Article Text

Download PDFPDF

Antidrug antibodies (ADAb) to tumour necrosis factor (TNF)-specific neutralising agents in chronic inflammatory diseases: a real issue, a clinical perspective
  1. Fabien B Vincent1,
  2. Eric F Morand2,
  3. Kim Murphy1,
  4. Fabienne Mackay1,
  5. Xavier Mariette3,
  6. Christian Marcelli4
  1. 1Department of Immunology, Monash University, Central Clinical School, Alfred Medical Research and Education Precinct (AMREP), Melbourne, Victoria, Australia
  2. 2Centre for Inflammatory Diseases, Southern Clinical School, Monash University, Monash Medical Centre, Clayton, Victoria, Australia
  3. 3Department of Rheumatology, Université Paris-Sud, AP-HP, Groupe Hospitalier Universitaire Paris-Sud, INSERM U1012, 78 rue du Général Leclerc, Le Kremlin Bicêtre, France
  4. 4Department of Rheumatology, University Hospital Centre of Caen, avenue de la Côte de Nacre, Caen, France
  1. Correspondence to Dr Fabien B Vincent, B lymphocyte, BAFF and Autoimmunity Laboratory, Department of Immunology, Monash University, Level 2, AMREP Building, 89 Commercial Road, Melbourne, Victoria 3004, Australia; vincent.fabien{at}


The introduction of biologics, especially tumour necrosis factor (TNF) inhibitors, has revolutionized the management of chronic inflammatory diseases. However, at least one third of patients with these diseases, receiving TNF inhibitors either do not respond to treatment, or lose initial responsiveness. For a significant proportion, improvement of clinical response is achieved after switching to another anti-TNF drug, suggesting a basis for failure unrelated to the therapeutic target itself. A likely explanation for this is immunogenicity, as all biologics are potentially immunogenic, and the resulting anti-drug antibodies (ADAb) can theoretically decrease the efficacy of biologics and/or induce adverse events. Indeed, in these chronic inflammatory diseases, many studies have now established correlations between ADAb formation, low serum drug levels, and the failure or loss of response to anti-TNF antibodies. This article will review key findings related to ADAb, and propose a model wherein monitoring of drug levels and ADAb may be a predictive tool leading to a better choice of biologics. Such an approach could improve chronic inflammatory disease management toward a personalized and more cost-effective approach.

  • Anti-TNF
  • Rheumatoid Arthritis
  • Spondyloarthritis
  • B Cells
  • Psoriatic Arthritis

Statistics from

Request Permissions

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.


In the past 2 decades, biological therapies, especially tumour necrosis factor (TNF) inhibitors, have revolutionised the management of chronic inflammatory diseases, including rheumatoid arthritis (RA), ankylosing spondylitis (AS), psoriatic arthritis (PsA), psoriasis and Crohn's disease (CD). Three TNF inhibitors, infliximab (IFX), adalimumab (ADA) and etanercept (ETN), have dominated management of these diseases. Two additional TNF inhibitors have become available more recently, namely, certolizumab pegol (CZP) and golimumab (GLM). Despite an acceptable responder rate of 60%–70% across these diseases,1–19 there remains a substantial proportion of patients for whom anti-TNF agents fail to demonstrate efficacy (primary failure) or induce significant side effects.1 ,2 ,5 ,7 ,8 ,18 ,20–27 Patients may also lose TNF inhibitor effectiveness over time despite an initial good response (secondary failure).1 ,7 ,8 ,18 ,20 ,24 ,28–33 In the absence of response to a first anti-TNF, alternative strategies are routinely used, such as a change in the current TNF inhibitor dose, shortening infusion intervals (notably for IFX), switching to another anti-TNF or switching to a drug with a different mechanism of action. There is at present no accepted rationale for the approach to primary and secondary failure of anti-TNF drugs, and this presents an important issue from both a clinical and economic perspective.

