Size of the data marker corresponds to the relative weight assigned in the pooled analysis using random-effects analysis. A, Association with European League Against Rheumatism (EULAR) response at 6 months or less. B, Association with EULAR response at 6 months or more. C, Association with Assessment in AS International Working Group response. D, Association with clinical response in inflammatory bowel disease. AAB– indicates seronegative patients; AAB+, seropositive patients; ACT, Active Ulcerative Colitis Trial; ADA, adalimumab; IFX, infliximab; OR, odds ratio.
Size of the data marker corresponds to the relative weight assigned in the pooled analysis using random-effects analysis. AAB– indicates seronegative patients; AAB+, seropositive patients; ACT, Active Ulcerative Colitis Trial; OBS, observational study; OR, odds ratio.
Size of the data marker corresponds to the relative weight assigned in the pooled analysis using random-effects analysis. ACT indicates Active Ulcerative Colitis Trial; ADA, adalimumab; CTZ, certolizumab; INF, infliximab; OR, odds ratio.
eTable 1. Excluded articles and reasons for exclusion
eTable 2. Individual results of the influence of AAB in efficacy
eTable 3. Individual results of the influence of AAB in hypersensitivity reactions
eTable 4. Individual results of effect of concomitant treatment in development of AAB
eTable 5. Individual results of the influence of AAB on the concentration of biological drugs
eTable 6. Efficacy measures included in the review
eFigure. Results of literature search
Maneiro JR, Salgado E, Gomez-Reino JJ. Immunogenicity of Monoclonal Antibodies Against Tumor Necrosis Factor Used in Chronic Immune-Mediated Inflammatory ConditionsSystematic Review and Meta-analysis. JAMA Intern Med. 2013;173(15):1416-1428. doi:10.1001/jamainternmed.2013.7430
Copyright 2013 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.
Knowledge of the immunogenicity of biologic agents may be helpful for the development of strategies for treatment of chronic immune-mediated inflammatory diseases.
To summarize the influence of antibodies against biologic agents (AABs [seropositivity]) on efficacy and safety in immune-mediated inflammatory diseases.
MEDLINE, EMBASE, Cochrane Library, and the Web of Knowledge were searched for articles published in English, Spanish, French, Italian, or Portuguese between 2000 and March 2012. The search strategy focused on synonyms of diseases, immunogenicity, and biologic agents. Abstracts from 2001 to 2011 of the European League Against Rheumatism and American College of Rheumatology congresses were also included.
The selection criteria were (1) observational or interventional studies in rheumatoid arthritis, juvenile idiopathic arthritis, inflammatory bowel disease, spondyloarthritis, and psoriasis; (2) studies including patients who received biologic agents; and (3) studies collecting data on AABs.
Data Extraction and Synthesis
Data collected included publication details, study design, characteristics of patients and treatments, presence of antibodies, and definition of response.
Main Outcomes and Measures
The primary end point was the association of AABs with response to treatment. Secondary end points were the association of AABs with safety, the association of AABs with concentration of the drug, and the influence of use of concomitant immunosuppressive therapy in the formation of AABs.
The search captured 10 728 articles and abstracts. By hand and reverse search, 31 articles were additionally included. After evaluation of the full reports, 60 references were selected. They included 59 studies of anti–tumor necrosis factor monoclonal antibodies: 1 with etanercept, 2 with rituximab, and 2 with abatacept. In rheumatoid arthritis but not in inflammatory bowel disease or spondyloarthritis, seropositive patients presented worse clinical response at 6 months or less (odds ratio [OR], 0.03; 95% CI, 0.01-0.21), and at 6 months or more (0.03; 0.00-0.30) by meta-analysis. In rheumatoid arthritis, discontinuation of the biologic agent for all reasons was more common in seropositive patients (OR, 3.53; 95% CI, 1.60-7.82). In all conditions, seropositive patients had a higher risk of hypersensitivity reactions (OR, 3.97; 95% CI, 2.36-6.67). Overall, concomitant treatment with disease-modifying antirheumatic drugs, including azathioprine, decreased the risk of seropositivity (OR, 0.32; 95% CI, 0.25-0.42).
