Article Text

Extended report
Long-term safety of etanercept and adalimumab compared to methotrexate in patients with juvenile idiopathic arthritis (JIA)
  1. Jens Klotsche1,2,
  2. Martina Niewerth1,
  3. Johannes-Peter Haas3,
  4. Hans-Iko Huppertz4,
  5. Angela Zink1,5,
  6. Gerd Horneff6,
  7. Kirsten Minden1,7
  1. 1German Rheumatism Research Centre Berlin, a Leibniz Institute, Berlin, Germany
  2. 2Charité Universitätsmedizin Berlin, Institute for Social Medicine, Epidemiology and Health Economics, Berlin, Germany
  3. 3German Center for Pediatric and Adolescent Rheumatology, Garmisch-Partenkirchen, Germany
  4. 4Professor Hess Children's Hospital, Bremen, Germany
  5. 5Charité Universitätsmedizin Berlin, Department of Rheumatology and Clinical Immunology, Berlin, Germany
  6. 6Asklepios Kinderklinik Sankt Augustin GmbH, Sankt Augustin, Germany
  7. 7Charité Universitätsmedizin Berlin, Children's university hospital, Berlin, Germany
  1. Correspondence to Dr Jens Klotsche, German Rheumatism Research Centre, a Leibniz Institute, Chariteplatz 1, Berlin 10117, Germany; Jens.Klotsche{at}


Importance Published evidence on the long-term safety of etanercept (ETA) and adalimumab (ADA) in patients with polyarticular juvenile idiopathic arthritis (pJIA) is still limited.

Objectives To investigate the rates of serious adverse events (SAE) and of events of special interest (ESI) under ETA and ADA treatment.

Design, setting and participants Patients with pJIA were prospectively observed in the national JIA biological register, Biologika in der Kinderrheumatologie, and its follow-up register, Juvenile arthritis Methotrexate/Biologics long-term Observation.

Main outcomes and measures We calculated the relative risks of SAE and ESI for ETA and ADA compared with methotrexate (MTX).

Results Among the 1414 patients treated with ETA (n=1414; 4461 exposure years (EY)) and ADA (n=320; 493 EY), significantly more SAE, infections and medically important infections were observed (ETA: 4.5, 5.7, 0.9; ADA: 4.7, 11.4, 0.4 per 100 EY) compared with those treated with MTX alone (n=1455; 2.907 EY; 2.6, 5.5, 0.5 per 100 EY). The risk for malignancies was not significantly increased for ETA and ADA compared with MTX (0.09, 0.27 and 0.07/100 person-years). Patients under ETA monotherapy developed more frequently incident inflammatory bowel disease (IBD) and incident uveitis (0.5 and 0.8/100 EY) than patients treated by ETA in combination with MTX (0.1 and 0.2/100 EY) or MTX alone (0.03 and 0.1/100 EY).

Conclusions and relevance Our data confirm the acceptable long-term tolerability of ETA and ADA in pJIA. However, whether the onset of IBD and uveitis during ETA monotherapy is a paradoxical effect or an inadequate response to therapy remains unclear and requires further investigation in this growing cohort.

  • DMARDs (biologic)
  • Methotrexate
  • Juvenile Idiopathic Arthritis
  • Infections
  • Autoimmune Diseases

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.


Juvenile idiopathic arthritis (JIA) is one of the most common chronic conditions of childhood. JIA comprises a group of heterogeneous forms of arthritis characterised by persistent joint inflammation lasting longer than 6 weeks and beginning before the age of 16 and has an unknown cause. According to the classification criteria of the International League of Associations for Rheumatology (ILAR),1 the term JIA covers seven mutually exclusive categories with differences in their clinical presentation, disease course and treatment response. JIA often extends into or relapses during adulthood, potentially leading to severe disability.2–5

Given the current inability to cure JIA, treatment aims towards disease remission, including controlling inflammation, reducing joint damage and avoiding long-term disability.6 Conventional pharmacological therapies consist of non-biological, disease-modifying antirheumatic drugs (nbDMARDs), among which methotrexate (MTX) is the most commonly prescribed first-line DMARD. However, the use of MTX may be individually limited by toxicity, incomplete response or even non-response in patients with JIA.4 ,7 Since 1999, the treatment options for JIA have been extended to biological DMARDs (bDMARDs), including the tumour necrosis factor (TNF)-α receptor antagonists, etanercept (ETA) and adalimumab (ADA, licensed in 2003 in Germany), as second-line treatments after nbDMARD treatment failure.

