Benign vs. malignant differentiation of well-circumscribed masses detected on Digital Breast Tomosynthesis

Abstract

Purpose To elucidate the mammographic and clinicopathological features of well-circumscribed masses (WCMs) detected on digital breast tomosynthesis (DBT). Materials and methods Findings of a total of 555 consecutive WCM lesions in 266 women detected on DBT between August 2016 and March 2021 were retrospectively reviewed. The clinicopathological and digital mammography (DM) features were correlated with the DBT features of the WCM lesions. Results The average age of the subjects was 55 years. Most of the WCMs were benign (531/555, 95.8%), with cancer accounting for only 4.3% (24/555) of all the cases. On DM, 41.6% (231/555) were classified as C1 (negative), and 3.5% (8/231) of these were eventually diagnosed as cancer. Univariate analysis identified WCM density on DBT, the maximum lesion diameter, and the maximum/minimum diameter ratio as significant factors for differentiating between benign and malignant lesions. Multivariable analysis identified only lesion density as a significant independent factor, with a sensitivity/specificity/accuracy/positive predictive value (PPV)/negative predictive value (NPV) of 45.8%/78.5%/77.1%/8.8%/97%. When the highest priority was placed on the sensitivity, a cutoff value for the largest diameter of 7.9 mm was found to be associated with a sensitivity/ specificity/ accuracy/PPV/NPV of 100%/38.6%/41.3%/PPV 6.9%/NPV 100%. Conclusion About 95% of WCMs detected on DBT were benign lesions. Our results suggest the false-positive rate may be lowered by excluding small lesions (‹7.9mm), without also missing any cases of breast cancer.