E2136. Can Audit of BIRADS-3 Lesions While on Surveillance Serve as a Quality Metric and Useful Benchmark of Practice Performance?
  1. Gopal Vijayaraghavan; Univ of Massachusetts Med School
  2. Prithwijit Roychowdhury; Univ of Massachusetts Med School
  3. Imani Williams; Univ of Massachusetts Med School
  4. Efaza Siddiqui; Univ of Massachusetts Med School
  5. John Roubil; Univ of Massachusetts Med School
BI-RADS-3 or a probably benign finding is an accepted assignment in the ACR BI-RADS lexicon and carries a less than 2% likelihood of malignancy. There is an established algorithm of surveillance for lesions assigned BI-RADS-3. While the imaging criteria for BI-RADS-3 assignment is well defined, given individual radiologist's and practice variations these guidelines are not always strictly followed. We therefore retrospectively audited the BI-RADS-3 in our practice to determine if the radiology reports met established criteria and to determine our cancer yield on surveillance.

Materials and Methods:
Approximately 30,000 screening exams are performed annually at our facility, a mixture of academic and community practice. All exams are interpreted by fellowship trained breast imagers. We do not assign a BI-RADS-3 on screens. Over a 4-year period from Jan 2014- Dec 2017, there were 11,718 BI-RADS-0 identified in our electronic data base, and of these 1360 were assigned a BI-RADS-3 on their subsequent diagnostic study. The BI-RADS assignment were confined to mammograms and ultrasounds. We have so far analyzed 323/1360 BI-RADS -3 on surveillance and share our initial results. Analysis of the remaining cases is ongoing.

57/323 cases were following a baseline screening exam and the remaining 266/323 had prior mammograms for comparison. The age distribution varied from 40 to 94 years with a mean age of 57.4 years. The follow up diagnostic study from a screening assigned BI-RADS-0 was a mammogram only in 110, ultrasound only in 19 and both modalities in 194 cases. We broadly classified the morphology distribution as under asymmetry/architectural distortion 164, calcifications 121, mass 13, others 25. The findings were on the right breast in 158, left breast in 149, and on both sides in 16 cases. The breast density breakdown of these 323 cases were as under category A-18, B-174, C-120, D-11. At the 6-month follow-up there were 306/323 cases, at 12 months 290/306 cases and at 24 months 264/290 cases. During the surveillance period 25 biopsies were performed; 16 under US guidance and 9 under stereotactic guidance. 6/25 biopsies were positive for cancer resulting in a 1.8% (6/323) upgrade rate. However, on follow-up at the 3rd year, 3 additional cancers were documented. 2/57 baseline screens were upgrade to BI-RADS-4 on follow-up, but both biopsies were benign.

Based on the initial analysis our cancer yield from BI-RADS-3 surveillance was 6/323 or 1.8% (<2%). However, if we add the additional 3 cancers detected in the 3rd year, then the upgrade rate is 9/323 or 2.7% (>2%). Like other established metrics in practice evaluation of breast imaging facilities, we feel that BI-RADS-3 data (benchmarking as a percentage of total screens, as a percentage of BI-RADS-0 and cancer upgrade rates while on surveillance), has potential in evaluation of quality and performance improvement. Likewise, larger studies are needed to evaluate if the <2% cancer probability in BI-RADS-3 or increasing the period of surveillance merits revisiting.