Semi-Quantitative [18F]FDG-PET/CT ROC-Analysis-Based Cut-Offs for Aortitis Definition in Giant Cell Arteritis.

ABSTRACT

[18F]fluorodeoxyglucose-positron emission tomography/computed tomography ([18F]FDG-PET/CT) is used to diagnose large vessel vasculitis in giant cell arteritis (GCA). We aimed to define a semi-quantitative threshold for identifying GCA aortitis from aortic atheroma or the control. Contrast enhanced computed tomography (CECT) was used as the reference imaging for aortic evaluation and to define aortitis, aortic atheroma and control aortas. [18F]FDG-PET/CT was performed on 35 GCA patients and in two different control groups (aortic atheroma (n = 70) and normal control (n = 35)). Aortic semi-quantitative features were compared between the three groups. GCA patients without aortitis on CECT were excluded. Of the GCA patients, 19 (54.3%) were not on glucocorticoids (GC) prior to [18F]FDG-PET/CT. The SUVmax, TBRblood and TBRliver aortic values were significantly higher in the GCA aortitis group than in the aortic atheroma and control groups (p < 0.001). Receiver operating characteristic curve analyses brought to light quantitative cut-off values allowing GCA aortitis diagnosis with optimal sensitivity and specificity versus control or aortic atheroma patients for each PET-based feature analyzed. Considering the overall aorta, a SUVmax threshold of 3.25 and a TBRblood threshold of 1.75 had a specificity of 83% and 75%, respectively, a sensitivity of 81% and 81%, respectively, and the area under the ROC curve (AUC) was 0.86 and 0.83, respectively, for aortitis detection compared to control groups in GCA cases with GC. A SUVmax threshold of 3.45 and a TBRblood threshold of 1.97 had a specificity of 90% and 93%, respectively, a sensitivity of 89% and 89%, respectively, with an AUC of 0.89 and 0.96, respectively, for aortitis detection compared to the control in GC-free GCA cases. Discriminative thresholds of SUVmax and TBRblood for the diagnosis of GCA aortitis were established using CECT as the reference imaging.