The value of prebiopsy FDG-PET/CT in discriminating malignant from benign vertebral bone lesions in a predominantly oncologic population

PURPOSE: To determine the value of prebiopsy (18)F-fluoro-2-deoxy-D-glucose positron emission tomography (FDG-PET)/computed tomography (CT) in discriminating malignant from benign vertebral bone lesions. MATERIALS AND METHODS: This retrospective study included 53 patients with 55 vertebral bone lesions that underwent FDG-PET/CT before CT-guided biopsy. Pathologic examination of the biopsy sample and a minimum follow-up of 1 year were used as reference standard. RESULTS: Sensitivity, specificity, positive predictive value, and negative predictive value of visual FDG-PET analysis (with lesion FDG uptake higher than liver FDG uptake as threshold for malignancy) in discriminating malignant from benign vertebral bone lesions were 91.3% (42/46), 22.2% (2/9), 85.7% (42/49), and 33.3% (2/6), respectively. The semiquantitative FDG-PET metrics SUV(max) and SUV(peak) achieved areas under the receiver operating characteristics curve of 0.630 and 0.671, respectively. Malignant lesions demonstrated bone lysis more frequently than benign lesions (60.9% (28/46) vs. 22.2% (2/9)), and this difference was nearly significant (P = 0.064). All other clinical and conventional imaging characteristics (including patient age, gender, previous diagnosis of malignancy, bone pain, weight loss, any CT abnormality, sclerosis, cortical destruction, bone marrow replacement, associated extraosseous soft tissue mass, and accompanying vertebral height loss, multiple bone lesions on FDG-PET/CT, and suspicious extraosseous lesions on FDG-PET/CT) were not significantly different (P = 0.143 to 1.000). CONCLUSION: FDG-PET/CT may steer the diagnosis (particularly thanks to a relatively high PPV and value of semiquantitative measurements), but cannot always classify vertebral bone lesions as malignant or benign with sufficient certainty. In these cases, biopsy and/or follow-up remain necessary to establish a final diagnosis.