Diagnostic Accuracy Of FDG PET For The Identification Of Vascular Graft Infection
Mu'ath Adlouni, BS, Ahmed Ahmed, MD, Mouaz Al-mallah, MD, Ezra Y. Koh, MD, Eric Peden, MD, Maham Rahimi, MD, PHD.
Houston Methodist Hospital, Houston, TX, USA.
OBJECTIVES: Vascular graft infection (VGI) is associated with high morbidity and mortality with early diagnosis being difficult due to the possible overlap of inflammation and infection on imaging. Despite the numerous diagnostic techniques available, a gold standard for the detection of VGI remains absent. The capability of fluorodeoxyglucose (FDG) positron emission tomography (PET) to diagnose VGI is variable in each institution and its accuracy depends on the cut-off value for maximum standardized uptake value (SUVmax) and the timing of FDG administration to image acquisition. This study aims to determine the diagnostic accuracy of FDG PET in the identification of VGI.
METHODS: We prospectively included 25 consecutive patients (64% males, mean age 66 ±10 years) who underwent FDG PET using an annual digital PET/CT system to identify suspected VGI. Patients were imaged 60 minutes after the injection of 5 mCi of FDG. Multiple beds were imaged for 10 minutes each. Exams were interpreted by a reader who was blinded to the gold standard diagnosis. The gold standard for defining infection versus inflammation was the final pathology on surgery or clinical consensus if surgery was not done. Receiver operator curves were constructed and the optimal cut point for standard uptake value (SUV) that differentiate infection from inflammation was determined using the method proposed by Youden et al.
RESULTS: Of the 25 patients included, 15 patients had infection either by surgical/pathological diagnosis or clinical diagnosis. From the receiver operator curves analysis, the following values gave the best balance between sensitivity and specificity: SUV max of 4.5, SUV mean of 5.3, and a Tissue-to-background ratio (TBR) of 1.6. An SUV max of 4.5 had a sensitivity of 93%, specificity of 90%, Positive Likelihood Ratio (LR+) of 7.88, and negative likelihood ratio (LR-) of 0.14 for determining VGI. The calculated SUV mean had a sensitivity of 90%, specificity of 90%, LR+ of 8.44, and LR- of 0.07. Lastly, the TBR threshold led to a sensitivity of 93%, specificity of 90%, LR+ 7.88, and LR- of 0.14.
CONCLUSIONS: Our data suggests that FDG PET is a reliable and accurate method for detecting VGI. Using the identified threshold values for digital PET SUV maximum, SUV mean, and TBR, FDG PET has a sensitivity and specificity of more than 90%.
Back to 2022 Abstracts