Conclusions and Relevance
Presence of antibodies against anti–tumor necrosis factor monoclonal antibodies confers a risk of discontinuation of treatment in rheumatoid arthritis and a risk of development of hypersensitivity reactions in all immune-mediated inflammatory diseases. The combined use of anti–tumor necrosis factor monoclonal antibodies and disease-modifying antirheumatic drugs reduces the development of antibodies and subsequent risks. Information on other biologic agents is fragmentary.
The development of biological therapies signified an advance in the treatment of immune-mediated chronic inflammatory diseases (IMIDs).1- 3 Nevertheless, several patients initially responding to a biologic agent develop acquired drug resistance or gradual drug failure, and some have to discontinue treatment with the biologic agent because of adverse events. In recent years, the immunogenicity related to these drugs, especially with monoclonal antibody as the cause of these unfavorable outcomes, has been studied. Results suggested that antibodies against biological therapies (AABs) affect the efficacy4- 8 and safety of the biologic agents.8- 10 Primarily, immunogenicity against anti–tumor necrosis factor (TNF) monoclonal antibodies has been extensively studied but also has been described against the TNF receptor etanercept, rituximab,11 tocilizumab,12 and abatacept.13,14 In contrast with anti-TNF monoclonal antibodies, the incidence of antibodies against etanercept is low,15 and the AABs are not neutralizing.16,17 A significant amount of information is available on rheumatoid arthritis (RA) and inflammatory bowel disease (IBD) but is scarce for other IMIDs. Overall, data regarding the presence of AABs and safety and efficacy are fragmentary. Knowledge of factors affecting the immunogenicity of biologic agents could help in the development of treatment strategies to prevent loss of efficacy and improved safety. The main objective of this study was to summarize the data on the presence of AABs and their relationship with the efficacy and safety of biological therapies in IMIDs.
A systematic literature review was performed to identify all publications that analyzed the immunogenicity of biological therapies in IMIDs. Data regarding the use of biologic agents in nonapproved indications were dismissed. The protocol of this review is available from the corresponding author; the Preferred Reporting Items for Systematic Reviews and Meta-analyses consensus was followed for the systematic review and meta-analysis.18
MEDLINE, EMBASE, Cochrane Library, and the Web of Knowledge were searched for articles published between 2000 and March 2012. The search strategy focused on synonyms of diseases, biologic agents, and immunogenicity, and was limited to articles published in English, Spanish, French, Italian, or Portuguese. We also included abstracts online from 2001 to 2011 from the European League Against Rheumatism (EULAR) and the American College of Rheumatology congresses.
The selection criteria were (1) studies in patients with RA, juvenile idiopathic arthritis, IBD, ankylosing spondylitis, psoriasis, psoriatic arthritis, or other spondyloarthropathies; (2) studies in patients receiving at least 1 biological drug; (3) studies collecting data on antibodies against biologic agents using some method of measurement; and (4) retrospective or prospective observational studies or intervention studies. Two reviewers (J.R.M. and E.S.) screened articles and abstracts for selection criteria independently. Once unrelated articles were excluded, the full report of selected studies was reviewed. Subsequently, articles not fulfilling all selection criteria were excluded. A table with reasons for exclusion was constructed (eTable 1 in the Supplement). For completeness, a reverse search of included articles and a hand search of published clinical trials of biological therapies were performed.
Data collected included publication details, study design, AABs status and method of measurement, definition of response, and characteristics of patients and treatments. The primary end point was the association of AABs with response to treatment. Secondary end points were the association of AABs with safety, the association of AABs with concentration of the drug, and the influence of use of concomitant immunosuppressive therapy in the formation of AABs.
We created an ad hoc checklist to analyze the risk of bias of the included studies in the meta-analysis. The checklist contains 30 items with a score range of 0 to 100, with 100 indicating the absence of risk of bias. This checklist (available from the authors on request) was based on the guidelines for assessing quality in prognostic studies on the basis of framework of potential biases proposed by Hayden et al.19 The level of evidence of the studies was assessed using the levels of evidence of the Oxford Centre for Evidence-Based Medicine.20
Meta-analyses were performed when at least 3 studies had comparable outcome measures using the random-effects method for computing odds ratios (ORs) of DerSimonian and Laird.21 For each available analysis, the effect was plotted by the inverse of its standard error to identify the risk of publication bias by visually assessing the symmetry of such funnel plots. Statistical significance was determined using the Egger test.22 Heterogeneity was tested as proposed by Higgins and Thompson23 and Higgins et al24 using I2. An I2 value greater than 40% was arbitrarily considered a high level of heterogeneity. If high statistical heterogeneity was present, possible explanations were investigated using sensitivity analysis and meta-regression. Meta-regression was aimed to determine the contribution of time to assess response, number of participants, quality of the data, time of disease duration, method of measurement of antibodies, design of the study, and levels of evidence. P < .10 was considered significant in meta-regression. Statistical analysis was conducted using commercial software (Stata, version 11.1for Windows (StataCorp LP) was used in all statistical analyses.