To date, published data suggest a favourable safety profile for ETA4 ,7–13 and ADA14–16 for the treatment of moderately or severely active polyarticular JIA. However, published safety reports from clinical trials and observational studies provide insufficient data due to their limited observation times under drug exposure and small sample sizes. Furthermore, there are no prospectively collected data available on biological drug safety in comparison to nbDMARDs in patients with JIA.

The objective of this study was to investigate serious adverse events (SAE), as well as events of special interest (ESI), in patients with JIA treated with ETA and ADA, compared with a biologic-naive MTX group in the German JIA biologic register, Biologika in der Kinderrheumatologie/Biologics in Paediatric Rheumatology (BiKeR), and its follow-up register, Juvenile arthritis Methotrexate/Biologics long-term Observation (JuMBO), which follows young adults after transitioning into the care of adult rheumatologists.


The paediatric register, BiKeR,17 and its follow-up register, JuMBO,7 are both ongoing, prospective, non-interventional, observational cohort studies that collect data on safety, effectiveness and treatment adherence in patients with JIA exposed to biological agents. Children with definite JIA according to ILAR1 beginning treatment with ETA or ADA have been consecutively enrolled in BiKeR since 1 January 2001. A cohort of patients not exposed to biologics but, instead, to MTX alone was built from 2005 to 2011.18 Young adults were transferred to the follow-up registry, JuMBO, for long-term follow-up into adulthood when they left paediatric care and BiKeR at 18 years of age. A detailed description of the enrolment process in BiKeR and JuMBO can be found elsewhere.7 ,17

Within this study, only patients with JIA with a polyarticular disease course, systemic JIA, enthesitis-related and psoriatic arthritis and at least one follow-up visit in BiKeR or JuMBO were included in the analyses. Patients were prospectively assessed semiannually by physician-reported and patient-reported outcomes via questionnaires. Physicians examined the health status of the patient at his/her office and sent the questionnaires to the coordinating study centre. Next, the physicians reported on current disease activity (global assessment), 72-joint count, erythrocyte sedimentation rate19 and the details of current nbDMARD and bDMARD use. The diagnosis, date of symptom onset and any past treatments were recorded at first documentation. Physician-derived 6-monthly follow-up details also included the dates and reasons for the cessation of drugs, if appropriate, as well as adverse events. Parents were requested to assess the activity of their child's disease using current pain levels on a 10-point-visual analogue scale and to complete the childhood health assessment questionnaire (CHAQ).20 Young adults aged >18 were requested to assess themselves the disability via the health assessment questionnaire (HAQ).21

Events, therapies and at-risk window

Safety analyses focused on SAEs and ESIs. SAEs are defined as events that were fatal or life threatening, led to a persistent or significant disability or incapacity or required a prolonged hospitalisation. ESIs include cases of death, malignancy, medically important infections (MII, infections that led to hospitalisation and/or required intravenous antibiotic treatment) and incident immune-mediated diseases. ESIs did not have to fulfil the appropriate criteria for seriousness; instead, they corresponded to the criteria considered for the Pharmachild register,22 excluding pregnancies. The Medical Dictionary MedDRA V.15.1 was used to code all of the events.

