E2020. Contrast-Enhanced Mammography: A Closer Look at BI-RADS 3 Lesions and Malignancy Rate
  1. Quinton Keigley; Mayo Clinic; UW Health (University of Wisconsin Hospitals and Clinics)
  2. Ilyse Nelson; Mayo Clinic
  3. Victor Pizzitola; Mayo Clinic
  4. Roxanne Lorans; Mayo Clinic
  5. Marina Giurescu; Mayo Clinic
  6. William Eversman; Mayo Clinic
  7. Bhavika Patel; Mayo Clinic
This study aims to evaluate the characteristics and outcomes of Breast Imaging Reporting and Data System (BI-RADS) category 3 lesions detected on contrast-enhanced mammography (CEM). Performance statistics were compared to prior research of “probably benign” breast lesions observed on magnetic resonance imaging (MRI).

Materials and Methods:
We conducted a single-facility, institutional review board exempt, retrospective study on CEM examinations with a final assessment of BI-RADS 3. Studies were interpreted by one of five fellowship-trained breast imagers during routine care. In our clinic, 1,204 CEMs were performed between April 2014 and October 2019. A total of 118 (9.8%) of these studies had been designated BI-RADS 3. Clinical notes, imaging findings, and pathology outcomes were evaluated for biopsy-proven malignancy or two years of negative imaging follow-up to establish a true negative. Eighty-six (86/118) patients met these criteria.

Included patients had a mean age of 56 years (range 32 to 83). A total of four patients (4/86; 4.7%) in the group were diagnosed with malignant lesions: one low-grade ductal carcinoma in situ and three low-grade invasive ductal carcinomas. The neoplasms presented as either an enlarging non-mass enhancement (1/4; 25%), new architectural distortion (1/4; 25%), or new masses (2/4; 50%). All tumors were surgically removed with a mean size of 13 mm (7 to 20 mm). The average time to diagnosis from the initial CEM BI-RADS 3 assessment was 13 months. All four patients had a prior history of breast cancer. The BI-RADS 3 assessment rate for included CEMs was 9.8% with a calculated cancer yield of 4.7%. Similar frequencies of “probably benign” impressions, ranging from 1 to 18%, were found in prior studies examining MRI. Furthermore, the cancer yield for BI-RADS 3 lesions found on MRI was comparable at 0.6 to 10%.

BI-RADS 3 performance metrics and outcomes with CEM are similar to reported ranges in MRI literature, supporting the notion that it may be useful for patients when MRI is contraindicated or access is limited. Continued experience with CEM over time will lead to increased confidence amongst readers, potentially refining the use of BI-RADS 3 assessments and improve cancer detection.