The search captured 10 728 articles and abstracts. After title and abstract screening, 95 articles were retrieved for full-text review. By hand search and reverse search, 31 articles were added.25- 55 A total of 66 articles were excluded after detailed review; 60 articles and abstracts were included in the present analysis (eFigure 1 in the Supplement).
In 64 studies from 60 articles, 13 982 patients were included. Thirty studies were randomized clinical trials and 34 were observational studies. Characteristics of the selected articles, including the demographics of patients, duration of disease, treatment, and treatment response definition, are presented in the Table. 4- 8,13,14,25- 45,47- 56,58- 78 Forty-two studies analyzed the association of AABs with clinical response. In 50 articles (78%), secondary end points, such as the association of AABs with safety, association of AABs with concentration of the drug, and influence of concomitant treatment in AABs formation, were reported.
The quality of data was higher than 70% in 37 of the studies (60%). The level of evidence was 4 in 10 studies and 3 in the other 54 studies. Individual results of studies are presented according to outcome in eTables 2 through 5 in the Supplement. Only data on anti-TNF monoclonal antibodies were sufficient to perform a meta-analysis; data on other biologic agents were insufficient.
The measures to evaluate efficacy in RA are the EULAR response criteria and the rate of improvement in the American College of Rheumatology criteria of response79,80 (eTable 6 in the Supplement).
Five studies with data on infliximab and adalimumab analyzed EULAR responses from 12 to 192 weeks.8,60,62,63,72 Significantly better response in seronegative patients was reported in most studies. In the meta-analysis of EULAR responses at 6 months, the OR of response for seropositive patients was 0.03 (95% CI, 0.01-0.21) with heterogeneity (I2 = 49.6%; P = .11) (Figure 1).8,49,53,60,62,63,65- 67,75,78 The type of biologic agent, design of the study, and level of evidence of the studies were identified by meta-regression as the sources of the heterogeneity. No asymmetries were found in a funnel-plot evaluation (Egger test P = .055). In the meta-analysis of EULAR response between month 6 and month 12, the OR for seropositive patients was 0.03 (95% CI, 0.00-0.30) with heterogeneity (I2 = 71%; P = .02). Only the length of follow-up was identified as a source of heterogeneity. The Egger test was significant for publication bias (P = .02).
Five studies with infliximab, adalimumab, golimumab, abatacept, and rituximab reported the association of AABs with the American College of Rheumatology 20 response.14,28,39,42,45 Only one study45 with adalimumab showed a significant difference in American College of Rheumatology 20 response comparing seropositive and seronegative patients. The meta-analysis included 3 studies with an OR of 0.57 (95% CI, 0.25-1.28) and heterogeneity (I2 = 62.8%; P = .07). No factors accounting for heterogeneity were found, and the funnel plot did not show publication bias (Egger test P = .24).
Efficacy in spondyloarthropathies was evaluated using the criteria response of Assessment in the Ankylosing Spondylitis International Working Group (ASAS)81 (eTables 2-5 in the Supplement). Four studies reported the association of AABs with ASAS response in spondyloarthropathies.65- 68 Two studies included data on adalimumab and 2 reported results on infliximab. Individual studies suggested that seronegative patients presented a better ASAS response than did seropositive patients. Three studies65- 67 were included in this meta-analysis, with an OR of 0.28 (95% CI, 0.08-1.03) (Figure 1). The heterogeneity was I2 = 52.9% (P = .12). No factor was identified as a source of heterogeneity. The funnel plot did not show publication bias (Egger test P = .13).
