CXCL13 predicts disease activity in early rheumatoid arthritis and could be an indicator of the therapeutic `window of opportunity'

Rheumatoid arthritis (RA) is a chronic autoimmune disease with joint inflammation and autoantibody production as key elements of its pathogenesis. The course of the disease is still difficult to predict. The encouraging results of early, intensive treatment of RA suggest the existence of a `window of opportunity' during which effective therapy can induce long-lasting remission [1]. Unfortunately, it is not known when this `window of opportunity' is open, and the search for informative biomarkers of early inflammation and triggers of memory development therefore becomes a pertinent issue in RA research.

T cells are present in elevated numbers in the synovial joints in RA where they form cellular infiltrates that resemble ectopic lymphoid aggregates with germinal center formation [2]. This suggests the presence of an ongoing antigen presentation and follicle formation in the synovium. The follicle is a well-organized structure, generated by follicular dendritic cells (FDCs), B cells, and follicular helper CD4 T (T_FH) cells. Within the follicle, B cells are activated and matured into long-lived plasma cells, which secrete high-affinity antibodies [3]. The production of autoantibodies is central in RA [4], and the processes leading to follicle formation in the RA synovium are therefore of great interest. The central role of ongoing immune activation in RA development is further supported by the fact that CTLA4 treatment reduces disease activity [5].

The chemokine C-X-C motif chemokine 13 (CXCL13) is necessary for follicle formation and is constitutively expressed in secondary lymphoid tissue, primarily by FDCs [6]. Further, CXCL13 expression is upregulated by tumor necrosis factor alpha (TNFα) and by T cell receptor stimulation [7],[8]. C-X-C chemokine receptor type 5 (CXCR5), the only known receptor for CXCL13, is expressed by naïve B cells and T_FH cells, and it controls the migration of these cells to the follicle [9]. The CXCL13-CXCR5 axis is critical to the generation of immunological memory based on long-lived plasma cells because the interaction between T_FH and B cells is necessary for the formation of plasma cells and autoantibody production [7],[10]. Recently, CXCL13 has risen to be a possible new marker of disease and inflammation in RA. CXCL13 is reported upregulated in RA patients, and is suggested to be connected with both disease activity and rheumatoid factor [11],[12].

In this study, we aim to investigate CXCL13's association with markers of disease activity in patients with early RA, who participated in a double-blind randomized clinical trial of two different treatment regimes.

In this study, we further investigated the role of CXCL13 in RA. We measured high CXCL13 plasma levels in early DMARD-naïve RA patients. Six months of anti-rheumatic treatment reduced plasma CXCL13 to levels observed in healthy volunteers. We also showed that baseline CXCL13 strongly correlated with SDAI, VAS and joint involvement at treatment initiation. These findings contribute to establishing a role for CXCL13 as a potential marker of inflammation in early RA. Our findings are in line with earlier published results on CXCL13 [11],[15],[16], but our study provides new knowledge suggesting CXCL13 as a marker of joint involvement in early RA.

CXCL13 is a pivotal chemokine in establishing an adaptive immune response. It attracts B cells in the secondary lymphoid tissue, which facilitates the generation of antibodies and local inflammation [6],[7]. The observed associations with joint involvement contribute to establishing activity within the lymphoid follicle in early RA as an important mechanism in the progression of RA. Because CXCL13 is produced by synovial cells, CXCL13 could serve as a marker that reflects local activity and inflammation [8]. CXCL13 was not associated with CRP or DAS28CRP. Rioja et al. [17] describes high CXCL13 and DAS28 levels in patients with active vs. inactive RA. In line with these findings, we observed that CXCL13 levels are high in untreated early RA patients (active RA), as is DAS28CRP and CRP. Treatment of early RA reduces disease activity, and thereby also DAS28CRP as well as CXCL13. Thus, although not associated with CRP, CXCL13 remains a potential marker of disease activity in early RA patients.

