Purpose: The aim of this study was the evaluation of quantitative and qualitative parameters for the diagnosis of aortic graft infection (AGI) using [18F]-FDG PET/CT.
Methods: PET/CT was performed in 50 patients with clinically suspected AGI. 12 oncological patients with aortic repair but without suspicion of AGI were included in the analysis to serve as control cohort. The [18F]-FDG uptake pattern around the graft was assessed using (a) a five-point visual grading scale (VGS), (b) SUVmax and (c) different graft-to-background ratios (GBRs). The diagnostic performance of VGS, SUVmax and GBRs was assessed and compared by ROC analysis.
Results: 28 infected and 34 uninfected grafts were identified by standard of reference. SUVmax and VGS were the most powerful predictors for the diagnosis of AGI according to the area under the curve (AUC 0.988 and 0.983, respectively) without a significant difference compared to GBRs. SUVmax and VGS showed congruent and accurate findings in 54 patients (i.e. either both positive or negative), yielding sensitivity and specificity (100%) in this subgroup of patients.
Conclusion: Quantitative analysis by SUVmax and qualitative analysis by VGS are highly effective in the diagnosis of AGI and should be tested as an outcome measure in prospective trials.
Keywords: CT; Hybrid imaging; Infection; Inflammation; Multimodality; PET.
© 2020. American Society of Nuclear Cardiology.