• rheumatoid arthritis
• methotrexate
• BAFF
• vaccination

Methotrexate (MTX) with its proven efficacy and safety profile remains as the anchor drug for the treatment of rheumatoid arthritis (RA).1 2 However, the impact of MTX alone or in conjunction with antitumour necrosis factor (anti-TNF) on humoral immune system and infection risk varies markedly among patients with RA, suggesting that other host factors influence the therapeutic response to MTX and/or anti-TNF treatment.3 A possible candidate is B-cell activating factor (BAFF), which promotes B-cell activation and differentiation for antibody production.4 When patients with RA received anti-TNF treatment, a high BAFF serum level prevented formation of antidrug antibody in patients taking MTX but not those who did not.5 Thus, in the presence of MTX, BAFF may exert a paradoxical anti-inflammatory effect. Here, we investigated whether high BAFF levels negatively impact vaccine response via the inhibitory BAFF–MTX interaction in patients with RA taking MTX.

Patients with RA according to the revised 1987 American College of Rheumatology from the randomised controlled trial (ClinicalTrials.gov identifier: NCT02897011) that aimed to investigate the effects of a 2-week MTX discontinuation on vaccine response to seasonal influenza vaccination were included in this study.6 Patients with RA were randomised to continue MTX or to hold MTX for 2 weeks after vaccination with 2016–2017 seasonal quadrivalent influenza vaccine that contained H1N1, H3N2, B-Yamagata and Victoria (GC Influenza, GC Pharma, South Korea). BAFF levels at vaccination and antibody titres to influenza antigens at baseline and 4 weeks after vaccination were measured (online methods and supplementary figure S1). A positive vaccine response was defined as a ≥4-fold increase in haemagglutination inhibition antibody titre.

Baseline characteristics of 316 patients (156 in the MTX-continue group and 160 in the MTX-hold group) were summarised in table 1. Baseline BAFF levels did not differ between the MTX-continue group and the MTX-hold group (866.1 (703.4–1036.2) vs 841.6 (688.4–108.9) pg/mL, p=0.741). The BAFF levels correlated with patient’s age, prednisolone dose and absolute lymphocyte counts but not with disease activity, rheumatoid factor titre, anticyclic citrullinated peptide-antibody titre or MTX dose (online supplementary table S1). In the MTX-continue group, vaccine responders had significantly lower BAFF levels than the non-responders except in response in ≥1/4 antigens (figure 1A, left panel). However, BAFF levels did not differ between the vaccine responders and the non-responders in the MTX-hold group (figure 1A, right panel). Similarly, the antibody titre changes relative to the baseline against individual antigen (except against H3N2) correlated inversely with the respective serum BAFF levels in the MTX-continue group but not in the MTX-hold group (figure 1B). The impact of the MTX and BAFF interaction on antibody formation was significant for H1N1 (p=0.047), Yamagata (p=0.019) and Victoria (p=0.045) but not for H3N2 (p=0.177). The inverse correlation between BAFF levels and antibody production seemed to be more robust in patients taking MTX>15 mg/week than those taking MTX<7.5 mg/week (online supplementary table S2). Use of biologics and corticosteroids did not influence antibody formation (online supplementary table S3).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

(A) Vaccine response depends on the MTX–BAFF interaction. BAFF levels were lower in the responders than the non-responders (according to number of antigen combination) in the MTX-continue group (left panel) but not when MTX was discontinued during peri-vaccination period (right panel). Box plot indicates the median and IQR, while whiskers indicate 10th and 90th percentile. P values were generated by Mann-Whitney test. (B) Correlation between BAFF levels and antibody formation against individual influenza strain. Fold changes in antibody titres relative to the baseline were plotted against their respective baseline BAFF levels in the MTX-continue group (upper panels) and the MTX-hold group (lower panels). Correlation was examined by using Spearman correlation. BAFF, B-cell activation factor; MTX, methotrexate; Pt No (NR/R), number of patients who responded and who did not respond to influenza antigen combination from ≥1/4 to 4/4 (responder/non-responder); r, Spearman rho.

MTX in the presence of higher (and not lower) BAFF levels negatively impacted vaccine response to seasonal influenza vaccination, further supporting the counter-intuitive, paradoxical immune suppressive effect of BAFF in the presence of MTX. This hypothesis-driven study is a first proof of concept to confirm the recent basic-translational finding by Bitoun et al that provides a biological explanation of MTX–BAFF interaction that induces a tolerance to biological disease modifying antirheumatic drugs (ie, TNF inhibitor) and new antigens such as vaccination by generating immune suppressive adenosine and regulatory B cells.5 This study supports that BAFF–MTX interaction at the time of antigen challenge is critical and that immune modulation by DMARDs depends on host immune factors. A soluble BAFF might serve as a surrogate marker of vaccination response in patients with RA taking MTX. Targeting the BAFF–MTX interaction might offer novel therapeutic approaches in RA treatment.