Dr. Emery has received consulting fees and/or honoraria (less than $10,000 each) from Schering-Plough, Centocor, and Wyeth.
Therapy of patients with rheumatoid arthritis: Outcome of infliximab failures switched to etanercept
Article first published online: 29 MAR 2007
Copyright © 2007 by the American College of Rheumatology
Arthritis Care & Research
Volume 57, Issue 3, pages 448–453, 15 April 2007
How to Cite
Buch, M. H., Bingham, S. J., Bejarano, V., Bryer, D., White, J., Emery, P., Reece, R. and Quinn, M. (2007), Therapy of patients with rheumatoid arthritis: Outcome of infliximab failures switched to etanercept. Arthritis & Rheumatism, 57: 448–453. doi: 10.1002/art.22617
- Issue published online: 29 MAR 2007
- Article first published online: 29 MAR 2007
- Manuscript Accepted: 10 JAN 2007
- Manuscript Received: 28 JUL 2006
- Arthritis Research Campaign Clinical Research fellowship
- Rheumatoid arthritis;
- Anti–tumor necrosis factor switching
The role of alternative tumor necrosis factor (TNF) antagonist therapies in the context of failure of initial TNF antagonist therapy in patients with rheumatoid arthritis (RA) has yet to be clearly defined. The goal of this study was to determine the efficacy of etanercept in patients who failed to respond to infliximab.
Ninety-five patients with RA who failed to respond to infliximab and methotrexate were treated with etanercept (with continuation of concomitant methotrexate). Thirty-four patients never achieved a response to infliximab (primary nonresponse), 38 had an initial response to infliximab but relapsed (secondary nonresponse), and 23 demonstrated toxicity. Disease Activity Score in 28 joints (DAS28), European League Against Rheumatism (EULAR) response, and American College of Rheumatology (ACR) response were determined after 12 weeks of etanercept.
After 12 weeks of etanercept, 38% of patients achieved an ACR 20% response (ACR20) on etanercept. Of these, 24% and 15% achieved ACR50 and ACR70 responses, respectively. In the primary infliximab nonresponse group, 42%, 30%, and 15% achieved ACR20, ACR50, and ACR70 responses, respectively; the percentages for the secondary nonresponse group were 34%, 21%, and 14%, respectively. Significant DAS28 reductions were observed in the entire cohort and nonresponse subtype groups. Sixty-one percent of the cohort achieved either a moderate or good EULAR score (67% of primary and 56% of secondary infliximab failures). No toxicity was observed in patients who stopped infliximab due to intolerance; 19 of 23 continued etanercept after week 12.
This study confirms that etanercept is effective in patients who fail to respond to infliximab and suggests a higher response in patients who have never had a response to infliximab.
Treatment of rheumatoid arthritis (RA) has been revolutionized by the introduction of tumor necrosis factor (TNF)–blocking therapies. Despite the success of these therapies, a significant proportion of patients treated with infliximab fail to respond (1). The subsequent treatment of these patients has not been well studied, with little published literature on the value of switching from one TNF antagonist to another. The absence of any licensed indication for switching TNF blockers has been made more relevant by the licensing for anti-TNF failures of abatacept and rituximab in the US, with the latter also having a license in Europe.
Three TNF antagonist drugs are currently licensed: infliximab, etanercept, and most recently, adalimumab; all have been shown to be highly successful treatments for patients with RA. Infliximab is a chimeric mouse/human monoclonal antibody and adalimumab is a fully human IgG1 monoclonal antibody, whereas etanercept is a dimeric TNF receptor (p75):IgG1 fusion protein. There are differences between the drugs in that infliximab and adalimumab bind to both soluble and membrane-bound TNFα, whereas etanercept binds not only to soluble and membrane-bound TNFα but also to lymphotoxin α (LTα). In addition, infliximab, which binds to both monomeric and trimeric (bioactive) TNFα forms, is postulated to cause cell lysis via antibody-dependent cell-mediated cytotoxicity or via complement activation (2, 3).
The Swedish cohort (4) is the largest in-house study to date exploring management of TNF antagonist failures, comprising a heterogeneous group of 31 patients including those switching from infliximab to etanercept and vice versa due to either inefficacy or toxicity. Improvements in either Disease Activity Score in 28 joints (DAS28) or swollen joint counts (SJCs) were observed.
