Diagnostic accuracy of positron emission tomography for assessment of disease activity in large vessel vasculitis

Aim To determine the diagnostic yield of F-18-fluorodeoxyglucose positron emission tomography (PET) in disease activity assessment of large vessel vasculitides (LVV). Methods Patients with LVV who had undergone PET (between 2004 and June 2010) or PET co-registered with computed tomography (PET/CT; since June 2010) were identified. Clinical disease activity was assessed using established scoring systems. PET images were reviewed by two blinded nuclear medicine physicians. Uptake of the aortic wall was compared to the liver uptake utilizing a visual 4-point score, with a vessel wall uptake similar or higher than liver uptake considered as active disease. Various target-to-background ratios were calculated. Receiver operator characteristics analysis was applied to determine the diagnostic accuracy of PET for detecting clinically active disease. Interobserver agreement of visual readings was measured with Cohen ' s kappa. Results Eighty examinations in 62 patients were analyzed, with a mean time between diagnosis and PET of 106 +/- 171 weeks. Fifty-seven cases were finally classified as clinically active and 23 cases as clinically inactive. With a cut-off value of 1.3, the aorta-to-liver ratio yielded a sensitivity and specificity of 84.2% and 82.6% (area under the curve 0.9). Overall, sensitivity and specificity of visual analysis were 68.4% and 91.3%, but sensitivity decreased to 54% in patients treated for more than 3 months. Interobserver agreement of visual rating was excellent (kappa: 0.93). Conclusion Positron emission tomography is specific and reliable in disease activity assessment of LVV, but lacks sensitivity for detecting active disease in patients under long-term immunosuppressive treatment.