Data were extracted from 11 studies.4,30,49,53,74,75,77,78 All but one included studies with infliximab. A significantly better clinical response in seronegative patients than in seropositive patients was reported in most studies. Five studies49,53,75,78 were included in the meta-analysis, resulting in an OR of 0.53 (95% CI, 0.16-1.72) with heterogeneity of I2 = 74.1% (P = .004) (Figure 1). The meta-regression identified quality of data and design, and level of evidence of studies as sources of heterogeneity. No asymmetries were found in funnel-plot analysis (Egger test P = .87). Meta-analysis was also performed after removing the retrospective studies,75,78 and no significant result with absence of heterogeneity was found (OR, 1.08; 95% CI, 0.78-1.50; I2 = 0.0%; P = .50).
Data on remission in IBD were reported in 3 studies; 2 studies with infliximab27,30 and one with certolizumab pegol.52 In one with infliximab and another with certolizumab, no seropositive patients achieved remission.30,52 The meta-analysis with these 3 studies showed a relative risk of 0.27 (95% CI, 0.04-1.80) with a heterogeneity of I2 = 63.4% (P = .06). The age and sex of patients, length of follow-up, design of studies, and quality of data were identified as sources of heterogeneity. No asymmetries were found in funnel-plot analysis (Egger test P = .25).
Loss of response was reported in 4 studies44,57,59,75 with anti-TNF monoclonal antibodies. Seropositive patients presented numerically higher rates of loss of response than did seronegative patients. Meta-analysis showed an OR of 3.0 (95% CI, 0.99-9.09) for loss of response for seropositive patients, with I2 = 31.8% (P = .22). No asymmetries were found in funnel-plot analysis (Egger test P = .90).
Significantly higher concentrations of AABs in patients with infusion reaction were reported in 2 observational studies.4,62 Seventeen studies reported data on serologic status of AABs (positive/negative) and risk of hypersensitivity reactions.4,8,25,26,30- 33,43,48,49,53,56,57,62,66- 68,75 Seropositive patients presented more hypersensitivity reactions than did seronegative patients treated with anti-TNF monoclonal antibodies irrespective of the diagnosis. One study43 with rituximab in patients with RA showed similar results. Meta-analysis was performed grouping patients treated with anti-TNF monoclonal antibodies for RA, IBD, and spondyloarthropathies (Figure 2).8,25,26,30- 33,48,49,53,56,57,66- 68,75 The resulting OR was 3.97 (95% CI, 2.36-6.67), with I2 = 62.5% (P < .001). The length of follow-up, prior use of TNF inhibitors, quality of data and design, and level of evidence of studies were identified as possible sources of heterogeneity in meta-regression. Analysis stratified by these factors was conducted. Studies with lower quality of data (≤0.60) presented nonsignificant results (OR, 6.97; 95% CI, 0.97-50.35; I2 = 62.2%). However, studies of moderate (0.61-0.79) or high (≥0.80) quality reported significant ORs of 3.97 (95% CI, 2.36-6.67) and 1.96 (95% CI, 1.36-2.84), with I2 = 0.0% and 14.2%, respectively. No publication bias was shown in funnel-plot analysis (Egger test P = .23).
Three studies5,57,62 reported data on discontinuation of treatment for all reasons in patients with RA. Data suggested that seropositive patients discontinued treatment more frequently than did seronegative patients. In meta-analysis, the OR for discontinuation of treatment in seropositive patients was 3.53 (95% CI, 1.60-7.82), with I2 = 34.5% (P = .22). No publication bias was found in funnel-plot analysis (Egger test P = .95).
Thirty studies analyzed the use of concomitant treatment and development of AABs. These studies suggested that patients who received concomitant anti-TNF monoclonal antibodies and nonbiological disease-modifying antirheumatic drugs experienced less immunogenicity than did patients who received only anti-TNF monoclonal antibodies. One study13 reported no significant difference in the rate of seropositivity comparing patients receiving abatacept and concomitant disease-modifying antirheumatic drugs and patients who received abatacept monotherapy. Eighteen studies on patients with different diseases receiving different anti-TNF monoclonal antibodies were included in the meta-analysis (Figure 3).4,27,30,32,40,51,53- 55,73,75,77,78 The resulting OR was 0.32 (95% CI, 0.25-0.42) without heterogeneity (I2 = 0.0%, P = .71). Funnel-plot analysis did not show risk of publication bias (Egger test P = .08).
