Background
Interleukin (IL)-33 is one of the most recently discovered members of the IL-1 cytokine family mediating its biological effects via its binding to its receptor suppression of tumorigenicity (ST)2 [
1]. IL-33 is upregulated in both resident cells and inflammatory infiltrating cells and is released in case of cell injury, thus acting as an alarmin [
2,
3]. IL-33 induces production of Th2 cytokines and eosinophilia, and may activate mast cells that can in turn release several pro-inflammatory cytokines [
4]. IL-33 investigations have been mainly devoted to asthma and allergy, with the development of a targeted IL-33/ST2 axis therapeutic strategy [
5].
IL-33 may also be involved in rheumatoid arthritis (RA) pathogenesis. IL-33 administration exacerbates collagen-induced and K/BxN serum-mediated murine arthritis, and disease severity is reduced in mice treated with sST2-Fc fusion protein or anti-IL-33 monoclonal antibody [
6‐
8]. Extracellular IL-33 is a critical enhancer of tumor necrosis factor (TNF)-induced RA synovial fibroblast activation [
9] and could activate osteoclastogenesis [
10,
11]. In patients with RA, biomarker studies have suggested that the serum level of IL-33 could reflect clinical activity [
12] and disease severity, or predict carotid plaque progression [
13]. However, the role of IL-33 could be paradoxical since, in K/BxN serum transfer-induced arthritis, ST2 but not IL-33 blockade may improve arthritis [
14,
15]. Moreover, IL-33-stimulated mast cells could also suppress monocyte activation [
16] and intracellular IL-33 also has anti-osteoclastogenic and anti-inflammatory properties [
11].
Some recent works have suggested a possible link between IL-33 and B-cell biology [
17]. In mice, IL-33 enhances immunoglobulin (Ig)M synthesis and markedly induces and activates B1 cells in an ST2-dependent manner [
18]. Additionally, IL-33 could also induce regulatory B cells to produce IL-10, attenuating mucosal inflammation in the gut [
19].
Using a transcriptomic approach, we have found that increased IL-33 mRNA expression in the whole blood of patients with RA was predictive of the response to rituximab (RTX), a targeted B cell-depleting agent [
20]. We aimed to investigate, using an accurate and simple enzyme-linked immunosorbent assay (ELISA), the possible association between a detectable serum level of the IL-33 protein and a response to RTX in RA patients in different cohorts.
Discussion
In this study, we have identified serum IL-33 detection as a novel biomarker associated with RTX response in RA, in addition to auto-antibody status, in real life patients.
This study was based on our previous observation of the upregulation of IL-33 mRNA expression in the whole blood that was associated with RTX response in a microarray study performed in patients randomly selected from the SMART trial [
20]. Since the IL-33 protein may be easily quantified in the serum, we first assessed serum IL-33 levels in the same population using the DuoSet ELISA IL-33 kit (R&D System) and preliminary reported in an abstract that the serum IL-33 level was associated with the RTX response [
23]. However, additional experiments in patients with Sjögren’s syndrome or RA have raised caution about the accuracy of this kit for sera measurements [
24]. Consequently, we discarded these preliminary results and examined serum IL-33 again with an accurate ELISA kit (Quantikine) validated for sera, in two separate and then merged populations.
Here, we have found in univariate and multivariate analysis a significant association between serum IL-33 detection and EULAR response in cohort 2. Furthermore, when we combined the two cohorts, the association was significant with an odds ratio in the same range as high serum IgG level (i.e., approximately 2). We thus confirmed at the protein level the results found at the mRNA level, which demonstrates that transcriptomic analysis with non a-priori hypotheses might open the way to new pathogenic pathways.
