Not all false positive diagnoses are equal: On the prognostic implications of false-positive diagnoses made in breast MRI versus in mammography / digital tomosynthesis screening

Abstract

Background: Breast magnetic resonance imaging (MRI) has been reported to frequently result in false-positive diagnoses, limiting its positive predictive value (PPV). However, for PPV calculation, all nonmalignant tissue changes are equally considered false-positive, although the respective prognostic importance, and thus patient management implications, of different pathologies may well differ. We investigated the pathology of false-positive diagnoses made by MRI compared with radiographic (digital mammography/tomosynthesis [DM/DBT]) screening. Methods: We conducted an institutional review board-approved prospective analysis of 710 consecutive asymptomatic women at average risk for breast cancer who underwent vacuum biopsy with or without surgical biopsy for screen-detected DM/DBT (n = 344) or MRI (n = 366) findings. We compared the frequency of false-positive biopsies (given by PPV3), as well as the types of nonmalignant tissue changes that caused the respective false-positive biopsies. In an order of increasing relative risk of subsequent breast cancer, pathologies of false-positive biopsies were categorized as nonproliferative, simple proliferative, complex proliferative, or atypical proliferative (including lobular carcinoma in situ/lobular intraepithelial neoplasia). The Mann-Whitney U test was used to compare distributions. Results: Histology yielded nonmalignant tissue in 202 of 366 biopsies done for positive MRI studies and 195 of 344 biopsies for positive DM/DBT studies, respectively, yielding a similar PPV3 percentages of 44.8% (164 of 202) and 43.3% (149 of 202) for both methods. However, the distribution of tissue types that caused false-positive diagnoses differed significantly (p < 0.0001). On the basis of MRI, high-risk atypical proliferative changes (40.1%; 81 of 202) were most common, followed by complex proliferative changes (23.8%; 48 of 202). In DM/DBT, low-risk, nonproliferative changes were the dominant reason for false-positive diagnoses (49.7%; 97 of 195), followed by simple proliferative changes (25.2%; 51 of 195). Low-risk nonproliferative changes resulted in false-positive diagnoses based on MRI as infrequently as did high-risk atypical proliferative changes based on DM/DBT (18.8% [38 of 202] vs. 18.0% [35 of 195]). The likelihood of a false-positive diagnosis including atypias was twice as high in women undergoing biopsy for MRI findings (81 of 202; 40%) as for those with DM/DBT findings (35 of 195; 18%). Conclusions: The prognostic importance, and thus the clinical implications, of false-positive diagnoses made on the basis of breast MRI vs. radiographic screening differed significantly, with a reversed prevalence of high- and low-risk lesions. This should be taken into account when discussing the rate of false-positive diagnoses (i.e., PPV levels of MRI vs. radiographic screening). Current benchmarks that rate the utility of breast cancer screening programs (i.e., cancer detection rates and PPVs) do not reflect these substantial biological differences and the different prognostic implications.