statistical comparisons were performed using a repeated measures linear model on the percent mis-sized. P-values < 0.05 were considered statistically significant. The dataset included 173 malignant breast lesions (mean size 23.8 mm, 43% of lesions were ≤2 cm in size) in 169 patients, two-thirds of which had heterogeneously or extremely dense breasts. Overall, the percentage of lesions mis-sized at DBT was significantly lower than at FFDM (19% versus 29%, p = 0.003). There was significantly less mis-sizing at DBT in both heterogeneously dense breasts (11.1% difference between DBT and FFDM, p = 0.016) and extremely dense breasts (15.8% difference, p = 0.024). DBT also had significantly less mis-sizing than FFDM in the subgroup of lesions that were ≤2 cm in size (14.7% difference, p = 0.005). DBT was significantly superior to FFDM for the evaluation of lesion size overall, and specifically for small lesions and for lesions in dense breasts. The superiority of DBT versus FFDM increased with parenchymal density.