Objectives: With the introduction of high-throughput biomarker measurements, traditional analysis of these markers is increasingly difficult. Using samples from a diverse group of patients, we tested the applicability of cluster analysis to these data. Using this method, we aim to visualize some of the patterns specific to certain disease groups. In particular, we focus on juvenile idiopathic arthritis (JIA), a multifactorial autoimmune disorder that ultimately leads to chronic inflammation of the joints.
Methods: Cytokine measurements were performed using multiplex immunoassays. Using heuristic clustering methods, we set out to compare the pattern of 30 cytokines in plasma and SF of JIA, RA, OA, or diabetes type II patients and healthy controls.
Results: Analysis shows that oligo- and polyarticular JIA have similar biomarker profiles, both in plasma and SF. Systemic onset JIA (SoJIA) has a profile distinct from other JIA subtypes, suggesting that they involve different inflammatory processes. SoJIA samples do, however, cluster together with RA in SF, suggesting that these two conditions have similar cytokine profiles. Furthermore, we identify several clusters of ILs and chemokines that are co-expressed, suggesting that they are co-regulated.
Conclusions: We show that previously undetected clusters of cytokines and patients can be identified by applying cluster analysis to multiplex data. Cytokine clusters identified in plasma and SF samples were quite different, which underscore the differential cytokine signalling in these two compartments, and suggest that plasma samples may not be suitable for estimating joint biomarker profiles and inflammation.