Development and initial validation of the ONCOREUM score to differentiate childhood cancer with arthropathy from juvenile idiopathic arthritis

Objective: To develop and validate a weighted score, the ONCOREUM score, that aids physicians in differentiation of cancer with arthropathy from juvenile idiopathic arthritis (JIA). Study design: Data were extracted from the ONCOREUM Study, a multicenter, cross-sectional investigation aimed at comparing children with cancer and arthropathy to children with JIA. Three statistical approaches were applied to develop the ONCOREUM score and assess the role of each variable in the diagnosis of cancer with arthropathy, including 2 approaches based on multivariable stepwise selection (models 1 and 2) and 1 approach on a Bayesian model averaging method (model 3). The β coefficients estimated in the models were used to assign score points. Considering that not missing a child with cancer is a mandatory clinical objective, discriminating performance was assessed by fixing sensitivity at 100%. Score performance was evaluated in both developmental and validation samples (representing 80% and 20% of the study population, respectively). Results: Patients with cancer and arthropathy (49 with solid tumors and 46 with hematologic malignancies without peripheral blasts) and 677 patients with JIA were included. The highest area under the receiver operating characteristic (ROC) curve (AUC) in the validation data set was yielded by model 1, which was selected to constitute the ONCOREUM score. The score ranged from -18 to 21.8, and the optimal cutoff obtained through ROC analysis was -6. The sensitivity, specificity, and AUC of the cutoff in the validation sample were 100%, 70%, and 0.85, respectively. Conclusions: The ONCOREUM score is a powerful and easily applicable tool that may facilitate early differentiation of malignancies with articular complaints from JIA.