RESUMO
RATIONALE AND OBJECTIVES: Magnetic resonance imaging (MRI) characteristics of Breast Imaging Reporting and Data System (BI-RADS) 4 and 5 lesions were evaluated to identify characteristics that may improve the positive predictive value (PPV) of a biopsy. MATERIALS AND METHODS: MRI BI-RADS 4 or 5 breast findings in 101 women who underwent biopsy leading to a diagnosis of cancer within 1 year (PPV3; n = 115 cases) were reviewed. Patient history, lesion morphology, enhancement pattern and kinetics, and T2 characteristics were examined. RESULTS: The PPV3 of all BI-RADS 4 and 5 breast lesions seen on MRI was 22.6% (26 of 115). Excluding lesions with second-look imaging correlates decreased the PPV3 to 11.8% (6/51). Of the MRI-guided biopsies, 20.9% (24 of 115) yielded a high-risk lesion, altering surgical management. MRI lesion type did not significantly affect the PPV3: the PPV3 was 26.3% (15 of 57) for masses, 21.4% (9 of 42) for non mass enhancement (NME) and 12.5% (2 of 16) for suspicious foci. The PPV3 for lesions found on diagnostic MRI in women with newly diagnosed cancer was 30.8% (20 of 65) which was statistically significantly greater compared to a PPV3 of 11.9% (5 of 42) for lesions identified on high-risk screening MRI. CONCLUSIONS: Suspicious MRI findings identified on a second-look examination are more likely malignant than those seen only on MRI. Suspicious MRI findings discovered in patients with concurrent malignancy have a greater PPV3 than those detected on high-risk screening MRIs. However, the type of MRI finding (mass vs. NME vs. foci) does not significantly affect the PPV3 and should not be used as a discriminator for determining biopsy threshold.
