Ultrasound correlation modifies malignancy risk of architectural distortion on digital mammography and tomosynthesis:...

Architectural distortion (AD) on mammography is worked up with targeted ultrasound (US), yet how much a US correlate changes malignancy risk by digital mammography (DM) versus digital breast tomosynthesis (DBT) is unclear.

Quantify malignancy risk of AD by modality and US correlation and compare DBT with DM.

PRISMA/MOOSE review (PROSPERO CRD4201088536). Five databases (2010-2025) were searched for studies reporting histopathology of AD stratified by modality and US findings. Malignancy risks were pooled with mixed-effects logistic regression; risk differences (RD) with 95% CIs compared US-positive versus US-negative lesions and DBT versus DM.

Fourteen studies (1707 AD lesions) were included. For DM-detected AD, malignancy risk was 50.7% (95% CI 9.7-90.8) with a US correlate and 26.5% (3.8-76.9) without (RD +24.1 points; 95% CI +9.8 to +38.5; p = 0.0010). For DBT-detected AD, risk was 56.8% (39.2-72.7) with a correlate and 20.3% (11.5-33.5) without (RD +36.4 points; 95% CI +27.3 to +45.6; p < 0.0001). Among US-positive lesions, DBT differed from DM (difference 16.7 points; p = 0.0017); among US-negative lesions, modality differences were not significant (difference 8.7 points; p = 0.097).

A targeted US correlate strongly increases malignancy risk for AD on both DM and DBT; absence of a correlate does not imply low risk by itself.