Any biologics, whether entirely of human origin, chimeric or ‘humanised’, can cause an immune response, leading to the formation of antidrug antibodies (ADAb). ADAb are also referred to as human antichimeric antibodies or human antihuman antibodies, depending on the nature of the drug. The generation of ADAb is increasingly recognised as a mechanism explaining the failure of anti-TNF drugs in chronic inflammatory diseases. This lack of clinical response in patients with ADAb may be explained by immune complex formation between TNF inhibitors and ADAb suppressing the drug and restricting its therapeutic role. This may be related to an increase of its clearance due to the presence of immune complexes, which leads to lowering the serum drug levels34–36 or to a direct neutralisation of the biological interfering with the fixation of the latter on TNF.37 The neutralisation of the functional part of the drug by ADAb, leading to the impossibility of fixation of the drug to TNF and thereby inhibiting its biological activity, has recently been also demonstrated.37 Thus, the presence of ADAb indirectly influences disease activity through limiting both the pharmacokinetics and pharmacodynamics of the drug. In addition to these considerations, innate inter-individual variability of pharmacokinetics means that the use of a standard dose probably results in unnecessary overtreatment of some patients to overcome the negative effect of immunogenicity in a subset. Although ADAb are associated with loss of response, ADAb status does not preclude clinical efficacy, as some patients with ADAb are good responders in these diseases.

In this article, we have performed a systematic review of the literature related to TNF-specific therapies from 1994 to present with a particular focus on ADAb. We also offer a model which could lead to more effective use of TNF neutralising agents.

Chronic inflammatory diseases

Chronic inflammatory diseases are a diverse range of conditions characterised by local or systemic inflammation. RA is characterised by synovial inflammation and symmetric arthritis that leads to progressive erosion of cartilage and bone.38 AS39 and PsA40 are spondyloarthritides involving the axial skeleton, entheses and/or peripheral joints, whose natural evolution leads to disabling joint deformation, bone erosion and ankylosis. Psoriasis is an immune-mediated, inflammatory papulosquamous skin and nail disease of which plaque psoriasis is the most common form.41 CD is a chronic, transmural inflammatory disorder which may involve any part of the gastrointestinal tract from mouth to anus, but is mostly confined to the ileum, caecum and the colon.42 These five inflammatory diseases differ greatly in their clinical presentation, but all have been shown to respond to treatments that inhibit the action of TNF, that is, the biological TNF inhibitors.

TNF and the TNF inhibitors

TNF can be produced by several cell types, most notably activated T cells and macrophages.43 TNF is produced as a transmembrane protein (tmTNF) and associates as a homotrimer on the cell surface, where it can be cleaved by a metalloproteinase (TNF-α convertase) to a soluble form (sTNF). Both sTNF and tmTNF bind TNF receptor 1 or 2, resulting in the activation of a series of intracellular signalling events, which result in the transcriptional activation of multiple pro-inflammatory genes.44 ADA, IFX, CZP, GLM and ETN block both forms of TNF, by inhibiting their binding to TNF receptors 1 and 2.45–48 When these drugs bind tmTNF, all except CZP, which does not contain a fragment crystallisable (Fc) domain, they have the potential to induce effects mediated by the Fc portion, such as complement-dependent cytotoxicity or antibody-dependent cell-mediated cytotoxicity.45 ,46 ,49 Whether TNF inhibition produces its beneficial effects by binding to systemic or tissue TNF remains unclear. Regardless, TNF inhibitors are efficacious in the chronic inflammatory diseases, lowering disease activity or inducing remission, inhibiting structural joint damage in some and improving symptoms, physical function and health-related quality of life.1–19 ,50 ,51 The effectiveness of the highly specific inhibitors of TNF demonstrates unambiguously that TNF plays a central role in the pathogenesis of RA, AS, PsA, psoriasis and CD.44 ,52–55

Monoclonal anti-TNF antibodies

ADA (Humira, Abbott Laboratories, North Chicago, Illinois, USA) and GLM (Simponi; Centocor OrthoBiotech Inc., Pennsylvania, USA) are human type immunoglobulin G1 (IgG1) monoclonal antibodies (mAb) targeting both sTNF and tmTNF. IFX (Remicade, Centocor Ortho Biotech Inc.) is a chimeric human-murine mAb, which is comprised of human IgG1 constant regions and murine variable regions,56 whereas ADA and GLM both have human IgG1 constant and variable regions (fully humanised mAb).

