In recent years, we are witnessing a growing demand for 18F-FDG PET/CT scans in rheumatologic patients with suspected large-vessel vasculitis (LVV), particularly in cases with non-specific presentation such as fever, malaise and weight loss. In addition to visual analysis of PET/CT images, various semi-quantitative parameters have been proposed to evaluate FDG uptake in vessels walls, to achieve more reproducible results among different operators: anyway, there’s no unanimity about the best one. Aim of the study was to evaluate the accuracy of 18F-FDG PET/CT in the diagnosis of LVV, comparing the results obtained by visual analysis and by using different semi-quantitative parameters: SUVmax of the vessel wall, vessel-to-liver (SUVmax e SUVmean of the liver), vessel-to-lung, vessel-to-blood-pool.
47 consecutive patients, undergone 18F-FDG PET/CT between May 2010 and April 2014 for fever of unknown origin or suspected vasculitis, were enrolled in the study. The diagnosis of LVV was confirmed in 17 patients by temporal artery biopsy, other imaging techniques and a follow-up of at least six months. All scans were acquired before starting of steroid treatment (to avoid any interference of the therapy) on an integrated PET/CT camera equipped with a full-ring dedicated PET scanner and a sixteen-slices CT scanner, 60 minutes after injection of 3,7 MBq/Kg of 18F-FDG. First, PET/CT images were visually analyzed; then, multiple regions-of-interest were positioned on large arteries (ascending and descending aorta, supra-aortic trunks, common iliac vessels) and on background regions (liver, lungs, lumen of vena cava) to measure the uptake of the tracer in vessels walls and to get the ratio values. The results obtained in positive and negative patients were compared and the diagnostic accuracy of the above-mentioned parameters was evaluated by means of ROC curves.
SUVmax of the vessel wall was significantly higher in positive patients than in negative ones and the same applies to the vessel-to-liver (SUVmax e SUVmean), vessel-to-lung, vessel-to-blood-pool ratios (p < 0.01). No differences were found in mean tracer uptake of liver, lungs and blood pool between the two groups. Visual analysis had a sensitivity of 94% and a specificity of 80% in the diagnosis of vasculitis. The performances of the semi-quantitative parameters were globally superior, with sensitivity varying between 94-100% and specificity between 87-97%. The best result was obtained with the vessel-to-liver ratio (using liver SUVmean), setting a cutoff value of 1.44: 100% sensitivity, 96.7% specificity and an AUC (area-under-curve) of 0.998.
The results of our study confirm that 18F-FDG PET/CT could play a leading role in the diagnosis of LVV; its diagnostic accuracy, already good when a simple visual analysis is performed, reaches higher levels when tracer uptake in vessels walls is evaluated with semi-quantitative parameters. Among them, in our series, best diagnostic results have been obtained with the vessel-to-liver ratio, using SUVmax to measure FDG uptake in the vessels walls and SUVmean to measure the “background” activity in the liver.