Objective: To evaluate the diagnostic performance of 18F-FDG PET/CT in the detection of stent graft infection (SGI).
Methods: In a retrospective study, two nuclear medicine physicians have independently analyzed 17 18F-FDG PET/CT examinations performed for clinical suspicion of SGI. The images were evaluated for the uptake pattern and intensity, and by the maximum standard uptake value (SUVmax), the target-to-background ratio with blood pool (TBRBP) and liver uptake (TBRhep) as a reference. The SGI was defined as the presence of focal hyperactivity with an intensity exceeding hepatic uptake. CT images were independently assessed for signs of SGI. Clinical review of all further patients' data served as the standard of reference.
Results: Nine cases were established as SGI by the clinical review. PET/CT correctly diagnosed SGI in eight and yielded a sensitivity of 89% and specificity of 100%. The mean SUVmax, TBRBP, and TBRhep values were 9.8 ± 4.0, 6.9 ± 2.6, and 4.6 ± 1.7 in the group of patients with true SGI, and 4.0 ± 1.1, 2.5 ± 0.4 (p < 0.001) and 1.9 ± 0.2 (p < 0.001) in true negative cases, respectively. CT alone showed a sensitivity of 78% and specificity of 100% and was concordant with PET/CT in 14 cases. The best performing threshold values of SUVmax, TBRBP, and TBRhep were 5.6, 3.5, and 2.2, respectively.
Conclusion: 18F-FDG PET/CT with expert evaluation, semiquantitative and quantitative image analysis with the proposed threshold values for SUVmax, TBRBP, and TBRhep has good diagnostic accuracy in the detection of SGI. We propose that visual grading scale for SGI should use hepatic uptake as a visual reference.
Keywords: Computed tomography; Fluorodeoxyglucose F18; Infection; Positron emission tomography; Sensitivity and specificity; Stent.