Differential diagnostic performance of PET/CT in adult-onset still’s disease and lymphoma: a retrospective pilot study

Abstract

Adult-onset still's disease (AOSD) and lymphoma are the common causes of fever of unknown origin (FUO) and show some similar clinical symptoms. This study aimed to establish a reliable and easy-to-used scoring model based on clinical information, laboratory characteristics and 18F-fluorodeoxyglucose positron emission tomography/computer tomography (18F-FDG PET/CT) images for the differential diagnosis of these two diseases. A development cohort including 70 AOSD and 37 lymphoma patients was used to establish a scoring model based on the features of PET/CT images. The scoring model was then validated in a validation cohort of 15 AOSD and 12 lymphoma patients. The features of involved bone marrow, spleen, lymph nodes, and other organs or tissues displayed on PET/CT images were compared. Multiple logistics regression and decision tree analysis were used to establish the scoring model. Four features that could significantly differentiate these two diseases were selected to establish a scoring model discriminating AOSD from lymphoma, including (I) white blood cell (WBC) count ≤10×109/L (1 point); (II) ferritin ≤ upper limit of normal (ULN) (1 point); (III) no abnormal bone marrow metabolism (1 point); (IV) total lesion glycolysistotal (TLGtotal) >9.0 (1 point). After decision tree analysis, it showed that a score ≤1 indicates AOSD. A score ≥3 strongly suggested lymphoma, with a sensitivity of 81.1% and specificity of 90.0% in the development cohort, and a sensitivity of 58.3% and specificity of 100% in the validation cohort. Our scoring model showed good diagnosis performance in distinguishing AOSD from lymphoma.