Objectives TNF inhibitors (TNFi) can induce anti-drug antibodies (ADA) in patients with autoimmune diseases (AID) leading to clinical resistance. We explored a new way of using methotrexate (MTX) to decrease this risk of immunisation.
Methods We treated BAFF transgenic (BAFFtg) mice, a model of AID in which immunisation against biologic drugs is high, with different TNFi. We investigated the effect of a single course of MTX during the first exposure to TNFi. Wild-type (WT) and BAFFtg mice were compared for B-Cell surface markers involved in MTX-related purinergic metabolism, adenosine production and regulatory B-cells (Bregs).
We translated the study to macaques and patients with rheumatoid arthritis from the ABIRISK cohort to determine if there was an interaction between serum BAFF levels and MTX that prevented immuniation.
Results In BAFFtg but not in WT mice or macaques, a single course of MTX prevented immunisation against TNFi and maintained drug concentration for over 52 weeks. BAFFtg mice B-cells expressed more CD73 and CD39 compared to WT mice. MTX induced adenosine release from B cells and increased Bregs and precursors. Use of CD73 blocking antibodies reversed MTX-induced tolerance. In patients from the ABIRISK cohort treated with TNFi for chronic inflammatory diseases, high BAFF serum level correlated with absence of ADA to TNFi only in patients cotreated with MTX but not in patients on TNFi monotherapy.
Conclusion MTX and BAFF interact in mice where CD73, adenosine and regulatory B cells were identified as key actors in this phenomenon. MTX and BAFF also interact in patients to prevent ADA formation.
- B cells
- rheumatoid arthritis
Statistics from Altmetric.com
TNF inhibitors (TNFi) have revolutionised the care of rheumatic diseases due to their high specificity allowed by their biologic nature. However, these biologic drugs cause immunogenicity that can lead to a neutralising immune response1 characterised by the production of anti-drug antibodies (ADA). ADA may also be generated against immune checkpoint inhibitors used in the treatment of cancer2 but at a lower frequency than in patients with autoimmune diseases (AID). Thus, generation of ADA may be influenced by the immune status of the patient.
Methotrexate (MTX) is used as an antirheumatic drug and leads to a higher therapeutic effect of TNFi, if used concomitantly.3 This effect might be explained by the diminished production of ADA when patients are treated with TNFi and MTX.3 4 But MTX does not prevent ADA in all patients and the mechanism of action of MTX for preventing ADA formation still remains to be elucidated. Used as a disease-modifying antirheumatic drug, the main proposed mechanism of action of MTX is an increase in adenosine, a powerful anti-inflammatory agent, by decreasing its conversion into inosine and the conversion of AMP into IMP. This occurs mainly by inhibition of AICAR transformylase which leads to a release of AMP outside the cell. AMP is further converted into adenosine by CD73, an ecto-5′-nucleotidase5 6 (online supplementary figure 1). In a mouse model of Pompe’s disease, MTX was successfully used to prevent ADA formation against alpha-glucosidase, a highly immunogenic recombinant enzyme used as a treatment for this disease. Surprisingly, a single course of MTX at the time of first enzyme exposure was able to significantly reduce anti-enzyme antibody concentration. This MTX-induced tolerance to alpha-glucosidase was transferred to naïve mice using splenic B cells.7
Supplementary file 1
The BAFF (B cell activating factor of the TNF family) or BLyS (B-lymphocyte stimulator) cytokine, discovered in 1999, is a key driver of B cell activation. BAFF is secreted mainly by monocytes, macrophages and neutrophils, and may explain pathogenic B cell activation in several systemic AID and lymphomas.8 BAFF targets three receptors present in B cells: BAFF-R, TACI and BCMA. BAFF transgenic (BAFFtg) mice develop AID reminiscent of systemic lupus erythematosus (SLE) and primary Sjögren’s syndrome (pSS) with an increased risk of developing lymphoma.9 In humans, patients with pSS and SLE have elevated serum levels of BAFF,10 11 sometimes correlated with the serum level autoantibodies.10 12 13 Interestingly, belimumab, an anti-BAFF monoclonal antibody, was approved in 2011 for the treatment of SLE and open studies are encouraging in pSS.14
The study aimed to elucidate the role of MTX in the immune response against TNFi in a setting of high immunisation against TNFi using the model of BAFFtg mice and to translate the observed findings to prevent ADA formation in patients.