This association was independent of strong predictive factors associated with RTX response, especially the presence of RF or anti-CCP antibodies, which strengthens the possible interest in serum IL-33 assessment. We previously reported that patients with a presence of auto-antibodies and high serum IgG level had a better response to RTX in comparison with patients having none of these characteristics [
25]. If serum IL-33 detection is added to these two predictive factors, the likelihood of response to RTX reaches 29.6 (95% CI 1.3–675) in comparison with absence of these three factors. Indeed, 100% of patients displaying these three factors simultaneously were responsive to RTX in our combined cohort. However, these patients represented 9% of the study population.
Auto-antibody status was associated with RTX response in SMART, as previously reported [
25], but also in the combination of cohorts. The absence of association in cohort 2 alone may be explained by the presence of these auto-antibodies in almost all the patients (97%) limiting analysis on these markers. Lastly, as in every trial, it is easier to have a response when the starting DAS28 is high and high disease activity was associated with a better response to RTX, as previously observed in SMART [
25].
Despite a number of strengths, and especially the use of a replication cohort, this study has several limitations. First, serum IL-33 assessment needs the use of an accurate assay such as the Quantikine kit, but the relatively low frequency of patients with detectable IL-33 level (33.5% in the merged population) justifies qualitative IL-33 detection rather than quantitative IL-33 values. Second, we have no data on the association between serum IL-33 detection and response to other biologic agents in RA, limiting our findings to RTX. Third, the association between serum IL-33 detection and response to RTX is statistically significant, but the usefulness of such a measurement for clinical practice needs to be further investigated. Finally, the presence of RF and anti-CCP auto-antibodies in 97% of the patients in cohort 2 might limit analysis on these markers, as well as the difference in terms of high disease activity (DAS 28-CRP >5.1) frequency between the two cohorts (cohort 1 = 77% versus cohort 2 = 49%).
Acknowledgements
We thank the other members of the scientific committee of the SMART study: Pr. J. Sibilia (Strasbourg, France), Pr. J. Tebib (Lyon, France), Pr. B. Combe (Montpellier, France), and Pr. X. Le Loët (Rouen, France).
We thank all the SMART investigators: Dr. I. Azais, Poitiers; Dr. J.C. Balblanc, Belfort; Dr. F. Berenbaum, Paris; Dr. P. Bertin, Limoges; Dr. M.-C. Boissier, Bobigny; Dr. P. Bourgeois, Paris; Dr. A. Cantagrel, Toulouse; Dr. P. Carli, Toulon; Dr. P.-Y. Chouc, Marseille; Dr. M. Couret, Valence; Dr. L. Euller-Ziegler, Nice; Dr. P. Fardellone, Amiens; Dr. P. Fauquert, Berck/Mer; Dr. R.-M. Flipo, Lille; Dr. P. Gaudin, Echirolles; Dr. J.-L. Grauer, Aix en Provences; Dr. A. Heraud, Libourne; Dr. P. Hilliquin, Corbeil; Dr. S. Hoang, Vannes; Dr. E. Houvenagel, Lomme; Dr. D. Keita, Paris; Dr. K. Lassoued, Cahors; Dr. L. Le Dantec, Lievin; Dr. J.-M. Le Parc, Boulogne; Dr. L. Lequen, Pau; Dr. F. Lioté, Paris; Dr. C. Marcelli, Caen; Dr. O. Meyer, Paris; Dr. J.-L. Pellegrin, Pessac; Dr. A. Perdriger, Rennes; Dr. G. Rajzbaum, Paris; Dr. S. Redeker, Abbeville; Dr. J.-M. Ristori, Clermont-Ferrand; Dr. A. Saraux, Brest; Dr. G. Tanguy, La Roche sur Yon; Dr. T. Thomas, Saint-Priest-en-Jarez; Dr. L. Zabraniecki, Toulouse, Dr. C. Zarnitski, Montivilliers, France.
We thank Dr. Pascale Boisseaux and François Gavini (Roche, France) for supporting this ancillary study.
The authors thank Laura Smales (BioMed Editing, Toronto, Canada) for editing the manuscript and Statitec (Toulouse, France) for independent statistical analysis.