Reporting expectations for BI-RADS category 5 classifications are feasible, achievable
When using the American College of Radiology’s Breast Imaging-Reporting and Data System (BI-RADS) to classify a breast lesion, radiologists are expected to only use the category 5 assessment when the likelihood of malignancy, or positive predictive value (PPV), is at least 95 percent. Is that a feasible expectation? Researchers examined data from thousands of consecutive breast imaging examinations at a single institution to find out, sharing their findings in a new study published in Academic Radiology.
The authors explored more than 22,000 consecutive examinations performed between January 2010 and September 2015. Overall, they found, 1.1 percent of those exams were classified as BI-RADS category 5. Malignancy was confirmed in 233 of the 239 tumors, giving the institution a probability of breast caner of 97.5 percent.
In addition, of the 220 exams that involved both ultrasound and mammography, more than 98 percent had four or more suspicious descriptors.
“This study is important because it shows the clinical decision made by BI-RADS to set a high PPV cut point for BI-RADS 5 is feasible and attainable by using four or more suspicious descriptors for each lesion classified as BI-RADS 5,” wrote author Melissa Min-Szu Yao, MD, department of radiology at Wan Fang Hospital in Taiwan, and colleagues. “Although this is a single institution study at an academic practice, BI-RADS descriptors are widely accepted and used throughout the United States and in many other countries, hence these results should be generalizable and transferrable to any practice properly using BI-RADS.
The team added that they did have sex false-positives at surgical excision, but “imaging review of these six cases revealed solid masses with multiple suspicious imaging descriptors that were indistinguishable from those associated with breast cancer.”
Michael has more than 18 years of experience as a professional writer and editor. He has written at length about cardiology, radiology, artificial intelligence and other key healthcare topics.