IFX binds both sTNF and tmTNF with high affinity, forming a stable complex that blocks the association of TNF with its receptor.57 IFX has a half life of 9.5 days,58 which is lower than the 21 days half life of endogenous human IgG1.59 IFX and ADA bind cell surface tmTNF and have the potential to induce cell lysis via initiation of complement-dependent cytotoxicity and antibody-dependent cell-mediated cytotoxicity. Efficacy and safety of IFX have been demonstrated in patients with RA,7 AS,5 PsA,10 psoriasis3 and CD.1 ADA is approved for treatment of RA,8 PsA,18 psoriasis,17 CD16 and AS.6

Similar to ADA, GLM is a fully human bivalent IgG1 κ mAb targeting and neutralising TNF by binding to both sTNF and tmTNF. The constant regions of the heavy and light chains of GLM are identical to the corresponding human constant regions of IFX, but unlike IFX the variable regions of GLM have fully human sequences. Its actions are comparable with other anti-TNF with Fc components (IFX, ADA, ETN). Its efficacy and safety profile was demonstrated in patients with RA,14 PsA13 and AS.12

CZP (Cimzia, UCB, Brussels, Belgium) is a recombinant polyethyleneglycolylated, fragment antigen binding (Fab) fragment of a high-affinity human anti-TNF mAb. CZP is devoid of the Fc portion of IgG, unlike IFX, ADA and ETN that all contain an IgG Fc region. Instead, the conjugation of a polyethyleneglycol chain to the Fab fragment increases the half life of the product to a mean of 14 days.60 The efficacy and safety of CZP have been demonstrated in RA11 and CD.2

Soluble TNF receptor

ETN (Enbrel, Amgen Inc., California, USA) is the only soluble TNF receptor currently in clinical use. ETN is a fully human protein comprised of the Fc fragment of human IgG1 fused to two extracellular domains of human TNFR2/p75.34 This dimeric fusion protein binds and neutralises s/tmTNF and lymphotoxin-α.61 ETN is approved for the treatment of RA,9 PsA,15 psoriasis19 and AS4 but lacks efficacy in CD.

Antidrug antibodies

Detection of ADAb

A number of methods can be used to detect and characterise ADAb. Standard direct and indirect ELISAs, while capable of high throughput and relatively inexpensive, may be more prone to false-positive results and non-specific binding. Improved assays for detecting ADAb include the two-site (or bridging) ELISA and radioimmunoassay (RIA) antigen binding test (ABT). The two-site ELISA involves binding the drug of interest to a solid support matrix, adding patient sera and then detecting using a biotinylated form of the drug of interest. This type of assay is both highly specific and sensitive, although it may favour the detection of IgM antibodies and may not detect monovalent IgG4 antibodies.62 The ABT involves immobilising antibodies in patient serum to a sepharose-bound reagent and adding a radiolabelled drug that is captured if drug-specific antibodies are present. The ABT can detect clinically relevant IgG1 and IgG4 antibodies and has low background, but the use of radioactivity is a disadvantage.35

The detection of ADAb is confounded by the presence in the serum of high concentrations of rheumatoid factor (RF) and by the presence of the drug itself. RF is IgM antibodies that forms complexes with the Fc portion of IgG that can thereby mask the epitopes bound in these assays. This may explain the interference caused by RF with immunoassays, although this can be overcome by treatment of samples with RF neutralisation.35 The presence of drug in a sample that contains ADAb can result in immune complex formation that both increases the clearance of the drug and associated ADAb in vivo and hinders in vitro detection.63 It is likely that ADAb are underestimated as they may be detected only if the molar concentration of antibody exceeds that of the drug in the serum.28 ,62 When screening for ADAb both dosing and timing of drug administration should be considered in the context of drug pharmacokinetics.