For this analysis, patients were only eligible if they had ever exposed to ETA (n=1414), ADA (n=320) or MTX without a bDMARD (n=1455). The time at risk began with the first application of the index drug ETA, ADA or MTX in each patient. The time of drug exposure ended at the recorded discontinuation date, at the last available visit or on the date of being switched to another nbDMARD or bDMARD, whichever occurred first. An SAE or ESI was attributed to ETA, ADA or MTX when the treatment was either ongoing or terminated <3 months prior to the event (3-month at-risk window approach).23 All single treatment episodes of MTX, ETA and ADA were counted across the observation time to calculate the years of drug exposure. In cases of concomitant treatment with MTX (ETA+MTX, ADA+MTX), only the time under the bDMARD was counted. This approach resulted in 4461 exposure years (EY) for ETA, 493 EY for ADA and 2907 EY for MTX. The ever-exposed approach was applied when considering malignancies and mortality. Patients were considered to be ever exposed to ETA or ADA for the time period from the start of anti-TNF-α treatment to the end of follow-up. The risk period ended for MTX either at the end of follow-up or at the date of switching to a biological drug (MTX: 2751 person-years (PY); ETA: 6619 PY; ADA: 745 PY).


To estimate the risk under a specific treatment, event rates per 100 EY for SAEs and ESIs or PY for malignancies and mortality were calculated. The possible effects of treatment decision on the outcome (confounding by indication) were modelled by propensity score methods, adjusting for differences in clinical parameters between anti-TNF-α and MTX at the beginning of BiKeR. The likelihood of being treated with anti-TNF-α instead of nbDMARDs was estimated by a logistic regression model, including the predictors JIA category, disease duration, physician global, CHAQ/HAQ, patient-reported pain, history of uveitis and number of previously used nbDMARDS. For each patient and each 6-month follow-up period, the probability of switching from MTX to anti-TNF-α or from ETA to ADA (as well as vice versa) was estimated by a logistic regression model using the covariates from the propensity score model. Weights were calculated as ratios of both probabilities and were combined for different 6-month time periods, as described by Molenberghs and Kenward,24 for the final weighting. Balanced samples of patients were obtained using an inverse probability of treatment and censoring weights (IPTCW)25 with respect to the treatment decision at baseline and the switching of drugs during follow-up. Generalised estimation equations26 were applied to the IPTCW-weighted sample of patients to analyse the risk of adverse events among the three groups, ETA, ADA and MTX. Sensitivity analyses using alternative risk windows are presented in the online supplementary text. Standardised incidence ratios (SIR) for malignant diseases were calculated based on data of the German-based Society for Paediatric Oncology and Haematology.27


Patient characteristics

Between 1 January 2001 and 31 December 2012, 1162 patients started treatment with ETA and 46 with ADA. A total of 1055 patients were included in the biologic-naive MTX control group (table 1). More than 40% of the patients had a polyarticular onset of JIA (ie, 36% rheumatoid factor (RF)-negative and 8% RF-positive polyarthritis). Patients with systemic JIA on a polyarticular disease course comprised 7.6% of the cohort, and approximately 50% of them showed extra-articular manifestations. Patients had a mean age of 11.3 years at enrolment. Patients starting on ETA or ADA had a significantly higher disease activity (mean 57.7) than MTX patients (mean 48.5). At baseline, 78% of the patients on ETA or ADA were on a combination therapy with MTX. The rate of concomitant glucocorticoid use significantly differed between the MTX (21.5%) and ETA and ADA group (59.1%) at treatment initiation. The mean observation time was 5.5 years (min=1 month, max=13.1 years) for the total cohort of 2263 patients with JIA and 8.9 years for the JuMBO subcohort (n=646), including only patients with JIA above the age of 18.

Table 1

Patient characteristics of the cohort by index drugs at study enrolment

Serious adverse events

Patients experienced 75 SAEs (2.6 events/100 EY) under MTX, 199 SAEs (4.5 events/100 EY, relative risk (RR)=2.2, p<0.001) under ETA and 23 (4.7 events/100 EY, RR=2.2, p=0.006) under ADA treatment, as shown in table 2. Approximately 18% of the SAEs were JIA-related (JIA flare or joint operation). A possible relation to drug exposure was stated for 8% of the SAEs.

