Malignant Outcomes of Architectural Distortion on Tomosynthesis: A Systematic Review and Meta-Analysis.

Abstract

<b>BACKGROUND.</b> The literature has reported varying rates of malignancy for architectural distortion (AD) on digital breast tomosynthesis (DBT). <b>OBJECTIVE.</b> The purpose of this study was to evaluate the PPV for malignancy of AD on DBT without a known cause and to assess the presence of an ultrasound (US) correlate for malignant AD through systematic review and meta-analysis of the literature. <b>EVIDENCE ACQUISITION.</b> This meta-analysis included all studies published in Em-base, MEDLINE, and Evidence-Based Medicine Reviews databases through July 15, 2020, that assessed the rate of malignancy in patients with AD on DBT without a known cause that was deemed BI-RADS category 4 or 5. Rates of benign or high-risk lesions and the presence of a US correlate for malignant AD were assessed. Core needle biopsy or surgical pathology was used as the reference standard for lesion diagnosis. The pooled PPV and 95% CI were estimated using a random-effects model. <b>EVIDENCE SYNTHESIS.</b> Thirteen retrospective, observational studies were included, yielding 857 ADs seen on DBT. Of the 857 ADs, 339 were breast malignancies, yielding a pooled PPV for malignancy of 34.6% (95% CI, 24.5-46.3%). The pooled PPV for invasive malignancy was 34% (95% CI, 25-45%) and for ductal carcinoma in situ was 5% (95% CI, 4-7%). Of the 857 ADs, 235 (27.4%) were benign lesions, 282 (32.9%) were high-risk lesions, and 1 (0.1%) was a nonbreast metastatic lesion. From the studies that assessed for US correlates, 217 of 277 malignant ADs (78.3%) had a US correlate. <b>CONCLUSION.</b> The pooled PPV for malignancy of AD on DBT without a known cause is high at 34.6%, warranting tissue sampling. <b>CLINICAL IMPACT.</b> A needle biopsy should be performed for ADs on DBT without a known cause. Because most malignant distortions have a corresponding finding on US, a US examination should be performed to look for a correlate, but the absence of a correlate does not obviate a biopsy.