Interleukin (IL)-33 may play a role in rheumatoid arthritis (RA) pathophysiology as shown by human studies and murine models [1]. Previously, we demonstrated that detectable serum IL-33 predicts clinical response to rituximab independently of auto-antibody status [2].

Here, we aimed to investigate whether the prediction of therapeutic response using serum IL-33 level is generalizable to all biologic agents, including TNF inhibitors (TNFi) and non-TNFi in RA.

We set up an ancillary study of the ROC (Rotation or Change of Biotherapy After First Anti-TNF Treatment Failure for RA) trial (NCT01000441) which compared the efficacy of TNFi vs non-TNFi in patients with insufficient response to a first TNFi [3]. Three hundred patients were randomized, and treatment efficacy was evaluated at 24 weeks according to EULAR response, showing that a non-TNFi was more effective in achieving EULAR response than a TNFi. Serum IL-33 level was assessed before treatment using an accurate enzyme-linked immunosorbent assay (ELISA IL-33, Quantikine, R&D Systems) [4]. Statistical analyses used Prism (Mann-Whitney and Fisher tests for quantitative and qualitative values, respectively). Serum IL-33 level was defined as detectable when > 6.25 pg/mL (lower threshold).

Results were analyzed for 267 patients with available serum and clinical data (Table 1). Serum IL-33 level was detectable for 109/267 (40.8%) patients (mean ± standard deviation serum level was 49.7 ± 61.0 pg/mL when detectable) (Table 2). IL-33 detection was associated with auto-antibody positivity: rheumatoid factor (RF) and/or anti-cyclic citrullinated peptide antibody (anti-CCP), either combined or analyzed separately (Table 3). Auto-antibody positivity was not associated with response to the different treatment: TNFi (N = 132, odds ratio (OR) = 1.1, 95% confidence interval (CI) = 0.39–3.16), non-TNFi (N = 130, OR = 1.5, 95% CI = 0.40–5.62), or different sub-groups of non-TNFi (data not shown). There was no association between IL-33 detection and response to TNFi as well as to non-TNFi drugs overall or analyzed separately (Table 2). Likewise, there was no difference when comparing the levels of serum IL-33 between responders and non-responders in TNFi and non-TNFi groups (data not shown).

Table 1 Characteristics of the patients included in the ancillary study of the ROC trial
Full size table
Table 2 EULAR response, IL-33 detectability rates and association between IL-33 detection and response to tumor necrosis factor inhibitor (TNFi; including adalimumab, certolizumab, etanercept and infliximab) and non-TNFi (including abatacept, rituximab, and tocilizumab) in patients from the ROC study
Full size table
Table 3 Association between IL-33 detection and auto-antibody positivity
Full size table

Thus, this new study confirms the association between serum IL-33 detection and seropositivity in RA patients. However, it did not replicate the association between IL-33 detection and response to rituximab. This may be due to a lack of power related to the number of patients who received this treatment (N = 37), but it may also reflect the difficulty of studying IL-33 as a possible predictor of response given its association with seropositivity, which is a well-known factor associated with response to some biologics such as rituximab or abatacept [5].

In conclusion, we confirm that serum IL-33 detection is associated with auto-antibody positivity but is not a predictive marker for response to TNFi and non-TNFi in RA.