Table 2

Incidence of serious adverse events and infections in children, adolescents and young adults treated with methotrexate, etanercept or adalimumab (3-month at-risk window approach)


Patients developed 471 infections (MTX: 160, 5.5 events/100 EY; ETA: 255, 5.7 events/100 EY, RR=1.4, p=0.015; ADA: 56, 11.4 events/100 EY, RR=2.2, p=0.001) as reported in table 2. Overall, 41 MII were recorded in the ETA group (0.9 events/100 EY, RR=2.1, p=0.03), 2 in the ADA group (0.4 events/100 EY, RR=0.8, p=0.87) and 15 in the MTX group (0.5 events/100 EY). The rate of MII was increased for ETA with concomitant MTX use (1.03 events/100 EY) versus ETA monotherapy (0.7 events/100 EY). Similar rates for sepsis were seen for MTX (0.03 events/100 EY) and ETA (0.07 events/100 EY, p=0.540).

Twenty-nine of the infectious events were attributed to bacterial infections, and 107 to viral infections. Among the latter category, there were 16 varicella zoster virus infections; 6 chickenpox cases were reported in the MTX group (0.2 events/100 EY) compared with 8 events under ETA (0.2 events/100 EY, RR=0.9, p=0.849) and 2 events under ADA (0.4 events/100 EY, RR=1.7, p=0.57). More detailed information on chickenpox and macrophage activation syndromes can be found in the online supplementary text. Tuberculosis and other opportunistic infections were not observed under the ETA or ADA treatment in either cohort.

Malignancies and death

A total of eight malignant diseases were observed during follow-up (see table 3 and online supplementary table S1). Among the patients with malignancies, two occurred in the MTX group (0.07 events/100PY), six in the ETA group (0.09 events/100PY, RR=1.03, p=0.948) and two in the ADA group (0.27 events/100PY, RR=1.92, p=0.612), both of whom had also been exposed to ETA before switching to ADA. A yolk sac carcinoma was noted within the first month of ETA treatment, it was likely present before the exposure to ETA. Among all patients ever treated with ETA without concomitant MTX use, only one malignancy was reported (0.02 events/100 PY). The other five were treated with etanercept in combination with MTX and glucocorticoids (0.08 events/100PY). The SIR was 3.6 (95% CI 1.22 to 9.05) for overall malignancies within the ETA exposure group, 10.7 (95% CI 1.23 to 43.06) within the ADA group and 2.9 (95% CI 0.33 to 11.6) within the MTX group compared with a German reference population.

Table 3

Rates of malignancies and mortality in the study cohort (at-risk window: ever exposed)

Four patients with JIA died in the ETA cohort (0.06 events/100PY; two sepsis, one macrophage activation syndrome, one carditis) and one child died from leukaemia in the MTX cohort (0.04 events/100 PY, table 3, online supplementary text and table S1).

Incident immune-mediated diseases

Incidental uveitis was observed in 21 patients under ETA (0.5 events/100 EY, RR=4.8, p=0.012), in 1 (0.2 events/100 EY, RR=1.8, p=0.617) under ADA and in 3 under MTX (0.1 events/100 EY) (table 4). ADA monotherapy was started 2 weeks before uveitis onset after ETA had been discontinued due to ineffectiveness. Patients under ETA monotherapy developed incident uveitis (n=15; 0.8/100 EY) more frequently than those on ETA in combination with MTX (n=6; 0.2/100 EY). ETA was discontinued in eight patients (38.1%) due to the uveitis. Among patients with incident uveitis, 14 (66%) were antinuclear antibody positive, 7 had enthesitis-related arthritis, 6 extended oligoarthritis, 3 psoriatic arthritis and 2 RF-negative polyarthritis. Patients developed uveitis after a median of 5.2 years (IQR: 3.6–10.3) after the onset of JIA.