Mice and treatment
BAFFtg C57BL/6J mice were kindly provided by F Vincent and F Mackay (Monash University Australia) and have been described previously.9 Wild-type (WT) C57BL/6J mice purchased from Janvier Labs (Le Genest-Saint-Isle, France) were used as controls in this study. Genotyping was performed by PCR using primers spanning the SV40 polyA tail of the BAFF transgene construct. The experiments included male and female mice in equal numbers. The experiments were performed either on 2-month-old mice or older mice, depending on the length of the treatment with TNFi. Experiments were performed on matched numbers of males and females with a total of 140 mice. Treatment with MTX (5 mg/kg) was administered intraperitoneally within minutes of the first TNFi injection and repeated only twice at 24 and 48 hours, as reported in the model of Pompe’s disease.7 The dose of 5 mg/kg of MTX administered here three times is equivalent to a dose of 0.1 mg/kg in humans. This is based on pharmaco-kinetics analysis from the literature in rodents and humans.15 16 Adalimumab was administered weekly at 20 mg/kg and etanercept biweekly at 8 mg/kg for 52 weeks. TN3 (a hamster anti-mouse TNFα antibody) was injected once a week at 20 mg/kg for 52 weeks also. To reverse MTX-induced tolerisation an anti-CD73 (clone TY23, BioXcell, West Lebanon, USA) monoclonal antibody was used. Animals were treated with anti-CD73 biweekly either for 1 or 9 weeks, with a first intraperitoneal injection of the anti-CD73 antibody 100 µg just before receiving the MTX plus adalimumab regimen.
Macaques and treatment
Adult captive-bred 3–5 year-old male cynomolgus macaques (Macaca fascicularis) were used.
Animals were treated with four 2 mg/kg intravenous adalimumab monthly injections which were similar to the schedule described in the Food and Drug Administration preclinical studies. Animals in the MTX group were injected prior to the adalimumab injection with 0.25 mg/kg MTX at day 0 and day 1.
Patients and treatment
Patients from the ABIRISK17 (Anti-Biopharmaceutical Immunization: prediction and analysis of clinical relevance to minimize the RISK), designed to prospectively identify risk factors of ADA against TNFi, were included in this study. Authorisation from each local ethics committee was obtained as well as patient’s informed consent. Patient’s underlying disease was rheumatoid arthritis (RA) (fulfilling 2010 American College of Rheumatology/European League Against Rheumatism criteria) or inflammatory bowel disease (IBD). Patients were considered treated with MTX when this treatment was concomitantly administered with the newly introduced TNFi. In ABIRISK patients, ADAs were screened at 1, 3, 6 and 12 months after TNFi therapy for rheumatologic cohorts (RA) and at 6 weeks and 3, 6 and 12 months after TNFi therapy for IBD cohort. Patients with at least one time-point with detectable ADA were considered immunised (ADA+).
All analyses of flow cytometry and animals were performed using non-parametric test (Mann-Whitney and Fisher tests). In the group comparisons of the human data a Welch’s unequal variances t-test was used. This test is well suited for when groups have unequal variance and as in this case unequal sample sizes.18 All authors had access to primary clinical trial data. Analyses were performed using GraphPad Prism V.7 software (La Jolla, USA).
Further methods are available in online supplementary materials.