New methods designed to overcome these limitations are being developed, although none is accepted clinically as yet. An alternate radioimmunoassay that detects bound ADAb using a radiolabelled F(ab′)2 region of IFX or ADA overcomes the problems associated with RF and detects antibodies to the now exposed hinge region of the F(ab′)2 molecule.64 The use of anti-λ detection antibodies in an ELISA assay was trialled,65 although this proved little better than the standard ELISA. To overcome the problems associated with immune complexes, disassociation protocols have been described. An ‘acid treatment’ disassociated ADAb–ADA complexes prior to antibody detection via an ABT; this assay demonstrated ADAb in 21 of 30 RA patients receiving ADA while the standard ABT detected these antibodies in only five patients.66 Imaeda et al disassociated immune complexes via affinity chromatography before detecting anti-IFX antibodies via a two-site ELISA; anti-IFX antibodies were detected in 16 of 58 CD patients compared with two patients detected via a standard two-site ELISA.67 The benefit of this assay is the lack of radioactivity and hence the ease of performing it in a conventional laboratory, although the multi-step process may be undesirable in the clinical setting.

Of note, the prevalence of ADAb has been found to vary widely depending on the type of assay used. Unfortunately, much of the existing data come from clinical trials of products using assays developed by the manufacturers, without complete information about the methods used (tables 15).

Table 1

Incidence, clinical and biological associations of anti-IFX antibodies

Table 2

Incidence, clinical and biological associations of anti-ADA antibodies

Table 3

Incidence, clinical and biological associations of anti-CZP antibodies

Table 4

Incidence, clinical and biological associations of anti-GLM antibodies

Table 5

Incidence, clinical and biological associations of anti-ETN antibodies

Anti-IFX antibodies

As it is the only currently marketed chimeric anti-TNF mAb, IFX is at least theoretically more immunogenic than other products. The reported rate of development of ADAb to IFX in clinical studies ranges from 6% to 61%, as summarised in table 1.

The presence of ADAb toward IFX is generally associated with reduced serum IFX concentration, with decreased clinical response to IFX and with increased adverse events (table 1). Of note, no studies report associations between serum IFX concentration and ADAb in PsA or CD, and one study reported lack of association between the presence of ADAb and clinical efficacy in CD.83 Interestingly, Kavanaugh et al showed that the prevalence of ADAb in patients with PsA was similar regardless of whether patients were receiving an escalating dose of IFX or a stable dose.86 ADAb can be detectable early after IFX treatment initiation, being detectable in half a group of RA and spondylo-arthritis (SpA) patients within 3.7 months of treatment.71 Interestingly, in RA, low serum IFX concentration even 1.5 months after initiation of the drug was significantly correlated with ADAb formation and predicted subsequent treatment failure.32 In contrast, in AS it has been reported that anti-IFX ADAb are detected most often after 6 months of therapy, but can arise even 2 years after commencement.76 This difference may be explained by a lower IFX dose (3 mg/kg in RA vs 5 mg/kg in SpA). In the study of Gottlieb et al, the rate of anti-IFX ADAb in patients with psoriasis was higher among those on 3 mg/kg IFX than those receiving 5 mg/kg.89 In CD, scheduled IFX therapy was associated with lower frequency of anti-IFX ADAb compared with episodic treatment.123 Of note, anti-IFX ADAb have been detected in patients with CD up to 4 years after IFX discontinuation,124 but this could be due to the limitations of some assays if the drug is still present in the serum.

Anti-ADA antibodies

Although ADA is a fully human mAb, it is immunogenic. The frequency of detection of anti-ADA ADAb varies from 0.04% to 87%, as summarised in table 2. Of note, the study reporting the lowest frequency of ADAb incidence had a duration of only 4 weeks.100 The presence of anti-ADA Abs has been associated with low or undetectable serum trough ADA levels and reduced response to ADA.17 ,22 ,30 ,75 ,91 ,92 ,95–99 ,101 Papp et al showed an association between the presence of anti-ADA Abs and a reduced response following treatment discontinuation and relapse in patients with psoriasis.98 In the Karmiris et al study, early and late discontinuations were associated with low ADA level, although there was no direct relationship between serum ADA level and short-term efficacy in CD.99 Of note, during long-term follow-up in RA, patients with anti-ADA ADAb achieved minimal disease activity or remission significantly less often and exhibited more often treatment failure, compared with those without ADAb.91 Immunisation against ADA occurs early after treatment induction, as anti-ADA Abs were detected in 67% of anti-ADA ADAb-positive patients within the first 28 weeks in RA.91