Table 4

Incidence of events of special interest in children, adolescents and young adults treated with methotrexate, etanercept and adalimumab (3-month at-risk window approach)

Incidental inflammatory bowel disease (Crohn's disease/colitis, IBD) was observed in 14 patients (MTX: 1, 0.03 events/100 EY; ETA: 12, 0.27 events/100 EY, RR=8.1, p=0.044; ADA: 1, 0.2 events/100 EY, RR=5.7, p=0.221). Higher rates of IBD were observed under ETA monotherapy (n=8; 0.5/100 EY) compared with ETA in combination with MTX (n=4; 0.1/100 EY). IBD developed after a median of 8 years (IQR: 4.8–11.2) after the disease onset. Among the 12 cases in the ETA group, 5 had extended oligoarthritis, 3 enthesitis-related arthritis, 3 RF-negative polyarthritis and 1 psoriatic arthritis. Finally, 16% were human leucocyte antigen-B27 positive.


Here, we presented safety data comparing treatment with ETA and ADA to a biologic-naive MTX control group, which were obtained via the BiKeR and JuMBO registers. Due to the current drug-licensing situation, both registers have provided data on the long-term drug safety in a real-life setting, so far primarily for ETA and ADA. To our knowledge, this is the largest cohort study on JIA to date. However, the results for ADA should be interpreted with some caution due to the smaller numbers of patients and years under drug exposure.

Compared to the reference patient population (MTX group), ETA and ADA were well tolerated by the patients. The SAE rate for ETA (4.5 SAE per 100 EY) is comparable to that in previous reports from other cohort studies4 ,11 ,17 ,28 ,29 (range: 2.9 to 6.3 SAE per 100 EY) and from an open-label extension trial.10 In addition, the SAE rate for ADA in our study was in accordance with the reported rates (4.1 SAE per 100 EY) from clinical trials for ADA.14 ,16 Treatment with both anti-TNF-α inhibitors led to a doubled SAE rate compared with MTX, as already shown by Giannini et al4 in a smaller cohort. A possible explanation for this difference may be the potentially less severe disease course within the MTX group. Moreover, two-thirds of the patients within the ETA and ADA groups were treated in combination with MTX. The shorter disease duration for the MTX group may not explain the lower SAE rate in patients treated with only MTX. Studies in patients with rheumatoid arthritis showed a stable or even decreased SAE rate for MTX.30 ,31

The US Food and Drug Administration placed a ‘black box warning’ regarding the possible risk of malignancy in children under treatment with anti-TNF-α inhibitors after spontaneous reports32 emerged in 2009. We found no increased incidence of malignancies among patients exposed to ETA or ADA compared with MTX. However, the standardised incidence rates for ETA, ADA and MTX were all increased compared with a German reference population. This result is in agreement with the reports of Ruperto and Martini,33 Beukelman et al34 and Simard et al,35 each of whom attributed the overall increased risk for malignancies to the autoimmune disease itself, rather than to treatment with TNF-α inhibitors. In addition, the concomitant and prolonged use of other immunosuppressive drugs than TNF-α inhibitors may also impact the risk of malignancies.36

A number of recent studies4 ,10 ,28 ,37 have suggested a higher risk for infection in patients with JIA exposed to biological drugs. In our cohort, the rate of MII was significantly increased in patients treated with ETA or ADA compared with MTX. This finding may be explained by a higher disease activity and a higher treatment rate with glucocorticoids in patients treated with ETA or ADA.38 ,39

A number of studies28 ,40 ,41 emphasised that incident IBD, psoriasis and uveitis appeared after the initiation of ETA. We observed an increased rate of incident IBD under ETA monotherapy. However, these eight patients had been previously treated with a combination of ETA and MTX. Because the rate of incident IBD is not significantly different between patients treated with MTX monotherapy and ETA in combination with MTX, it is reasonable to hypothesise about the effectiveness of MTX for the treatment of juvenile IBD or regarding any protective effects of this drug.42 The association of IBD, ETA and concomitant drug use requires further research. Dallocchio and colleagues43 speculated that an already established subclinical IBD may be activated by the use of TNF-α inhibitors.43