The ReAct trial (5), a recent collaborative study (organized by Abbott Immunology), enrolled 4,241 patients with moderate to severe RA and an insufficient response to standard therapies from 450 sites in 11 European countries plus Australia. Patients were treated with adalimumab 40 mg subcutaneously every alternate week in addition to their prior traditional disease-modifying antirheumatic drug (DMARD) therapies. Comparable 12-week responses according to the American College of Rheumatology 20% criteria (ACR20; ∼60%) were achieved in the biologic-naive group (n = 3,553), the group with any previous biologic treatment (n = 688), the group with etanercept as the prior biologic treatment (n = 114), and the group with infliximab as the previous biologic treatment (n = 358). Forty-two percent of patients who had failed to respond to both infliximab and etanercept (n = 78) achieved an ACR20 response. Although this study represents an industry-sponsored, open-label registry, it suggests the general effectiveness of adalimumab in patients who have failed to respond to a previous anti-TNF agent.
Our study cohort comprised 95 patients who failed to respond to infliximab and switched to etanercept for defined reasons. This enabled a more precise assessment of the value of switching from infliximab to etanercept in the context of infliximab failure, providing insight into the role of using additional TNF-blocking agents.
PATIENTS AND METHODS
The present study, which was approved by the Leeds Research Ethics Committee guidelines, was based at the Leeds Teaching Hospitals Resistant Rheumatoid Arthritis Clinic. More than 500 patients attending the clinic from June 2000 to September 2004 have been treated with ≥1 biologic agents (>600 treatment courses) including infliximab and etanercept.
All patients receiving biologic therapy fulfilled the revised 1987 ACR (formerly the American Rheumatism Association) criteria for a diagnosis of RA (6). Eligibility for TNF-blocking agents was according to the British Society of Rheumatology Guidelines (7) (failure of treatment with 2 previous DMARDs including methotrexate and a DAS28 score >5.1 ). Prospective data entered into our biologic agents database included the components of the ACR core set of measures and measurements for DAS28 score and European League Against Rheumatism (EULAR) response. A total of 95 consecutive patients with RA treated initially with infliximab (all receiving 3 mg/kg at weeks 0, 2, 6, and then every 8 weeks as per UK licensing guidelines) who later switched to etanercept due to lack of response (either primary, secondary, or with toxicity) were identified. The groups that switched from infliximab to etanercept were therefore defined as follows: the primary infliximab nonresponse group (n = 34) included patients who never achieved an ACR20 response to infliximab, the secondary infliximab nonresponse group (n = 38) included those who achieved an ACR20 response transiently before relapsing, and the infliximab toxicity group (n = 23) included those who discontinued infliximab therapy due to toxicity and switched to etanercept (analyzed for comparison).
All patients receiving infliximab were taking concomitant methotrexate (dosage range 7.5–25 mg weekly, as either an oral or subcutaneous preparation) as per prescribing guidelines, and continued with methotrexate on switching to etanercept. Intraarticular corticosteroids were not administered within 1 month prior to starting etanercept. A period of 8 weeks between the last infliximab infusion and initiation of etanercept therapy (25 mg subcutaneous twice weekly) was maintained. The ACR and DAS28 scores with EULAR response were calculated after 12 weeks of etanercept therapy. A recently published (9) subgroup analysis and response rates of patients with primary nonresponse to infliximab with no C-reactive protein (CRP) suppression who were switched to etanercept were also included in this study.
Subgroup immunohistochemistry study.
A proportion of patients from this cohort also underwent medical arthroscopy of a swollen knee joint by 1 of 2 medical arthroscopists (RR and MQ) prior to commencing treatment with any biologic agent. Multiple synovial biopsy samples from representative inflamed sites were obtained using grasping forceps 3.5 mm in diameter and a Hopkins 2.7-mm 30° arthroscope (Storz, Tuttlingen, Germany). Based on a previous observation suggesting that expression of LTα could account for failure on infliximab and response to etanercept (10), immunohistochemical evaluation of LTα expression using a standard 3-stage staining procedure (ChemMate [DakoCytomation, Carpinteria, CA]; following incubation with monoclonal antibody, anti-LTα [monoclonal mouse IgG1 clone 5802.21 R&D Systems, Minneapolis, MN]) was undertaken. Sections were randomly analyzed and the histologic features were scored in blinded fashion (MHB). Sublining layer LTα was scored in each field semiquantitatively on a 5-point scale (0–4, where 0 = minimal infiltration and 4 = maximal infiltration), obtaining an average value (11).
Statistical analysis was performed using SPSS for Windows, version 12.0 (SPSS, Chicago, IL). P values less than 0.05 were considered significant. Baseline characteristics (age, CRP level, Health Assessment Questionnaire [HAQ], tender joint count 28 [T28] and swollen joint count [S28]: all parametric) were calculated for the entire group, with further calculation and comparison between the primary and secondary nonresponse groups (one-way analysis of variance [ANOVA]); the percentage with positive rheumatoid factor (RF) was also determined. ACR response after 12 weeks of etanercept therapy was calculated. In addition, DAS28 scores at baseline and week 12 were measured, enabling week 12 EULAR response calculation. Significant differences between baseline and week 12 DAS28 scores were determined by paired t-test. Wilcoxon's signed rank test was used to assess significant difference in synovial LTα expression between the 2 response groups.
