Blood chemokine profile in untreated early rheumatoid arthritis: CXCL10 as a disease activity marker
Chemokines and chemokine receptors are involved in leukocyte migration and inflammatory processes, and are therefore suggested to be involved in RA pathogenesis. Indeed, in this study we found that the patients with ueRA were separated from HC in multivariate discriminant analysis based on the blood chemokine profile. A combined profile of chemokines and T-cell subsets led to stronger separation between ueRA and HC compared to chemokine or T-cell subset profiles alone. The levels of chemokines related to Th1 (CXCL9, CXCL10), Th2 (CCL22), T follicular helper and B cells (CXCL13), and macrophages (CCL4) were significantly higher in ueRA compared to HC, pointing to the role of these chemokines and immune cells in early RA pathogenesis. Among the discriminator chemokines, only the levels of CXCL10 correlated with several clinical disease activity parameters, including DAS28-CRP, DAS28-ESR, CDAI, SJC in 66 joints, CRP, and ESR.
In line with our results, previous studies have reported higher levels of CXCL10 in serum or plasma in patients with both established RA [16] and early RA [13, 14] compared to healthy controls. However, a significant proportion of these patients were treated with DMARD when chemokine levels were assessed. Similarly, higher serum levels of CCL22 in patients with RA compared to healthy controls have been reported previously; however, in that study, the majority of the patients were treated with DMARD and there was no information concerning the disease duration [22]. Various treatments have been shown to affect the levels of chemokines in the circulation [15–18], which suggests that evaluation of chemokines in untreated patients might be advantageous.
Among the evaluated chemokines, only the plasma levels of CXCL10 correlated with the clinical disease activity in our cohort. In a previous study in established RA, a correlation between CXCL10 and DAS28-CRP has been seen, but no correlation with other clinical disease activity measures such as swollen joint counts, CRP, and ESR was observed [15]. Another study in established RA did not observe any correlation between CXCL10 and clinical disease activity measures [18]. Thus, ours is the first study that defines CXCL10 as a disease activity marker in early RA by demonstrating correlation between CXCL10 and multiple clinical disease activity measures. The reduction in CXCL10 with symptom duration and correlations with multiple clinical disease activity measures in early RA but not in established RA suggests that CXCL10 plays a more critical role in the early stages of the disease, and can function as a disease activity marker in early RA.
CXCL10 (traditionally known as IFN?-inducible 10-kd protein or IP-10) can be secreted by several cell types such as endothelial cells, fibroblasts, monocytes, neutrophils, dendritic cells, mesenchymal cells, keratinocytes, astrocytes, and so forth [23–25]. CXCL10 can be induced in response to IFN?, IFN-?/?, IL-1? or tumor necrosis factor (TNF) [26, 27]. Interestingly, the type I interferon (IFN) signature has been demonstrated as a biomarker of preclinical RA [28], and granulocytes are shown to be a major contributor of this signature in early arthritis [29]. Since IFN-? is a potent inducer of CXCL10 [26], high levels of CXCL10 in early RA, despite lower proportions of Th1 cells, may suggest that type I IFN-dependent immune mechanisms could be important in preclinical and early RA. Moreover, the role of Th17 cytokines is well established in neutrophil activation and, according to recent models, Th2 and Th17 pathways may amplify each other in certain conditions [30]. Thus, the dominance of Th2 and Th17 milieu may induce a granulocyte-mediated type I IFN response, which may further induce downstream secretion of CXCL10 and other chemokines contributing to early RA pathogenesis.
CXCL10 can potentially regulate inflammation at several levels, contributing to RA pathogenesis and disease activity. CXCL10 and CXCR3 play crucial roles in leukocyte homing to inflamed tissue and also in perpetuation of inflammation and tissue damage [25]. In particular, CXCL10 promotes directional migration of activated T cells, monocytes, and NK cells, induces integrin activation, and promotes T-cell adhesion to endothelial cells [31, 32]. It can therefore coordinate recruitment of various immune cells to the site of inflammation. CXCL10 can also induce RANKL expression in RA synoviocytes and CD4+ T cells [33], which may induce bone resorption. It has been shown that stimulation of fibroblast-like synoviocytes (FLS) with CXCL10, CXCL9, and CCL2 enhances the proliferation of these cells, which may lead to synovial hyperplasia [34]. Chemokines such as CXCL10, CXCL9, CCL5, CCL4, and CCL2 can also stimulate FLS and chondrocytes to release inflammatory mediators including cytokines, matrix metalloproteinases (MMPs), and other enzymes, leading to degradation of the extracellular matrix and cartilage [34–36]. Higher levels of CXCL10 and other chemokines have indeed been detected both in synovial fluid and synovial tissue of RA patients compared to osteoarthritis patients [37, 38]. In patients with established RA, a chemokine gradient with higher levels in synovial fluid compared to blood has been reported for CXCL10, CXCL9, CCL2, CCL3, CCL4, and CXCL8 [38]. Among these, differences in the levels of chemokines were most dramatic for CXCL10 and CXCL8 (10 times higher concentration in RF synovial fluid compared to RA serum for both chemokines) [38]. The relative levels of chemokines between synovial and peripheral blood compartments in early RA and how a chemokine gradient affects the selective recruitment of immune cell subtypes in the joints of patients with early RA has not been established.
