1665. Can BI-RADS 3 Assessments Be Used for New Mammographic Findings? Results from the National Mammography Database (NMD)
Authors * Denotes Presenting Author
  1. Cindy Lee *; NYU Langone Health
  2. Jeremy Berg; University of Pittsburgh School of Medicine
  3. Wendie Berg; University of Pittsburgh School of Medicine
The BI-RADS Atlas reserves use of probably benign, BI-RADS 3 (BR3), assessments for specific imaging findings known to have = 2% likelihood of malignancy (1). Its use was validated for baseline mammographic findings without prior comparison, but not for new findings on subsequent mammograms; we sought to explore outcomes of such within the NMD. The NMD is the largest national registry of mammograms in the United States (2, 3).

Materials and Methods:
This retrospective cohort HIPAA-compliant study included all women recalled from screening mammography followed by BR3 assessment at diagnostic evaluation from January 2009 to March 2018, from 471 NMD facilities. We included only the first BI-RADS 3 occurrences for women = age 25 who underwent biopsy or = 2-year imaging follow-up. Women who had breast cancer or were biopsied at their first BI-RADS 3 visit were excluded. Outcomes of biopsies and follow-up were analyzed. Cancer yield, biopsy rate and PPV3 at 2-years were calculated for women with and without prior mammographic comparison. Linear regression with weighted age binning was performed to assess for statistical significance in measures between two groups.

143,374 screening and diagnostic mammograms performed in 43,628 women (median age 55 years; range 25-90) with BR3 assessment were analyzed. Overall, biopsy rate was 11.2% (95% CI 10.9-11.5), CY was 1.86% (1.73-1.98) and PPV3 was 16.6% (15.5-17.6) 57.7% of women (25,160) with BR3 findings had prior mammographic comparison with presumed new findings. Among women with prior comparison, BR3 findings had significantly higher cancer yield (2.24 vs 1.34%, p <0.0001), biopsy rate (11.64 vs 10.64%, p = 0.0011) and PPV3 (19.22 vs 12.57%, p<0.0001), compared to those without prior (Figure). The difference in cancer yield was 0.51% (0.16-0.86) between women with and without prior comparison at the same age (p =0.0095).

Nearly 60% of women given BR3 assessment on recall from screening had prior mammographic comparison with presumed new findings. The BR3 findings in women with prior comparison had significantly higher cancer yield than those without prior at the same age, with the cancer yield exceeding 2% over age 55. Cancer yield of probably benign findings significantly increases with age (p<0.0001) and availability of prior comparison (p=0.0095), both of which are important for management of these findings.