A total of 95 patients failed to respond to infliximab and were switched to etanercept (34 primary failures, 38 secondary failures, and 23 with toxicity).
The mean ± SEM age of the entire cohort was 57.2 ± 1.47 years, and 71% of patients were RF positive (Table 1). Baseline mean ± SEM CRP level, HAQ, T28, and S28 were 60 ± 7.37 mg/liter, 2.16 ± 0.64, 14 ± 1.0, and 9 ± 0.86, respectively, with an overall mean ± SEM DAS28 score of 6.41 ± 0.13, findings consistent with a considerably disease-active and treatment-resistant cohort.
|Characteristic||All infliximab NR||Primary infliximab NR||Secondary infliximab NR|
|Age, years||57.2 ± 1.47||57.2 ± 2.42||57.2 ± 1.8|
|RF positive, %||71||69||74|
|CRP level, mg/ liter||60.0 ± 7.37||63 ± 12||58 ± 9|
|HAQ||2.16 ± 0.64||2.15 ± 0.84||2.18 ± 0.97|
|T28||14.0 ± 1.0||16.0 ± 1.5||13.0 ± 1.3|
|S28||9.0 ± 0.8||10.0 ± 1.3||9.0 ± 1.2|
|DAS28||6.41 ± 0.13||6.46 ± 0.22||6.36 ± 0.16|
Comparison of disease characteristics for individual nonresponse subtypes (primary/secondary).
No significant differences (as determined by one-way ANOVA test) in the above baseline parameters were noted. Mean ± SEM age was 57.2 ± 2.42 years and 57.2 ± 1.8 years for the primary and secondary nonresponse groups, respectively; 69% of the primary nonresponse group were RF positive compared with 74% in the secondary nonresponse group toxicity data. Baseline mean ± SEM CRP levels in the primary and secondary nonresponse groups were 63 ± 12 mg/liter and 58 ± 9 mg/liter, respectively, and baseline mean ± SEM DAS28 scores were 6.46 ± 0.22 and 6.36 ± 0.16, respectively.
Duration of infliximab.
The primary infliximab nonresponse group, by definition, stopped infliximab at week 12. The secondary infliximab nonresponse group (n = 38) had been taking infliximab for a median of 60 weeks (range 24–216 weeks) before stopping due to loss of efficacy and switching to etanercept. Patients stopping due to toxicity did so after a median of 12 weeks (range 2–102 weeks), with 16 of 23 stopping within 12 weeks.
Week 12 ACR response to etanercept.
On analyzing the entire infliximab nonresponse group (n = 72) taking etanercept for 12 weeks, 38% of patients achieved an ACR20 response, with 24% and 15% achieving ACR50 and ACR70 responses, respectively (Figure 1). Of the primary infliximab nonresponders (n = 34), 42% of patients achieved an ACR20 response, with 30% and 15% achieving ACR50 and ACR70 responses, respectively. All of the secondary infliximab nonresponse patients (n = 38), by definition, achieved at least an ACR20 response while taking infliximab. However, 36% of these patients never achieved a higher response throughout the exposure to infliximab, 47% achieved an ACR50 response as their best response at some point, and another 17% achieved an ACR70 response at some point. The mean ± SEM DAS28 score of the 38 secondary nonresponse patients at point of relapse was 6.26 ± 0.19. On subsequently switching to etanercept, 34% of secondary nonresponse patients achieved an ACR20 response, 21% achieved an ACR50 response, and 14% achieved an ACR70 response.
Week 12 DAS28 score and EULAR response to etanercept.
In all infliximab nonresponders (n = 72), the mean ± SEM DAS28 score at baseline (prior to commencing etanercept) was 6.4 ± 0.13 (Figure 2). A significant mean reduction in DAS28 score of 1.47 was observed, resulting in a score of 4.9 ± 1.8 after 12 weeks of etanercept (P < 0.0001). In terms of EULAR response, 61% of all infliximab nonresponders achieved a moderate EULAR score and 12% archived a good EULAR score (Figure 3).
In the primary infliximab nonresponse subgroup (n = 34), a significant reduction in mean ± SEM DAS28 score was also observed from 6.46 ± 0.22 at baseline to 4.97 ± 0.38 after 12 weeks of etanercept (reduction of 1.43; P < 0.0001). Sixty-seven percent achieved a moderate EULAR score and 13% achieved a good EULAR score.
In the secondary infliximab nonresponse group (n = 38), a significant reduction in mean ± SEM DAS28 score was also observed from 6.36 ± 0.16 at baseline to 4.79 ± 0.29 at week 12 of etanercept therapy (reduction of 1.53; P < 0.001). Fifty-six percent achieved a moderate EULAR score and 12% achieved a good EULAR score.