An association between anti-ADA ADAb and adverse events has not been widely documented, with negative studies in RA,92 psoriasis17 ,98 and CD.99 However, Korswagen et al reported a higher incidence of venous and arterial thromboembolic events in RA patients with anti-ADA Abs.125

Anti-CZP antibodies

The frequency of detection of anti-CZP ADAb ranges from 3% to 25%, as summarised in table 3. The highest incidence of anti-CZP ADAb was reported by Reich et al in psoriasis in a study in which patients were retreated after a period off medication.105 This effect of drug re-exposure was also reported in CD, in which there was a higher risk of anti-CZP ADAb in patients who had interrupted therapy than in those with continuous therapy.106

In common with the antimonoclonal ADAb discussed above, anti-CZP ADAb have been associated with a modest reduction in therapeutic response in RA,103 whereas no such association was reported in psoriasis.105 Anti-CZP ADAb were associated with low serum trough CZP levels in one study in CD.106 In most studies, the frequency of ADAb-positive patients was too low to identify any Ab effect on response. The association of ADAb with adverse events was analysed in only one study in CD in which no association with anti-CZP ADAb was observed.106

Anti-GLM antibodies

The reported frequency of anti-GLM ADAb ranges from 0% to 7%, as summarised in table 4. Most studies of GLM ADAb were short-term and reported ADAb frequency too low to analyse any association among ADAb and clinical response, trough serum GLM level, and adverse events. Two studies in RA and PsA showed no association between ADAb and clinical response,13 ,14 whereas another reported a non-significant association in RA.115 An association between ADAb formation and lowered trough serum GLM levels has been reported in RA14 ,111 and in AS.12 There have been only two long-term studies in AS116 and PsA,51 and no association among ADAb formation and clinical response, serum trough GLM level, and adverse events was reported. In the GO-RAISE study,12 median serum GLM concentrations were greater in AS responders than non-responders. However, no association between serum GLM concentration and response to drug was found after dose escalation in non-responders. In the 2-year results of this study, of the patients with anti-GLM Abs (4.1%), 50% became ADAb-negative beyond 24 weeks of treatment.116 No studies have reported an association of GLM ADAb and the occurrence of AEs.13 ,14 ,115

Anti-ETN antibodies

The reported data suggest that ETN is less immunogenic than anti-TNF mAbs. ETN is a dimeric fusion protein comprised of two TNF receptors linked to the Fc portion of IgG1. Only the junction between these domains constitutes a non-human sequence and may therefore be immunogenic. In contrast, even wholly human mAb may have multiple epitopes within the Fab variable region against which an immune response can be directed. The frequency of detection of anti-ETN Abs reported ranges from 0% and 18%, as summarised in table 5. In each case where detectable, anti-ETN ADAb were non-neutralising. To date, no association has been reported among the formation of anti-ETN ADAb and trough serum ETN levels, clinical response, or adverse events, even in long-term studies of over 3 years.117 Jamnitski et al recently reported a significant positive association between serum trough ETN level and clinical response to ETN in RA, without any anti-ETN Abs detected.63