In the DMARD treated CXCL13-high group, the baseline CXCL13 levels correlated inversely with disease activity markers at 12 months. A priori, one would not expect high levels of CXCL13 to correlate inversely with disease parameters. Rosengren et al. [11] described plasma CXCL13 levels to decrease in accordance with disease activity, indicating CXCL13 and disease parameters to be positively correlated. However, Rosengren et al. examined patients with established RA. Bugatti et al.[15] find fewer patients in clinical remission after one year of treatment, if baseline levels of CXCL13 were high. In line with Bugatti et al.'s study, Meeuwsisse et al. [16] show that high CXCL13 is associated with increased radiographic destruction. We do not find any association with radiographic progression. Our results are of course controversial in comparison with both Meeuwisse et al. and Bugatti et al.'s findings. Though the average disease duration in our cohort is only 3 months, where disease duration in Bugatti's cohort is 1 year and 2 years in Meeuwisse' cohort. We suggest this difference is of major importance, as these very early RA patients comprise a more uniform cohort, because spread in disease increases significantly over time. Our different findings can be explained by the fact that our patients are still in the earliest phases of disease initiation. Also supporting the difference in the patient cohorts is that 67% of patients in Bugatti et al.'s article reached low disease activity after one year, whereas this percentage was 76 to 80% in the OPERA cohort. Again supporting a difference is when patients are treated aggressively and as early as after just 3 months of disease. Jones et al. [12] recently showed, that CXCL13 is associated with rheumatoid factor in RA patients, supporting its importance in antibody production. In our cohort of patients with very early RA, and we did not observe CXCL13 to be associated with rheumatoid factor. Thus, we propose that a high, plasma CXCL13 level in treatment-naïve early RA is a possible indicator of newly developed and reversible inflammation. It is likely that these very early RA patients have neither established a full memory response, nor fully developed a lymphoid follicle antigen response at this earliest stage of disease. This would imply that the memory process to some degree could be halted, possibly by aggressive treatment regimes. In the DMARD + ADA treated CXCL13-high group we do not see this inverse correlation with disease markers. Several studies on TNF^-/- mice elucidate the importance of TNF receptors such as TNF-R1 in fully establishing an immune response [18]-[20]. Thus TNF is required for differentiation of follicular dendritic cells and an antibody response. This could explain the lack of associations in the DMARD + ADA treated group and reflect the difference in treatment response between the two groups. Thus, the DMARD + ADA-treated patients had decreased disease activity after 12 months of treatment compared with the DMARD-treated patients [13]. This supports the hypothesis that adding adalimumab to the treatment regime impairs the development of disease progression and possibly also immunologic memory, while disease progression in the DMARD group is ongoing. We also showed that sustained remission (measured by DAS28CRP <2.6) at 2 years of follow-up, was associated with higher baseline CXCL13. This finding could further support that high baseline CXCL13 may be an indicator of recent-onset and active disease, and that an `open window' for successful treatment does exist when the disease is in its earliest phase. We analyzed if patients with high CXCL13 simply were treated more aggressively, and therefore achieved sustained remission. This was not the case, as evaluated by number of intra-articular steroid injections and addition of hydroxychloroquine and/or sulphasalazine. When we repeated the above analysis, using CRP with a cut of 8 mg/L as a definition of remission, no difference in baseline CXCL13 was observed. This supports the theory that CXCL13 especially reflects joint involvement, and is not just connected to CRP. Based on these very early RA patients from the OPERA cohort, we propose that an initial high level of CXCL13 could be a potential indicator that the patients are more treatment-responsive and thereby within the so-called `window of opportunity'. Adding adalimumab to the treatment regime seems to further improve the chance for remission after two years, especially with high baseline CXCL13. Our findings may therefore also contribute to the explanation of the disease-modifying effects of early aggressive treatment.

Our study suggests that plasma CXCL13 is a marker of early inflammation in general and especially of joint involvement in early RA. Early RA patients with high baseline CXCL13 levels could form a particular patient group whose disease is still very responsive to treatment. This responsiveness could indicate that patients are in the earliest disease stage and within the `window of opportunity' where they probably respond better to early aggressive treatment than patients whose disease has progressed.

All authors contributed to writing the manuscript. SRG performed the statistical analysis, the data interpretation and the drafting of the manuscript. KKS conducted the practical experiments. MLH, KSP, KHP, PJ, and MØ designed and carried out the OPERA trial including: planning the trial, inclusion of patients, sample preparation and follow-up visits. TKR participated in the design of the study and in optimizing and performing the practical experiments. TWK participated in optimizing and performing the practical experiments and in data interpretation. MH and BD planned and supervised the project, participated in drafting of the manuscript and data interpretation. The manuscript was commented on and approved by all the authors. All authors read and approved the final manuscript.