BI-RADS and MRI Useful in Predicting Breast Cancer

May 15, 2012

Breast Imaging and Reporting Data Systems terminology is useful in predicting malignancy in breast lesions that are detected by MRI.

Breast Imaging and Reporting Data Systems (BI-RADS) terminology is useful in predicting malignancy in breast lesions that are detected by MRI, according to a study published online in the journal Radiology.

BI-RADS, published by the American College of Radiology in collaboration with other healthcare organizations, was developed to standardize the lexicon used by radiologists when interpreting breast imaging reports. “The BI-RADS lexicon for breast MRI provides descriptors and assessment categories that can be used to help predict the likelihood of cancer,” said Mary C. Mahoney, MD, director of breast imaging at the University of Cincinnati Medical Center in Ohio.
Researchers evaluated the performance of BI-RADS for MRI and identification of breast imaging features most predictive of cancer among 969 women who had recently been diagnosed with cancer in one breast.

Researchers found that a BI-RADS score of five was assigned to 14 women, 11 of whom underwent follow-up imaging. Cancer was identified in 10, for a positive predictive value of 71 percent. Eight-three women were assigned a score of four, considered “suspicious abnormality, biopsy should be considered.” Of these, 67 women had follow-up imaging that found 17 cancers, a positive predictive value of 20 percent.

Irregular shape, spiculated margins, or marked enhancement were lesion features most predictive of cancer. For non-three-dimensional lesions, features most predictive of cancer were located in a milk duct or clumped enhancement.

Mahoney noted that there is still wide variability in how MRI for breast cancer is preformed, and a lack of protocol standardization makes comparing results difficult. Recommendations from this study may be incorporated into future editions of BI-RADS, she said, to ensure MRIs are more easily transferred between institutions.