We here show that the levels of CXCL10 levels correlated strongly with those of CXCL11 and moderately with CCL2 and CXCL9. However, only CXCL10 levels correlated with clinical disease. This suggests that the inflammation and disease severity in patients are directly associated with the levels of CXCL10 but not with the other CXCR3 ligands or chemokines. Supporting this notion, a recent genetic association study based on single nucleotide polymorphisms (SNPs) showed that the CXCL10 GG genotype was an independent factor associated with increased probability of extra-articular manifestation development, while CXCL9 genotypes did not display such an association [39]. In a study with synovial fibroblast cell lines from RA patients, significant CXCL9 and CXCL10 secretion, but not CXCL11, could be observed upon IFN? stimulation, and only CXCL10 secretion, but not CXCL9 and CXCL11, was found upon TNF or IL-1? stimulation [27]. However, when these cells were stimulated with a combination of IFN? and TNF, significant secretion of all three chemokines (CXCR3 ligands) was observed [27]. Thus, despite differential regulation of CXCL10 and other chemokines in chronic inflammation, association among several chemokines can be seen due to synovial hyperplasia and paracrine activity/positive feedback loop among various cytokines and chemokines.
It has been reported previously that the serum levels of CXCL13 can act as a disease activity marker in early RA [40, 41]. However, we did not observe any significant correlations between CXCL13 and clinical disease activity measures in our cohort. The reasons for these different results are unclear, but could be due to differences in the percentages of ACPA+ and RF+ patients which varied between our cohort and patient cohorts in previous studies. The patient cohort in the present study had 79% ACPA+ and 77% RF+ patients, while the study by Bugatti et al. [41] had 49% ACPA+ and 57% RF+ patients, and the study by Greisen et al. [40] had 63% ACPA+ and 71% RF+ patients. However, the levels of CXCL13 were not significantly different between ACPA+ and ACPAneg or RF+ and RFneg subgroups in our patient cohort.
Lower levels of eotaxin were found in female patients compared to male patients and HC females in our study cohort. Interestingly, eotaxin associated negatively with clinical disease activity in female patients while it displayed a positive association pattern in male patients. High serum levels of eotaxin have been shown to associate with less radiographic progression in an early rheumatoid arthritis patient cohort which had 77.4% female patients [42]. Based on our results, it would be interesting to evaluate sex-based differences in immune cells and soluble inflammatory mediators in larger cohorts of patients.
A negative association pattern was seen between chemokines and the corresponding T-cell subsets in the present study, which could have several explanations. Chemokine-mediated apoptosis of T cells could be one reason, as it has been shown that chemokines are able to induce apoptosis in T cells depending on the co-stimulating signals and the balance of downstream signaling pathways [43, 44]. Alternatively, binding of chemokines to the chemokine receptor-expressing cells may lead to lower concentrations of chemokine protein in the blood. Another possibility is the migration of T cells to target organs such as joints, but this should depend more on the chemokine gradient between organ tissue and blood and not on the absolute plasma levels of chemokines.
Targeting of chemokines and chemokine receptors have given favorable results in preclinical studies performed in animal models of arthritis [4]. However, human clinical trials targeting CCR2, CCR5, CCL2, and CXCL8 by small molecules or monoclonal antibodies have failed to demonstrate clinical efficacy in RA (reviewed in [4]). In contrast, one publication of a phase II clinical trial using anti-CXCL10 monoclonal antibody (MDX-1100) in established RA patients who had responded inadequately to methotrexate (MTX) treatment showed that the response rate was significantly higher in MDX-1100-treated patients at week 12 according to the American College of Rheumatology 20% criteria for improvement compared to the placebo group [45].