A short course of MTX prevents ADA formation and maintains TNFi concentration for over a year in BAFFtg mice but not in cynomolgus monkeys or WT mice
BAFFtg mice treated with three different TNFi (adalimumab, etanercept and TN3) had an undetectable concentration of the drug due to ADA as soon as 8 weeks after initiation (figure 1). Intraperitoneal MTX administration at a dose of 5 mg/kg on days 0, 1 and 2 was able to maintain drug concentration of all tested TNFi, and to prevent the appearance of ADA in almost all animals. Strikingly, this tolerance was maintained for over a year with all three TNFi despite recurrent administration of TNFi without MTX. Thus, one course of three MTX injections was sufficient to prevent TNFi immunogenicity and induce long-term tolerance in BAFFtg mice. Following the same procedure as in BAFFtg mice, the rate of immunisation was low in WT mice (3/10 WT mice immunised) but MTX did not seem to have an effect (5/10 WT mice immunised with MTX) (online supplementary figure 2A,B).
To provide evidence for translation to patients we tried to reproduce these results in cynomolgus monkeys. In striking contrast, monthly adalimumab administration to cynomolgus monkeys with or without a 2-day MTX regimen showed no effect on immunogenicity. All animals showed a rapid appearance of ADA and drug concentrations were undetectable as soon as the second injection (online supplementary figure 2C,D). This suggests that high BAFF levels observed in BAFFtg mice compared to WT and macaques (online supplementary figure 3) are required for MTX to exert its effect
CD73 overexpression and increased adenosine production by B cells from BAFFtg mice
Since high BAFF levels seemed necessary for MTX-induced immune tolerance, we sought to compare the main molecular players involved in MTX mechanism of action between BAFF and WT mice. Between different metabolic pathways modulated by MTX, the adenosine synthesis pathway seemed particularly interesting in the context of tolerance induction studies. We thus analysed the expression of CD73, an ecto-5′nucleotidase involved in adenosine generation from AMP on a subset of peritoneal B1 cells (B220+ CD23 low). As shown in figure 2A, the expression of CD73 was markedly increased in the BAFFtg mice B1 subset compared with WT mice. Similar striking results were obtained in whole B220+ splenocytes in whom 21.7% of cells were CD73+ in BAFFtg mice versus 10.9% in WT mice (mean, p=0.0018) (figure 2B). Interestingly, the expression of the CD39 enzyme that acts upstream to CD73 in the adenosine synthesis cascade was also increased in B cells from BAFFtg mice compared to WT mice (figure 2C).
Functional characteristics of splenic B cells were next assessed: CD73 enzymatic activity on AMP exposure was quantified. Sorted splenic B cells were first exposed to low (20 µM) or high (200 µM) concentration of AMP for 2 hours, then adenosine and inosine productions were quantified. BAFFtg B cells were able to convert significantly more AMP into adenosine than WT mice (255.5 vs 120.7 ng/mL of adenosine, respectively, p<0.0001, at 20 µM AMP). This capacity was even increased at higher AMP concentrations (figure 2D). Cells without AMP exposure were used as controls and showed no difference in adenosine production between BAFFtg and WT mice. Since adenosine can be rapidly degraded into inosine, the sum of adenosine+inosine production was also assessed and consistently showed the same difference between BAFFtg and WT mice (figure 2E). Taken together, these data suggest that in the highly autoimmune context of BAFFtg mice, MTX could promote tolerance induction through stimulation of extracellular ATP/AMP degrading system and adenosine synthesis pathway in B cells.
