Utility of Breast MRI for Further Evaluation of Equivocal Findings on Digital Breast Tomosynthesis.

Abstract

The purpose of this study is to evaluate the utility of MRI as a problem-solving tool for equivocal findings on diagnostic digital mammography (DM) and digital breast tomosynthesis (DBT). Breast MRI examinations performed from March 2011 to November 2014 were retrospectively reviewed to identify those examinations that were performed to further assess equivocal findings on combined DM and DBT (DM/DBT) examinations. All patients underwent diagnostic ultrasound in conjunction with their DM/DBT examination. Imaging reports were retrospectively reviewed for BI-RADS findings and assessments of diagnostic DM/DBT and diagnostic MRI examinations. A review of the electronic medical records provided information on demographic data, cancer diagnoses, and pathologic findings. Differences in the positive predictive values and negative predictive values of DM/DBT and MRI were compared using a generalized estimating equation for correlated binary data. Of 5330 MRI examinations performed during the study, 67 (1%) were performed for evaluation of an equivocal finding, including 27 asymmetries (40%), 16 focal asymmetries (24%), five masses (7%), and 19 architectural distortion (28%). MRI correlates were identified in 22 of 67 examinations (33%). Biopsies yielded a cancer diagnosis for five of 67 patients (7%). For MRI, the positive predictive value and negative predictive value were 19% and 98%, respectively, whereas for DM/DBT they were 6% and 90%, respectively (p = 0.009 and p = 0.059, respectively). The frequency of recommendations for breast MRI to evaluate equivocal findings decreased exponentially in the 3 years after DBT implementation. As clinical implementation of DBT becomes increasingly widespread, breast radiologists need an algorithm for addressing the small number of inconclusive findings that remain equivocal despite thorough DM/DBT and ultrasound examinations. Breast MRI is a useful adjunctive tool for these selected cases.