The aim of this study is to compare FDG and FDG-labeled leukocyte (WBC) PET/CT in the diagnosis of infection using different SUV and visual thresholds for interpretation. Patients, material, method: 49 consecutive patients (27 men, 22 women, mean age: 55.7 years, range: 16-89 years) with suspected musculoskeletal infection (n = 34), vascular graft infection (n = 5), aortitis (n =1 ), endocarditis (n = 1), mass lesion which is suspicious for infection or malignity (n = 6), and fever of unknown origin (n = 2) underwent both FDG and WBC-PET/CT. Images were evaluated by both visual analysis (grade 1-3) according to uptake intensity and quantitative grading (grade 1-3) based on lesion to background SUVmax values. Final diagnosis was made by histopathological, microbiological analysis or clinical-radiological work-up.
Results: The diagnosis of infection was made in total 24 patients, of whom 14 were diagnosed by histopathological and the rest by clinical-radiological work-up. WBC-PET/CT imaging with the visual threshold of 1b as infection positivity (for truncal lesions uptake equivalent to liver or lumbar vertebrae uptake; for extremity lesions uptake significantly higher than neighbouring soft tissue uptake or higher than neighbouring bone marrow uptake) was found to have the highest diagnostic accuracy (AUC: 0.874, CI: 0.771-0.997, p < 0.001). The optimal SUV threshold was found to be 8.8 (p = 0.006; sensitivity: 72.7%, specificity: 82.8) and 5.3 (p < 0.001; sensitivity: 81.8%, specificity: 79.3%) for FDG and WBC-PET/CT, respectively by ROC curve analysis.
Conclusion: WBC-PET/CT is more valuable than FDG PET/CT in the imaging of infection. Visual threshold of >1b seems to be more suitable for detection of infection.
Keywords: FDG-labeled leukocyte PET/CT; infection.