BAFFtg mice have an increased pool of regulatory B cells and this population expands further with MTX treatment
Bregs are important players in the negative regulation of immune responses and might be involved in tolerance induction mechanisms. We thus compared the regulatory B cell population between BAFFtg and WT mice. Surprisingly, B10 cells defined by their ability to produce interleukin (IL)-10 were significantly increased among splenocytes in BAFFtg mice, despite their autoimmune phenotype as compared to WT mice (3.4% vs 1.16% of CD19+ cells, respectively, p=0005, figure 3A). When BAFFtg mice were subjected to MTX tolerisation, B10 cells were further significantly increased compared with untreated BAFFtg mice (6.3% vs 3.4%, respectively, p=0046, figure 3A). Moreover, B10 cells from BAFFtg mice showed higher CD73 expression: 41.1% vs 26% in WT of CD19+ IL10+ cells (p=0009, figure 3B). BAFFtg and WT mice were next compared for their percentage of Breg precursors among splenocytes. BAFFtg mice had a similar percentage of CD21hiCD24hi CD23+ cells compared to WT mice. But, when BAFFtg mice were tolerised with MTX against adalimumab, this percentage significantly increased from 33.4% to 52.5% (p=0.02, figure 3C).
In vivo CD73 inhibition prevents MTX-induced tolerance
Animals tolerised with the MTX regimen at the beginning of adalimumab treatment were treated or not with anti-CD73 antibodies. Anti-CD73 antibodies were administered twice a week either during the first week only (beginning before MTX and adalimumab administration) or continuously for 9 weeks. The two methods of anti-CD73 antibodies administration yielded the same results and were therefore pooled. After 9 weeks of treatment with one cycle of MTX and adalimumab, the mice treated with anti-CD73 antibodies lost their ability to induce tolerance to adalimumab. Anti-CD73-treated mice had a significantly lower serum concentration of adalimumab (0 vs 6.8 ng/mL, p=0.02, figure 4A). The number of immunised animals was also significantly higher when the anti-CD73 regimen was added to MTX and adalimumab in BAFFtg animals (p=0.03, figure 4B).
Translation to patients: in MTX-treated patients, BAFF is higher in patients without ADA
BAFF quantification was performed at baseline before the first administration of TNFi in the serum of 292 patients included in the ABIRISK consortium (135 RA and 157 IBD, characteristics in table 1) and these patients were prospectively followed for 18 months with regular quantification of ADA. In IBD, the percentage of patients with ADA was the same in patients cotreated or not with azathioprine/6-mercaptopurine (12% and 15%, respectively) and few patients were treated with MTX. In patients with RA, ADA occurred in 10% of the MTX-treated patients and in 25% of the non-MTX-treated patients (p<0.05, Fisher exact test). In MTX-treated patients, the mean serum level of BAFF was significantly lower (0.49 ng/mL) in ADA-positive patients than in patients without ADA (0.69 ng/mL, p=0.02, figure 5A). Conversely, in the non-MTX-treated patients, there was no significant difference in the serum level of BAFF between patients with or without ADA (p=0.88, figure 5B). Other baseline data (age, smoking, activity of the disease, cotreatment with steroids, C-reactive protein level) which could have influenced ADA occurrence were not different among RA MTX-treated patients between ADA+ and ADA− patients cotreated with MTX+TNFi (online supplementary table 1). Only three etanercept-treated patients had a transient low level of ADA only at month 1, with normal PK and thus without any clinical relevance since they are not neutralising antibodies. The removal of all patients treated with etanercept reinforced the association between low-serum BAFF level and ADA (0.68 ng/mL vs 0.46 ng/mL, p=0.0006, figure 5C). This removal did not affect the results in the non-MTX-treated patients (p=0.61, figure 5D). This confirms the interaction between MTX and BAFF to prevent ADA formation in TNFi-treated patients.
In this study, we showed that a single short regimen of MTX abolished immunogenic response to TNFi in BAFFtg mice but not in WT mice or in macaques. This suggests that high BAFF levels observed in BAFFtg mice compared to WT mice and macaques (online supplementary figure 3B) are required for MTX to exert its effect. Our data suggest an interaction between BAFF and MTX for this active tolerisation induction. We demonstrated that BAFFtg mice had constitutively elevated levels of CD39 and CD73 on B cells. Using a transgenic mouse model was the most reliable way to increase BAFF homogeneously but other strategies could be explored such as treatment of WT mice with exogenous BAFF or with BAFF-containing plasmids, but these approaches will be probably much less reproducible. This increased CD73 expression in B cells is required for the tolerisation action of MTX. MTX is in turn responsible for IL-10 producing Breg induction and increased adenosine production by B cells. This interaction between BAFF and MTX suggested in the BAFFtg mouse model was confirmed in a prospective human study. Hence, high BAFF serum level was protective against ADA formation to TNFi only in patients cotreated with MTX but not in patients on TNFi monotherapy.