Effect of associated immunosuppressive agents on ADAb

The concomitant use of immunosuppressive agents such as methotrexate (MTX), azathioprine (AZA) or 6-mercaptopurine (6-MP) with TNF inhibitors has been shown to reduce the frequency of ADAb formation, as summarised in tables 15. There are no published data available regarding the ability of leflunomide to prevent ADAb formation in these diseases. Co-prescription of MTX clearly reduces ADAb formation directed against monoclonal TNF inhibitors.8 ,28 ,30 ,72 ,75 ,80 ,82–84 The effect of MTX in preventing ADAb toward IFX has been described in RA,32 ,71 ,72 CD,78 ,80–84 PsA86 and in SpA,71 but not in psoriasis. Anti-ADA ADAb are also reduced with MTX use in RA30 ,95 ,91 ,126 This observation was also reported in psoriasis97 and in CD,16 although no statistical analysis was performed. Interestingly, the negative effect of MTX on anti-ADA Abs was reported in one study to be dose-dependent.126 The development of ADAb to CZP is also reduced in association with the use of MTX, AZA and 6-MP, but this has only been reported in CD.2 ,108 Concomitant use of MTX was associated with lower incidence of anti-GLM Abs in RA,114113 PsA13 ,51 and AS.12 No study has analysed associations between the use of MTX and anti-ETN ADAb formation since the frequency of anti-ETN Abs is very low. An association between ADAb incidence and use of corticosteroids has been reported in CD in which intravenous hydrocortisone premedication significantly reduced anti-IFX Abs concentrations, but not anti-IFX Ab formation.81

An important question is the possible usefulness of MTX in AS in decreasing immunogenicity, as it is not currently recommended in this condition. The concomitant use of MTX has been shown to reduce ADAb formation in SpA patients receiving IFX in two retrospective studies which did not differentiate AS within SpA.71 ,76 One randomised, placebo-controlled, double-blind study in AS patients receiving GLM showed this association but did not provide any statistical analysis, probably due to the low incidence of ADAb.12 In a small, randomised controlled study having included around 60 patients in each arm and treated with IFX in an ‘on demand’ scheme, the adjunction of MTX did not change statistically the percentage of responders. Nevertheless, there was a numerical difference of response between MTX-positive and MTX-negative patients.127 Based on these data, it is not possible to recommend concomitant use of MTX in AS, but the use of MTX just to avoid immunogenicity has to remain open. Further prospective and long-term studies are needed to definitively characterise the potential anti-immunogenic effect of MTX in AS.

Why are ADAb not observed in all patients receiving TNF inhibitors?

Variability among studies

Variation of ADAb incidence among studies, even when studying the same TNF inhibitor in the same disease, may be explained in part by the use of different assays, as outlined earlier. Moreover, the cut-off above which the patient is considered to have ADAb may vary depending on the study, making it difficult to compare studies. Advancing knowledge in this area, which may become even more important with the introduction of potentially more immunogenic biosimilar anti-TNF drugs, requires quality control and standardisation of assays such as is done for routine clinical lab testing. This is the objective of a European Union Innovative Medical Initiative called ABIRISK, which commenced in 2012.

Another explanation for variation between studies might be the timing of sampling, in that the time between the last injection of TNF inhibitor and the assay sample will affect the concentration of drug in the sample and thus the assay sensitivity. The development of ADAb also appears to vary between diseases. A higher incidence of ADAb in AS than in RA may relate to less frequent use of MTX in AS, as co-prescription of MTX reduces ADAb formation directed against monoclonal TNF inhibitors.8 ,28 ,30 ,83 ,75 ,84 ,72 ,80 ,82 Variation of ADAb frequency may also be influenced by the variability of drug doses. Comparison of two studies in RA in which patients received IFX revealed a fivefold lower frequency of ADAb in patients who received higher doses.28 ,128 Similarly, in CD, Hanauer et al showed that maintenance IFX therapy was associated with a lower frequency of ADAb than episodic treatment.83

Biological variability

A key question is why the formation of ADAb is not observed in all patients receiving TNF inhibitors even under identical conditions in the same study. Patient genetics may be a factor in ADAb formation.95 ,129 For example, anti-ADA Abs formation appears to be related to polymorphisms in the gene encoding IL-10, which may regulate immune activation during ADAb formation.130 Difference in baseline characteristics may also be considered, even if they are still not understood. In one study, patients with anti-ADA ADAb exhibited higher C-reactive protein levels, longer disease duration and more erosive disease at baseline than those without ADAb.91 Although the concomitant use of immunosuppressants is reported to reduce the frequency of ADAb formation, and it is tempting to hypothesise direct effects on T or B cell function leading to reduced antibody production, the mechanism of this relationship is not yet demonstrated. Interestingly, in RA patients after switching from IFX to the B cell depleting antibody rituximab, anti-IFX Abs production was not stopped by a single course of rituximab.131

