Bianca Carpentier MD, Presenter: Nothing to Disclose

Deborah Anne Cunningham MD, Abstract Co-Author: Nothing to Disclose

Justine Lavoye, Abstract Co-Author: Nothing to Disclose

Carlos Gonzalez MD, Abstract Co-Author: Nothing to Disclose

Marie Elaine Quinn MD, Abstract Co-Author: Nothing to Disclose

B. Nicolas Bloch MD, Abstract Co-Author: Nothing to Disclose

To assess the influence of inter-reader variability of radiologists interpreting mammograms during the course of a two-year BI-RADS 3 follow up protocol.

In this IRB approved retrospective study, medical records from February 1, 2002 to May 31, 2009 were reviewed, using a dedicated mammography database and the electronic medical record database. All mammograms performed within this period, classified as BI-RADS 3, were included and evaluated for change in classification, time and reason of change, and interpreting radiologist during the 2-year course of follow up. Mammographic reports of the BI-RADS 3 lesions, which were upgraded to BI-RADS 4 or 5 on follow up exams, were reviewed by a radiologist.

121,862 mammograms were performed at our institution between February 2002 and May 2009, with 12 interpreting radiologists. 8,403/121,862 mammograms (7%) were classified as BI-RADS 3. During the course of a two-year follow up, 32/8,403 (0.4%) exams were changed to BI-RADS 4 or 5. All of the 32 (100%) upgraded BI-RADS 4 or 5 lesions showed cancer on subsequent biopsies. However, 11/32 (34%) did not show any objective change in size or morphology, yet were upgraded to BI-RADS 4 or 5 by a subsequent and different radiologist, based on individual interpretation of the stable finding. None of the 11 (0%) mammographically stable initial BI-RADS 3 lesions were upgraded, if the same radiologist interpreted the subsequent mammogram. 9/11 (82%) were changed at the first 6-month follow up exam; 0/11 (0%) at the second 6-month (12-month) follow up; 1/11 (9%) at the third 6-month (18-month) and 1/11 (9%) at the fourth 6 months (24-month).

Inter-reader variability plays an important role in the assessment of mammographic BI-RADS 3 lesions (34% of upgrades to BI-RADS 4 or 5, were reader-dependent, without objective mammographic interval change). The majority of reader-dependent upgrades (82%) occurred at the first 6-month follow up. We propose, in order to reduce adverse reader-variability based misclassification and delay of cancer diagnosis, that a second reader reviews BI-RADS 3 lesions at the time of the initial diagnostic work-up (first BI-RADS 3 interpretation).

To reduce adverse inter-reader variability and misclassification of BI-RADS 3 (probably benign) lesions, we propose, that a second reader reviews BI-RADS 3 lesions at time of initial classification.

Carpentier, B, Cunningham, D, Lavoye, J, Gonzalez, C, Quinn, M, Bloch, B, The Role of Inter-Reader Variability in the Assessment of Mammographic Probably Benign Breast (BI-RADS 3) Lesions.  Radiological Society of North America 2012 Scientific Assembly and Annual Meeting, November 25 - November 30, 2012 ,Chicago IL. http://archive.rsna.org/2012/12035696.html