The main action of BAFF is to stimulate activated B cells. This is confirmed by mouse models and by the efficacy of belimumab, a therapeutic anti-BAFF antibody in human lupus. However, a recent work shows that when type 1 interferon and IL-6 increase activation of immune cells, there is a concomitant activation of regulatory B cells enriched among Breg precursor cells.19 We describe a similar regulation in BAFFtg mice. Here, we showed that CD73 expression, a marker known to be expressed on regulatory T cells and on peritoneal B1 cells,20 21 was enhanced in BAFFtg mice. CD73 was upregulated in B1 cells from the peritoneum, where the first contact between MTX and TNFi occurs. This was also confirmed in the spleen which has been shown to be responsible for MTX-induced tolerance transfer in mice.7 Moreover, we found an increase of CD73 in B10 cells in BAFFtg mice especially when they were tolerised with MTX. Interestingly, it has been previously shown that BAFF stimulation of B cells increased IL-10 production in the supernatant22 and that B cells from BAFFtg mice could secrete more IL-10.23 Since B10 cells require in vitro activation to be studied, we sought to also explore Breg precursors as described previously.19Again, Breg precursors were increased in BAFFtg mice tolerised with MTX. Interestingly, a possible role of IL-10 in ADA prevention could also be suggested based on the association of anti-adalimumab ADA with IL-10 gene polymorphisms.24
MTX is the cornerstone of RA treatment but its mechanism of action remains to be elucidated. Through inhibition of AICAR transformylase, MTX increases the release of AMP outside the cell and in presence of CD73 the transformation into adenosine, a powerful immunosuppressive agent (online supplementary figure 1). CD73 is required for low-dose MTX-induced immunomodulatory effect observed in the air pouch model.5 6 CD39, the directly upstream ectoenzyme of CD73, has also been shown to influence clinical response to MTX in patients with RA.25 The interaction between MTX and BAFF can be explained, at least partly, by the increase of CD73 expression on B cells from BAFFtg mice. Indeed, CD73, present on BAFFtg mice B cells has a functional effect: exposition of these cells to AMP induced a higher level of adenosine synthesis. This was confirmed by treating mice with an anti-CD73 antibody which markedly decreased the ability of MTX to induce tolerisation against TNFi.
The MTX and BAFF interaction hypothesis is further confirmed by results obtained in patients from the ABIRISK cohort. For the first time in a clinical study, the serum level of BAFF was assessed to predict immunogenicity. The results obtained in mice were confirmed since only in patients cotreated with MTX, a high-serum BAFF level was correlated with less immunisation against TNFi. This effect was identical for different TNFi in two different inflammatory diseases. It was even higher if we excluded the patients treated with etanercept in whom only three presented with transient ADA, which probably has no clinical signification.
Besides RA treatment and prevention of ADA formation, MTX is also known to diminish immune response against vaccines. This effect is stronger than that induced by TNFi and similar to that induced by rituximab, an anti-CD20 B cell depleting agent.26 But the time frame between MTX administration and the vaccine exposure, required to inhibit the immune response, remains unclear. The mechanism of action for impairing the vaccine response could be the same as the mechanism we propose for inhibiting the ADA formation: activation of specific Bregs. In this study, we showed that MTX had to be present at the very beginning of antigen exposure and that this specific window was sufficient to induce long-term tolerance. Interestingly, a recent randomised controlled trial showed that, in order to impair the response to vaccines, MTX had to be present just before and at the same time of vaccination.27 We show in this report that a similar time frame is required to impair ADA formation against TNFi. Conversely to the action on RA activity which may take up to 3 months, this effect of MTX on ADA or vaccine requires the presence of MTX at the time of first exposure to the antigen.
