Abstract
OBJECTIVES: Undifferentiated arthritis(UA) is clinically heterogeneous and differs in outcomes ranging from spontaneous resolution to RA-development. Therefore, we hypothesized that subgroups exist within UA and we aimed to identify homogeneous groups based on clinical features, and thereafter to relate these groups to the outcomes spontaneous resolution and RA-development. These outcomes can only be studied in UA-patients in which DMARD-treatment does not influence the natural disease course; these cohorts are scarce. METHODS: We studied autoantibody-negative UA-patients (not fulfilling 1987/2010 RA-criteria, no alternate diagnosis), included in the Leiden Early Arthritis Clinic between 1993 and 2006, when early DMARD-treatment in UA was infrequent. Latent class analysis was used to identify subgroups based on combinations of clinical features. Within these subgroups, test-characteristics were assessed for spontaneous resolution of arthritis and RA-development within 1 year. RESULTS: 310 consecutive UA-patients were studied. Five classes were identified: location and number of swollen joints were most distinguishing. Classes were characterized by: 1) polyarthritis, often symmetric; 2) oligoarthritis, frequently with subacute onset; 3) wrist-monoarthritis, often with subacute onset, increased BMI and without morning stiffness; 4) small-joint monoarthritis, often without increased acute phase reactants, and 5) large-joint monoarthritis, often with subacute onset. Studying the classes in relation to the outcomes revealed that patients without spontaneous resolution (thus having persistent disease) were nearly absent in the classes characterized by monoarthritis (specificity >90%). Additionally, patients who developed RA were infrequent in monoarthritis classes (sensitivity <7%). CONCLUSION: Using a data-driven unsupervised approach, five subgroups within contemporary UA were identified. These have differences in the natural course of disease.