Frontiers in Medicine (Sep 2022)
Diagnostic accuracy and validation of 18F-fluorodeoxyglucose positron emission tomography scores in a large cohort of patients with polymyalgia rheumatica
Abstract
BackgroundSeveral studies have shown that 18F-FDG PET may contribute to the diagnosis of polymyalgia rheumatica (PMR). Previously, we developed a composite PET score called the Leuven score, which was recently adapted to the more concise Leuven/Groningen score by van der Geest et al. The aim of this study is to validate and compare the diagnostic accuracy and cut-off points of both scores in a large cohort of PMR patients.MethodsPatients with a possible clinical diagnosis of PMR and a PET scan prior to the initiation of glucocorticoids between 2003 and 2020 were included retrospectively. The gold standard for the diagnosis of PMR was the judgment of two experienced clinicians after a follow-up of at least 6 months. FDG uptake was scored visually in 12 articular regions (scores 0–2) and a total skeletal score was calculated by summing the individual scores (maximum of 24 for the Leuven score and 14 for the Leuven/Groningen score). Receiver operating characteristic (ROC) analysis and the Youden index were used to determine the diagnostic accuracy and optimal cut-off points.ResultsA total of 162 patients with PMR and 83 control patients were included. Both PET scores showed high diagnostic accuracy in the ROC analysis (area under the curve 0.986 and 0.980, respectively). The Leuven Score provided a sensitivity of 91.4%, specificity of 97.6% and accuracy of 93.5% at its predefined cut-off point of 16. With the newly determined cut-off point of 12 the sensitivity was 98.8%, the specificity 95.2% and the accuracy 97.6%. The Leuven/Groningen score had a sensitivity, specificity and accuracy of 93.2%, 95.2%, and 93.9%, respectively, with the pre-specified cut-off point of 8, and 96.9%, 92.8%, and 95.5% with the optimal cut-off point of 7.ConclusionThe original Leuven score and the simplified Leuven/Groningen score both had excellent diagnostic accuracy. The latter may be easier to apply in clinical practice.
Keywords
