A preliminary study to propose a diagnostic algorithm for PET/CT-detected incidental breast lesions: application of BI-RADS lexicon for US in combination with SUVmax

Objectives To develop a diagnostic algorithm for positron emission tomography (PET)-detected incidental breast lesions using both breast imaging reporting and data system (BI-RADS) and maximum standardized uptake value (SUVmax) criteria. Methods Fifty-six PET-detected incidental breast lesions from 51 patients, which were subsequently investigated by breast ultrasound within 1 month of the PET study, constituted the study cohort and they were finally verified by tissue diagnosis or a 2-year follow-up. Based on the maximum specificity with sensitivity > 60.0% and maximum sensitivity with specificity > 60.0%, two SUVmax cutoff values were calculated at 2 and 3.7. BI-RADS >= 4 was considered as highly suspicious for malignancy. The diagnostic accuracies were estimated for SUVmax levels above or below the cutoff points combined with the BI-RADS suspicion level. Results Overall, 46 benign and 10 malignant lesions were studied. The diagnostic characteristics of SUVmax >= 2, SUVmax >= 3.7, and BI-RADS >= 4 were 80.0%, 60.0%, and 80.0% for sensitivity, 73.9%, 95.7%, and 92.7% for specificity, and 75.0%, 89.3%, and 90.2% for accuracy, respectively. When the SUVmax threshold was set at 2, combined with BI-RADS suspicion level, the sensitivity, specificity, and accuracy were 100.0%, 69.6%, and 75.0%, respectively. The results for SUVmax threshold set at 3.7 combined with BI-RADS were 90.0%, 91.3%, and 91.1% for the sensitivity, specificity, and accuracy, respectively. A diagnostic algorithm was accordingly generated. Conclusion The need for biopsy should be justified in low BI-RADS lesions presenting with high SUVmax at 3.7 or higher. The biopsy of patients with high B-IRADS and low SUVmax could be preserved.