The specific effect of different drugs in AABs immunogenicity was also investigated. In observational studies7 in IBD, patients who received azathioprine combined with anti-TNF monoclonal antibodies developed fewer AABs than did patients with anti-TNF monoclonal antibody monotherapy. Meta-analysis was not performed because of insufficient data. In the meta-analysis of the influence of concomitant methotrexate in AABs development, 13 studies7,29,34- 36,38,39,41,57,58,62,65 were included. The resulting OR was 0.31 (95% CI, 0.18-0.54), with an I2 of 44.3% (P = .04). The diagnosis of the disease was reported as a possible factor of heterogeneity in meta-regression. Stratified analysis was conducted after removing one study41 in patients with juvenile idiopathic arthritis and merging psoriasis/psoriatic arthritis, IBD, and ankylosing spondylitis as spondyloarthropathies. In patients with RA, the resulting OR for the presence of AABs in combined therapy was 0.15 (95% CI, 0.06-0.35), with I2 = 10.6% (P = .35). In patients with spondyloarthropathy, OR was 0.52 (95% CI, 0.28-0.95), with I2 of 33.3% (P = .12). The funnel plot did not show risk of publication bias (Egger test P = .50).
Four studies32,53,58 reported the effect of concomitant use of oral corticosteroids and AABs. No significant differences were found comparing monotherapy with combined therapy. Meta-analysis of these 4 studies was performed with a resulting OR of 0.73 (95% CI, 0.43-1.17) without heterogeneity (I2 = 0.0%; P = .41). Funnel-plot analysis did not show risk of publication bias (Egger test P = .455).
Nine studies5,7,8,37,58,64,66,76 analyzed the association of AABs with concentrations of infliximab, adalimumab, and golimumab in patients with RA, spondyloarthropathies, and IBD. Drug concentrations were lower in patients with higher titers of AABs, with a significant difference in most studies. Meta-analysis was not performed because of insufficient analogous comparisons.
In our review, we investigated the association of AABs with the efficacy and safety of biological drugs, as well as the role of concomitant therapy in the development of these antibodies. Main limitations of the study were the diversity in methods of determination of AABs and inconsistency in the time of antibody investigation and at assessing response. Other weaknesses were related to the variance of the design of studies included in the analysis (clinical trials as well as prospective and retrospective observational studies). To minimize this issue, our analysis of heterogeneity included not only quality of data but design and level of evidence of the studies. For completeness, diagnosis of disease as a putative heterogeneity factor was included in the meta-regression. Although OR is not the best estimate of association, we used it because OR was the most frequently provided measure in the different studies. Finally, this review focused mainly on the immunogenicity of anti-TNF monoclonal antibodies (infliximab and adalimumab) that have the highest rate of immunogenicity among biologic agents; therefore, conclusions may not be extended to other drugs because of the absence of information.
One of the benefits of meta-analysis is to extend conclusions beyond the populations that are included in a single study. However, meta-analysis can be limited by sampling bias, inadequate data, and biased outcome interpretation.
There is some disagreement on restriction of the analysis to randomized clinical trials. However, observational studies often represent the best available evidence. Observational studies are thought to overestimate treatment or exposure effects. Nevertheless, meta-analyses of observational studies continue to be valuable and common in assessing efficacy and effectiveness, and they are being published in increasing numbers.82
Heterogeneity may help to identify factors that influence the results of the outcomes that were not observable in individual trials.83,84 Our statistics included analysis of heterogeneity, risk of bias, and quality of data with stringent predefined criteria. Sensitivity analysis was performed by stratification of meta-analyses by variables causing heterogeneity. That was the case in the analysis of association of AABs with hypersensitivity reactions or in the analysis of influence of methotrexate in the development of AABs. Risk of publication bias was predefined as a significant result in the Egger test, and this occurred in the meta-analysis of the EULAR response between month 6 and month 12. Finally, the quality of data was more than 70% in 37 studies (60%).