Uveitis is associated with JIA,44 among other extra-articular manifestations. TNF-α antibodies provide a treatment option for uveitis, reducing flares,45 ,46 whereas ETA does not seem to be an effective option for patients with uveitis.47 ,48 Some studies have even highlighted the occurrence of incident uveitis during ETA treatment.47 ,49 In our cohort, increased incidence rates for uveitis were reported under ETA monotherapy compared with ETA in combination with MTX and MTX controls. The possible link between ETA and the onset of uveitis requires further investigation.50 ,51 ETA may not directly cause the development of uveitis, but the discontinuation of MTX upon successful arthritis control seems to pose a risk to the patient. Thus, patients treated with ETA should be closely monitored for uveitis.52 Patients with a history of uveitis were more likely to receive ADA in our cohort.53 ADA and infliximab are recommended as second-line therapeutic option among patients with JIA and uveitis.52

The study has strengths and limitations. Among its strengths are the large sample size and the inclusion of a biologic-naive control group. The small sample size, brief follow-up time and limited EY are still too low to draw robust conclusions, especially for ADA. The results for ADA should be interpreted with some caution due to the low number of patients and short duration of drug exposure. We were unable to distinguish between acute and chronic uveitis in our registry. The possible influences of disease duration, disease severity, concomitant morbidity, drug history, use of glucocorticoids and further immune response modifiers in the biological cohort cannot yet be estimated with sufficient precision.

The European JIA pharmacovigilance register Pharmachild, in which all of the safety data of the existing national JIA registers are accumulated, will represent a significant contribution to the evaluation of biological drug safety, including ESIs. The data included in both BiKeR and JuMBO will be considered in conjunction with those in the European Pharmachild register, and the documentation of ESIs are harmonised across the three registers.

In conclusion, this is the most comprehensive study of the long-term safety of ETA and ADA in JIA to date. The present data confirm the good tolerability of these substances, as reported in other studies. Because ETA is not efficacious for the treatment of the extra-articular manifestations of uveitis and IBD, ETA monotherapy may not prevent both diseases.


The support of Dr Nico Wulffraat in the JuMBO register is gratefully acknowledged. We thank Karin Weber (German Rheumatism Research Center) for her excellent patient monitoring.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:


  • Handling editor Tore K Kvien

  • Correction notice This article has been corrected since it was published Online First. The heading of the last column in table 2 has been corrected.

  • Contributors JK, GH and KM had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: JK, MN, H-IH, AZ, GH, and KM. Acquisition of data: MN, H-IH, AZ, GH and KM. Analysis and interpretation of data: JK, MN, H-IH, AZ, GH and KM. Drafting of the manuscript: JK, MN, H-IH, AZ, GH and KM. Critical revision of the manuscript for important intellectual content: JK, MN, H-IH, AZ, GH and KM. Statistical analysis: JK. Obtained funding: AZ, GH and KM. Administrative, technical or material support: MN. Study supervision: GH and KM.

  • Funding BiKeR register: joint unconditional grant from Pfizer, Abbvie and Roche; JuMBO register: joint unconditional grant from Pfizer and Abbvie; Pfizer, Abbvie and Roche had no role in the study design or in the collection, analysis or interpretation of the data, the writing of the manuscript or the decision to submit the manuscript for publication. The JuMBO register was supported by the Pharmachild project, what is currently funded by the European Union (EU) within the FP7 framework.

  • Competing interests None declared.

  • Patient consent Obtained.

  • Ethics approval BiKeR was approved by the ethics committee of the Medical Council of North-Rhine Westfalia, Duesseldorf, Germany. JuMBO was approved by the ethics committee of the Charité University Medicine Berlin.

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