In conclusion, just as IL-2 may activate effector or regulatory T cells depending on the dose or the environment, BAFF could be considered as a Janus cytokine able to balance between effector and regulatory B cells, depending on signals of the environment like MTX or other factors. In case of increased BAFF, using one single dose or one single cycle of MTX just before the first injection of a therapeutic monoclonal antibody may induce a specific tolerisation against this biologic drug for life. Another possible application would be to screen patients for BAFF levels and use MTX only in patients with elevated BAFF to prevent immunisation against TNFi and more generally to all immunogenic biologic drugs. Thus, it could be possible to prevent immunisation against any kind of therapeutic monoclonal antibodies in all domains of medicine by a very simple, cheap and safe procedure.
We thank Carole Nicco, Marine, Frédéric Batteux, Stephane Bloquet and Axel Perrot for their help in mouse housing. We thank the IDMIT infrastructure staff (Christophe Joubert, Benoit Delache, Sebastien Langlois and Jean-Marie Robert) for excellent technical assistance for macaque facilities. We also warmly thank all the clinicians who included patients with RA and IBD in the ABIRISK cohort.
Handling editor Josef S Smolen
Contributors SB, GN and XM designed the study, and supervised the experimental design and the data analysis. SB, GN, BL, PR, RK, RLG, APruvost, APaoletti, JP, KF, FM and AG made the experiments. PD and AHM performed statistical analysis. MA, AG, SHBA and MP gave substantial contributions from the ABIRISK cohort. SB and XM wrote the original draft of the manuscript. All authors reviewed and edited the manuscript, and provided final approval of the version published.
Funding The research leading to these results was supported by the Labex in Research on Medication and Therapeutic Innovation (LERMIT) (ANR10), the Innovative Medicines Initiative Joint Undertaking, ABIRISK (Anti-Biopharmaceutical Immunization Risk) project under grant agreement number 115303, the resources of which comprise financial contribution from the European Union’s Seventh Framework Program (FP7/2007-2013) and in-kind contributions from EFPIA companies, and the Fondation pour la Recherche Médicale DEQ20150934719: Sjögren’s syndrome and Autoimmunity-associated Lymphomas (SAIL). The IDMIT infrastructure is supported by the French government ‘Programme d’Investissements d’Avenir’ (PIA) under grant ANR-11-INBS-0008. SB was supported by two PhD grants from Société Française de Rhumatologie and INSERM.
Competing interests KF is an employee of GSK, PD is employed by SciCross and AHM is an employee of Sanofi.
Patient consent Not required.
Ethics approval The mouse study was approved by the regional Animal Care and Ethics Committee (Comité Régional d’Ethique sur l’Expérimentation Animale Île de France Sud, Fontenay-Aux-Roses, France; decision number 4281). Animal care and use was in accordance with the EU Directive 2010/63/EEU. The macaque study was approved by the regional Animal Care and Ethics Committee (Comité Régional d’Ethique sur l’Expérimentation Animale Île de France Sud, Fontenay-Aux-Roses, France; decision number A15_016). The CEA Institute was approved as compliant with ETS123 recommendations for animal breeding (European Union Directive 2010/63/EU, 22 September 2010) and with Standards for Human Care and Use of Laboratory Animals (Animal Welfare Assurance, OLAW No A5826-01). The study was also approved by the French department of education and research (MENESR; study number 2015070114504151v3) as defined in French law ‘décret 2013-118 from 2013 Feb 1st’. The human study was approved by the local IRB named ‘Paris Ile de France VII’ under number 13-048 and by the French agency for drugs (ANSM) under number 2013-A01268-37. The ABIRISK study was registered as study NCT02116504 by ClinicalTrials.gov.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.