It has been reported10 that AABs produce a decrease in drug concentration and a parallel reduction in efficacy. This could be the result of clearance of free drug by immune-complex formation. Herein, the greater response was greater in seronegative than in seropositive patients regardless of the type of anti-TNF monoclonal antibodies or type of disease. In contrast, analysis of the association of AABs with efficacy produced highly variable results. This could have several explanations. First, studies on different anti-TNF monoclonal antibodies were included in the meta-analysis. Dissimilar immunogenicity of chimeric and human monoclonal antibodies has been reported.85 However, the type of the anti-TNF monoclonal antibody was taken into account in the meta-regression and type was not identified as a heterogeneity factor. Second, different immunogenicity with different prevalence in AABs formation in diseases included in this review has been documented.86 In agreement, the association of AABs with efficacy was relevant in RA but not in spondyloarthropathies or IBD. In RA, the disease activity score reflects absolute values. It is used to determine and evaluate the status and course of disease activity in individual patients.87,88 The EULAR criteria are based on a significant change in disease activity score in relationship to the level of disease activity attained, whereas the American College of Rheumatology criteria are based only on change and do not consider the actual disease activity at the end point.79 Of relevance in clinical practice, the disease activity score at the time of evaluation correlates best with the decision to change medication therapy.89 Thus, our results concerning the effect of AABs on the EULAR response could be of clinical relevance. These results should be interpreted with caution because of the presence of publication bias and heterogeneity. The presence of AABs did not have a detectable effect on efficacy in patients with IBD. This may be the result of variability in the definition of response. Nonetheless, difference in the immunogenicity profile of IMID as discussed above could account for the result. The results of the effect of concomitant medication on the clinical response support the impact of the difference in the immunogenicity profile. Meta-regression to identify factors affecting heterogeneity showed that diagnosis was the cause as confirmed by the stratified analysis. On the whole, discontinuation of the biologic agent for all reasons reflecting effectiveness was affected by the presence of AABs.
Meta-analysis of the association of AABs with safety showed higher risk of hypersensitivity reactions in seropositive than in seronegative patients, yet there was a high level of heterogeneity. Interestingly, one of the factors accounting for the heterogeneity was prior use of TNF inhibitors. However, the stratified analysis did not reveal significant association with prior use of TNF antagonists. A previous article90 reported that development of antibodies against infliximab do not affect a patient’s response to adalimumab.
Biologic agent monotherapy is commonly used in clinical practice. A survey91 of prescribing RA practices found that 28% to 30% of patients with newly diagnosed RA received a biologic agent as monotherapy. Several trials suggest that TNF inhibitors may need to be given in conjunction with methotrexate to improve the clinical efficacy in patients who do not respond to methotrexate therapy.92 In RA, the presence of antibodies against anti-TNF monoclonal antibodies produces significantly worse clinical response and higher risk of discontinuation of treatment and development of hypersensitivity reactions. The use of combined therapy of anti-TNF monoclonal antibodies with disease-modifying antirheumatic drugs reduces the formation of AABs and these risks. Thus, monotherapy with anti-TNF monoclonal antibodies may not be suitable for treatment of RA. Information regarding other biologic agents and IMID is limited.
Accepted for Publication: March 24, 2013.
Corresponding Author: Jose Ramon Maneiro, MD, Rheumatology Department, Complejo Hospitalario Universitario de Santiago de Compostela, C/Choupana s/n, 15701 Santiago de Compostela, Spain (email@example.com).
Published Online: June 24, 2013. doi:10.1001/jamainternmed.2013.7430.
Author Contributions: All authors gave final approval of the version to be published.
Study concept and design: All authors.
Acquisition of data: Maneiro and Salgado.
Analysis and interpretation of data: All authors.
Drafting of the manuscript: All authors.
Critical revision of the manuscript for important intellectual content: Gomez-Reino.
Statistical analysis: Maneiro and Salgado.
Obtained funding: Gomez-Reino.
Study supervision: Gomez-Reino.
Conflict of Interest Disclosures: Drs Maneiro and Salgado have received grants from UCB Pharma. Dr Gomez-Reino is a member of the advisory boards of BMS, Pfizer, Roche, Schering-Plough, and UCB SA; has received lecture fees from BMS, Roche, Schering-Plough, and Wyeth; and has received research grants from Roche and Schering-Plough.
Funding/Support: The study was supported by an unrestricted grant from UCB Pharma, Spain.
Role of the Sponsor: The funding sources had no role in the design and conduct of the study; in the collection, analysis, and interpretation of the data; or in the preparation, review, or approval of the manuscript.