Positive predictive value of BI-RADS MR imaging

Area of Science:

• Radiology
• Oncology
• Medical Imaging

Background:

• Breast magnetic resonance (MR) imaging is crucial for detecting breast cancer.
• The Breast Imaging and Reporting Data Systems (BI-RADS) provides a standardized lexicon for reporting.
• Accurate assessment of BI-RADS categories and lesion features is vital for predicting malignancy.

Purpose of the Study:

• To evaluate the positive predictive values (PPVs) of BI-RADS assessment categories in breast MR imaging.
• To identify specific BI-RADS MR imaging lesion features that are most predictive of breast cancer.

Main Methods:

• Prospective, multicenter study evaluating contralateral breast MR imaging in women with recent breast cancer diagnoses.
• Assessed BI-RADS categories, morphologic descriptors (foci, masses, non-masslike enhancement - NMLE), and kinetic features for malignancy prediction.
• Estimated PPVs and used logistic regression to analyze the predictive ability of lesion combinations.

Main Results:

• Of 969 participants, most had BI-RADS 1/2 (71.3%), with fewer in categories 3 (10.9%) and 4/5 (10.0%).
• Thirty-one cancers were detected; overall PPV for BI-RADS 4/5 was 0.278 (category 4 PPV 0.205, category 5 PPV 0.714).
• For masses, irregular shape/margins and spiculated margins were highly predictive. For NMLEs, ductal, clumped, and reticular/dendritic enhancement were common in malignancies. Kinetic features were less predictive than morphology.

Conclusions:

• Standardized BI-RADS terminology allows quantification of malignancy likelihood in breast MR imaging.
• BI-RADS assessment categories and morphologic descriptors for masses and NMLE are valuable for estimating cancer probability.
• Careful evaluation of lesion features enhances the diagnostic accuracy of breast MR imaging.