CLINICAL AND BIOLOGICALLY-BASED APPROACHES FOR CLASSIFYING AND PREDICTING EARLY OUTCOMES OF CHRONIC CHILDHOOD ARTHRITIS

Background: Juvenile idiopathic arthritis (JIA) comprises a heterogeneous group of conditions that share chronic arthritis as a common characteristic. Current classification criteria for chronic childhood arthritis have limitations. Despite new treatment strategies and medications, some continue to have persistently active and disabling disease as adults. Few predictors of poor outcomes have been identified. Objectives: This thesis comprises two complementary studies. The objective of the first study was to identify discrete clusters comprising clinical features and inflammatory biomarkers in children with JIA and to compare them with the current JIA categories that have been proposed by the International League of Associations for Rheumatology. The second study aimed to identify predictors of short-term arthritis activity based on clinical and biomarker profiles in JIA patients. Methods: For both studies we utilized data that were collected in a Canadian nation-wide, prospective, longitudinal cohort study titled Biologically-Based Outcome Predictors in JIA. Clustering and classification algorithms were applied to the data to accomplish both study objectives. Results: This research identified three clusters of patients in visit 1 (enrolment) and five clusters in visit 2 (6-month). Clusters revealed in this analysis exposed different and more homogenous subgroups compared to the seven conventional JIA categories. In the second study, the presence or absence of active joints, physician global assessments, and Wallace criteria were chosen as outcome variables 18 months post-enrolment. Among 112 variables, 17 were selected as the best predictors of 18-month outcomes. The panel predicted presence or absence of active arthritis, physician global assessment, and Wallace criteria of inactive disease 18 months after diagnosis with 79%, 82%, and 71% accuracy and 0.83, 0.86, 0.82 area under the curve (AUC), respectively. The accuracy and AUC values were higher compared to when only clinical features were used for prediction. Conclusion: Results of this study suggest that certain groups of patients within different JIA categories are more aligned pathobiologically than their separate clinical categorizations suggest. Further, the research found a small number of clinical and inflammatory variables at diagnosis can more accurately predict short-term arthritis activity in JIA than clinical characteristics only