Accuracy of whole-body fluorine-18-FDG PET for the detection of recurrent or metastatic breast carcinoma.

This study assessed the diagnostic accuracy of whole-body PET on a patient and lesion basis using 18F-fluorodeoxyglucose (FDG) for the detection of tumor foci in patients with suspected recurrent or metastatic lesions of breast carcinoma.Whole-body FDG-PET imaging was performed on 57 patients with a previous history of breast carcinoma who were referred for a clinical suspicion of disease recurrence. Whole-body PET images were scored from 1 (definitely negative) to 5 (definitely positive) by three independent observers, and discrepancies were resolved by a fourth observer. Patients were clinically followed for up to 24 mo to assess the accuracy of PET diagnosis by biopsy, follow-up imaging and other diagnostic tests.PET scans showed that there were 41 sites indicating recurrent or metastatic disease in 29 patients. There were 38 sites in 28 patients that showed no evidence for malignant disease. On a patient basis, with scores 4 or 5 considered to be positive, sensitivity and specificity were 93% and 79%, respectively. The corresponding positive and negative predictive values were 82% and 92%. On a lesion basis, with scores 4 or 5 considered to be positive, the sensitivity was 85% and specificity 79%. The area index in receiver operating characteristic analysis was 0.91 for patient-based analysis and 0.88 for lesion-based analysis. To determine the cause for false-negative and false-positive findings more precisely, false-negative lesions with scores of 3 or lower and false-positive lesions with scores of 4 or higher were analyzed. Bone metastases had a significantly larger proportion of false-negative lesions than other nonosseous malignant sites (p < 0.05). False-positive lesions were due to muscle uptake (n = 5), inflammation (n = 4), blood pool activity in the great vessels (n = 2), bowel uptake (n = 1) and unknown causes (n = 6).The whole-body FDG-PET scan is a useful diagnostic test for detecting recurrent or metastatic lesions of breast carcinoma. However, the sensitivity for metastases to bone appears to be lower than that to other organs. Specificity may be improved by more strict attention to patient preparation and better recognition of physiologic skeletal muscle or artifactual uptakes.