Another area of variation among patients is the dose and immunosuppressant effect of the anti-TNF agent itself. Low serum TNF inhibitor concentrations are generally associated with presence of ADAb among non-responders, although this could be an effect of ADAb rather than the cause. Hypothetically, lower doses of anti-TNF could permit ADAb development and, conversely, higher dosages of anti-TNF result in suppression of this response. One study showed that high doses of IFX were associated with lower frequency of ADAb.72 These observations raise the possibility of increased ADAb formation in response to spacing of anti-TNF inhibitor doses when patients reach clinical remission, increasing anti-TNF clearance, and thus increasing the risk of secondary treatment failure. Supporting this hypothesis, systematic treatment was found to produce a better response than an on-demand regimen of IFX in AS.132 Additionally, Gottlieb et al showed that the incidence of infusion reactions after an extended period without treatment was threefold higher among patients with anti-IFX ADAb compared with those without ADAb.89 However, no direct evidence has been published so far demonstrating that spacing of TNF inhibitors leads to ADAb production and thereby enhances the risk of secondary failure. If such a relationship were shown, it could be useful to have data to guide the monitoring of serum TNF inhibitor and ADAb concentrations so as to determine the optimal spacing of injections, control the occurrence of immunogenicity and avoid secondary failure.

Clinical responses to the development of ADAB

The definition of failure of anti-TNF drugs is clinical. Notwithstanding the effect of ADAb on drug levels and thereby efficacy, it remains the case that there are many non-responders for whom no evidence of immunogenicity is demonstrated.28 ,30 A rationale exists for patients who experience primary failure to be switched to a drug with a different mechanism of action.133 ,134 It is currently not routine clinical practice to investigate whether ADAb contribute to primary failure. In the case of secondary failure, assessment of the presence of ADAb and their effects on drug concentrations is also not routine, with management strategies instead including dose escalation, switching to an alternative anti-TNF agent or switching to a drug with an alternative mechanism of action.

Anti-TNF intensification: a matter for debate

Adjustments in drug dosage or frequency of infusions, especially for IFX, is often proposed to patients who exhibit primary or secondary failure.33 ,135 Some studies have shown that therapeutic intensification significantly improves responses,136 whereas other showed little or no effect.25 ,50 ,91 ,137–139 Most of these studies did not monitor serum drug concentration or ADAb. In patients with SpA receiving IFX, clinical improvement in response to shortening infusion intervals occurred significantly more often in patients without ADAb.76 Of note, in this study, shortening infusion intervals led to reduced ADAb levels in 50% of patients.76 In another study, patients with RA who exhibited anti-IFX Abs and low clinical response underwent dose escalation or shortening of infusion intervals.68 There were two kinds of response to this dose escalation. In one group, in patients in whom ADAb did not disappear, some infusion reactions were observed in response to escalation. In contrast, in patients in whom ADAb became undetectable after dose escalation, clinical improvement and higher serum IFX levels were observed. In this second group, ADAb became detectable again and were associated with reduced clinical responses when IFX dose was subsequently decreased. In another study, in contrast, increasing IFX dose did not lead to increased serum IFX concentrations or decreased ADAb levels or clinical improvement.77 Interestingly, among RA patients on ADA who experienced a decrease of anti-ADA Abs to undetectable levels after increasing drug dose, minimal disease activity was not achieved.91

To switch or not to switch?

