Correlation Between 18F-FDG PET/CT Findings and BI-RADS Assessment Using Ultrasound in the Evaluation of Breast Lesions: A Multicenter Study.

ABSTRACT

RATIONALE AND OBJECTIVES: To analyze the correlation between ultrasound breast imaging reporting and data system (BI-RADS) category and fluorodeoxyglucose [18F] (18F-FDG) positron emission tomography/computed tomography (PET/CT) findings and their value in breast lesion diagnosis. MATERIALS AND METHODS: Cases involving hypermetabolic lesions identified by 18F-FDG PET/CT and ultrasound were retrospectively analyzed. The correlation between the maximum standardized uptake values (SUVmax) of the lesions and the BI-RADS grades was calculated. Histologic diagnosis or evidence at the end of a 2-year follow-up as the standard of truth were analyzed to determine the sensitivity, specificity, positive predictive values (PPV), and negative predictive values (NPV) of the diagnostic methods. Area under the curve (AUC) of BI-RADS, SUVmax, and BI-RADS/SUVmax combined were obtained using receiver-operating characteristic curve (ROC) analysis. RESULTS: Of 206 cases, 92 were benign and 114 were malignant. The difference between the SUVmax and the BI-RADS grades was statistically significant (p < 0.001). The critical value of the optimal SUVmax was 2.325, and the accuracy, sensitivity, specificity, PPV, and NPV were 84.5%, 91.2%, 76.1%, 82.5%, and 87.5%, respectively. For diagnosis using BI-RADS, these values were 85.9%, 98.2%, 70.7%, 80.6%, and 97.0%, respectively. ROC analysis of 206 breast lesions for distinguishing benign from malignant lesions yielded AUCs of 0.948, 0.896, and 0.977 for BI-RADS, SUVmax, and BI-RADS/SUVmax combined, respectively. The critical value of the optimal SUVmax in grade 3 and 4 lesions (as determined using BI-RADS) was 2.705, and the accuracy, sensitivity, specificity, PPV, and NPV were 82.6%, 77.8%, 85.7%, 77.8%, and 85.7%, respectively. For diagnosis using BI-RADS in cases with grade 3 and 4 lesions, these values were 68.5%, 94.4%, 51.8%, 55.7%, and 93.5%, respectively. In ROC analysis for distinguishing benign from malignant for BI-RADS grade 3-4 lesions, the AUC of BI-RADS, SUVmax, and BI-RADS/SUVmax combined were 0.731, 0.859, and 0.882, respectively. CONCLUSION: Both 18F-FDG PET/CT and ultrasound-dependent BI-RADS grading are effective for diagnosing breast lesions. However, in cases of BI-RADS grades 3 and 4, 18F-FDG PET/CT has better specificity and may be useful for further differential diagnosis.