Architectural distortion in the era of digital breast tomosynthesis: outcomes and implications for management.

Purpose: Digital breast tomosynthesis (DBT) has been shown to increase conspicuity of some mammographic findings, particularly architectural distortion (AD). The purpose of this retrospective study was to determine the positive predictive value of AD on diagnostic DBT, and evaluate associations between AD characteristics and histopathologic outcomes.

Methods: This IRB-approved, HIPAA-compliant study included diagnostic DBT exams performed between 1/2014 and 12/2015 that demonstrated AD. Imaging characteristics of AD, corresponding ultrasound and MRI exams, and pathology results were reviewed. Fisher's exact tests and a two-tailed t-test were performed.

Results: Seventy-seven cases of AD were visualized in 68 patients (ages 36-78 years, mean 54 years). Core biopsy of 74 cases of AD yielded malignant pathology in 26/74 (35%) tissue samples. Among 48/74 (65%) non-malignant cases of AD, 25 demonstrated a high-risk lesion, including radial scar in 20, with no upgrades to malignancy among 20 high-risk lesions that underwent surgical excision. Among 23 non high-risk causes of benign AD, stromal fibrosis was most common, present in 12 biopsy specimens. Associations between imaging characteristics of AD on DBT and histopathologic outcome were not statistically significant (one-view visualization 3/13 malignant, 10/13 benign, P = 0.52; DBT-only finding 1/11 malignant, 10/11 benign, P = 0.09). Of 48 cases with an ultrasound correlate, 22/48 (46%) were malignant versus 4/26 (15%) cases without an ultrasound correlate (P = 0.01).

Conclusions: AD on diagnostic DBT was malignant in over one-third of cases. The presence of an ultrasound correlate was associated with malignancy.