Switching from one TNF inhibitor to another can be associated with efficacy gains in inadequate responders,23 ,90 but anti-TNF naive RA patients showed better clinical responses than anti-TNF switchers.95 Patients who stopped their first anti-TNF agent due to adverse effects were more likely to respond than those who experienced inefficacy.140 Measurement of ADAb before switching may be of value in clinical decision making.95 ,133 In RA, anti-TNF naive patients and switchers with ADAb against their first anti-TNF showed similar responses to a second TNF inhibitor which were superior to the responses of switchers without detectable ADAb,95 ,133 suggesting that anti-TNF failures in the absence of ADAb may have a target-related reason for anti-TNF failure, and thus be better suited to a drug with a different mechanism of action, while those with ADAb may have a drug-related failure. Interestingly, the tendency to develop ADAb may endure despite a change of drug. In one study, patients with anti-IFX Abs more often developed anti-ADA Abs compared with anti-TNF naive patients when treated with ADA.95 West et al reported that when ADA was used as a second anti-TNF after IFX, serum anti-IFX concentrations were significantly increased among ADA non-responders.101

Monitoring serum drug and ADAb levels: toward a personalised therapy approach

Current clinical practice, in monitoring response, dose regimens and planning decision making after primary or secondary failure of anti-TNF therapy, does not include routine monitoring of ADAb or drug concentrations in patient serum. Thus, therapeutic intensification and/or drug switching is carried out without the knowledge of factors that may explain the failure. Knowledge of the predictive power of serum drug and ADAb concentration for responses to dose escalation, alternative TNF inhibitors or drugs with different mechanisms of action could radically change the way patients on anti-TNF drugs are managed.

Given the high cost of TNF inhibitors, and the consequent cost associated with the use of such drugs when destined to fail, the potential benefit of monitoring drug concentrations and ADAb in patients on anti-TNF therapy could be economically justified. Afif et al showed the utility of measurement of ADAb and serum TNF inhibitor concentrations in patients with CD receiving IFX.141 In patients with ADAb, changing to another TNF inhibitor was associated with a response in 92% of patients, whereas increasing the dose of IFX resulted in a response in only 17% of cases. Conversely, patients with subtherapeutic IFX concentrations were more likely to respond to dose escalation than to drug switching.141 In contrast, in AS, a recent study showed that the measurement of serum IFX concentration did not improve the control of disease activity.142 Mulleman et al143 demonstrated in a cohort of 24 anti-TNF non-responders suffering from RA that when serum anti-TNF levels were taken into account, therapeutic decisions were changed for almost half of the patients and that non-responders patients with low serum IFX concentrations benefited from dose escalation, whereas patients with high IFX concentrations responded to switching to another biologic. In a study involving 272 patients with RA receiving ADA over 3 years, the use of serum ADA and ADAb concentration monitoring in combination with disease activity score decreased the total cost of biological drugs by €643 275, while testing costs amounted to only €12 116.144 Such approaches could allow for more precise therapeutic decision making in anti-TNF non-responders and be highly cost-effective.


The literature provides strong arguments for the use of ADAb and drug concentration measurement as decision making tools in clinical practice. Despite this, routine screening for serum anti-TNF agents and ADAb is not regularly performed. We recommended that an international consensus be reached for further studies on the impact of ADAb and drug concentration measurement in inflammatory disease management. Detection method standardisation, definition of cut-off values for defining the presence of ADAb and definition of the therapeutic range of concentration for each anti-TNF agent are necessary for this field to advance. Assuming such standardisation can be reached, we propose a model for the clinical application of these measurements (figure 1). The economic cost of the use of expensive drugs in patients unlikely to benefit from them, as well as that of tissue damage occurring while patients are using ineffective treatments, argues in favour of significant research investment in the near term.

Figure 1

Decision tree algorithm for monitoring serum anti-TNF drug and ADAb concentrations in patients with primary or secondary failure of anti-TNF therapy. ADAb, antidrug antibodies; TNF, tumour necrosis factor.



  • Correction notice This article has been corrected since it was published Online First. The paragraph ‘sTNF receptor’ has been updated to read ‘Soluble TNF receptor’. The paragraph ‘Anticertolizumab antibodies’ has been updated to read ‘Anti-CZP antibodies’.

  • Contributors FV drafted the paper. FV, EM, KM, FM, XM and CM revised the draft paper. All authors gave final approval of the version to be published. FV accepts full responsibility for the work, had access to the data and controlled the decision to publish the manuscript. He is the guarantor.

  • Funding None.

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.