Comparison between visual and numerical metrics for the evaluation of patients with Takayasu arteritis with 18F-FDG-PET

    loading  Checking for direct PDF access through Ovid

Abstract

Introduction

The choice of metrics for defining active Takayasu arteritis (TAK) using fluorine-18-fluorodeoxyglucose (18F-FDG)-PET remains controversial.

Objective

The aim of this study was to compare in the same patients the diagnostic performance for the detection of active TAK of different metrics applied for the quantification of vascular 18F-FDG uptake with PET.

Patients and methods

Overall, 62 PET acquisitions were performed 90 min after 18F-FDG injection in 15 patients with TAK and analyzed retrospectively. The intensity of vascular 18F-FDG uptake was graded visually in comparison with the liver signal and with the numerical metrics, including maximum standard uptake value (SUV), maximum target to background ratio (TBR, ratio of SUVmax in the vessel wall and SUVmean of blood), most-diseased segment (MDS)-TBR (average of TBR from all active lesions), and global TBR (average TBR along the aorta and carotid arteries). The gold standard was disease activity identified using the National Institute of Health score for TAK.

Results

Using visual analysis, the definition of 18F-FDG-PET as positive in presence of at least one vascular lesion with a signal more than liver provided the best diagnostic performance for detecting active TAK with a specificity of 98%, a sensitivity of 62% and an accuracy of 89%. Using numerical metrics, SUVmax [SUVmax >3.3; area under the curve (AUC)=0.84] and TBRmax (TBRmax >2.3; AUC=0.84) offered the best diagnostic performance for the detection of active TAK in comparison with MDS-TBR (MDS-TBR>1.7; AUC=0.70) and global TBR (global TBR >1.4; AUC=0.51).

Conclusion

In this study, we found that the analysis of the vascular region with the highest 18F-FDG uptake using either visual or numerical metrics provided the best diagnostic performance for the detection of active TAK with PET.

Related Topics

    loading  Loading Related Articles