Introduction: Vascular prosthesis infection (VPI) is a life-threatening complication that occurs in 0.5-5% of prostheses. Low-grade infections in non-acute patients are a diagnostic challenge requiring a new method with good diagnostic accuracy. The aim of this work was to define the accuracy of (18)F-FDG PET/CT in these settings and to identify essential parameters of the evaluation.
Material and methods: PET/CT was performed prospectively in 76 consecutive patients with a total of 96 vascular prosthetic grafts in which infection was suspected. PET/CT scans were analysed in terms of the presence and intensity of focal and diffuse FDG uptake, the presence of an anastomotic pseudoaneurysm, the presence of an irregular boundary of infiltration, a combination of these, and the uptake ratio between the graft and blood background. The gold standard was based on operative/histopathological finding or a clinical follow up of >6 months.
Results: Among the various assessed parameters only focal FDG uptake and an irregular graft boundary were significant predictors of VPI. Focal intense FDG uptake together with an irregular boundary of the lesion on CT scan predicted VPI with 97% probability, while smooth lesion boundaries and no focal FDG uptake predicted a probability of VPI of less than 5%. Even in lesions with nondiagnostic inhomogeneous focal FDG uptake (18/96) an irregular boundary effectively helped in decision-making with a probability of 28% (smooth) or 77% (irregular) for VPI.
Conclusion: PET/CT gave reliable results with an accuracy >95% in 75% of prostheses. PET/CT can identify those prostheses (25% of prosthesis) for which its diagnostic accuracy is diminished to 70-75%. In our series PET/CT was an excellent diagnostic modality for suspected VPI.