Week 12 response to etanercept in primary infliximab nonresponse group with no CRP suppression.
Data from a recently published study of >200 patients that used the presence or absence of CRP level reduction to further dissect primary infliximab nonresponse and define additional subgroups (8) were represented in this study's cohort. The primary infliximab nonresponse group that failed to achieve significant (≥20% from baseline) CRP level suppression (n = 25) demonstrated impressive, higher response rates on etanercept; 68% achieved an ACR20 response, 61% achieved ACR50, and 23% achieved ACR70.
Week 12 response to etanercept in infliximab toxicity group (n = 23).
One patient developed sepsis and died following hospitalization for an unrelated problem. Of the 22 patients completing 12 weeks of etanercept following infliximab toxicity, none developed toxicity with etanercept and 19 patients continued treatment at week 12.
Comparison of synovial LTα expression in ACR50/ACR70 responders and ACR nonresponse with no CRP suppression.
To increase the likelihood of revealing any trend differences between infliximab responders and nonresponders (who subsequently switched to etanercept), synovial LTα expression was compared between 12 patients who achieved either an ACR50 or ACR70 response on infliximab and 5 patients who demonstrated clinical and biochemical nonresponse (no CRP level suppression). There was no significant baseline LTα difference between these 2 groups (median score in response group 0.48 [range 0–2.0] versus 0.75 [range 0.33–1.3] in the extreme nonresponse group).
This study addressed the increasingly relevant clinical issue of the value of switching from one TNF antagonist to another. This is of particular importance as the justification of switching to an alternative TNF antagonist comes under increasing scrutiny with the availability of newer biologic agents. Furthermore, clarifying the reasons for nonresponse to these drugs should provide further insight into the pathogenesis of RA.
These results suggest that in patients failing to respond to infliximab, switching to etanercept can result in a successful response (the magnitude of such response with regard to overall disease activity score is discussed later). This seems to be irrespective of whether patients demonstrate no response to infliximab by week 12 (primary nonresponse) or lose an initial response (secondary nonresponse), although response after primary infliximab failure appears to be slightly better. In patients who had stopped infliximab treatment due to toxicity, etanercept was safely administered.
Our findings seem to concur with previous studies to date. The Swedish biologics database provided the largest published cohort of patients switching TNF-blocking therapies to date (4). Thirty-one patients were studied, of which 18 received etanercept first and switched to infliximab (most due to inefficacy, but not all) and 13 received infliximab first and switched to etanercept (most [n = 11] due to toxicity). Mean best DAS28 scores as well as ACR-n and SJCs were noted. The etanercept to infliximab cohort demonstrated significant improvements in mean best DAS28 scores (4.8 to 3.6) and SJC on switching to etanercept, with nonsignificant improvements in best ACR-n. The infliximab to etanercept group also demonstrated a significant reduction in best DAS28 scores, with nonsignificant reduction in SJC and unchanged ACR-n. In a comparable study by Gomez-Puerta et al (12) of switching from infliximab to etanercept (n = 12, but in contrast to the Van Vollenhoven et al  study, all were switched due to loss of initial response to infliximab), 83% achieved a good or moderate EULAR response with significant DAS28 reductions. Preliminary results from the EMBARK study (13), a phase 4 study of etanercept in infliximab failures (n = 84: 16 primary nonresponders and 68 secondary nonresponders), demonstrated an overall median DAS28 improvement of 1.2 at 8 weeks.
The present study has the advantage of using a larger cohort and specifically addressing infliximab failures defined by the reason of failure as opposed to a heterogeneous disease cohort without defined reasons for infliximab discontinuation. It is clear that the optimal benefit of etanercept is demonstrated when administered with methotrexate (14). The present study continued with methotrexate enabling the evaluation of switching from infliximab to etanercept under constant conditions and allowing for more accurate comparison.
At our center, a standard infliximab dose is administered to all the patients with no dose increase, unlike many centers both in Europe and the US where the dose is often increased to 5 mg/kg before determining the patient to be a nonresponder. Our standard regimen enables analysis of the kinetics of loss of response; however, the opportunity to reverse loss of response by a higher dose is lost. This latter point may be particularly relevant for the nonresponse patients who demonstrate notable CRP level reduction; we have in fact demonstrated that more than half of this group achieve an ACR response with continuation of 3 mg/kg infliximab for a further 12 weeks (9) (and would likely achieve better response with concomitant dose escalation). The corollary of this is that patients not achieving even 20% CRP level reduction on an infliximab loading dose of 3 mg/kg are unlikely to respond satisfactorily to 5 mg/kg dosing.
Interestingly, the majority (64%) of secondary infliximab nonresponse patients were good (ACR50/ACR70) responders at some point during their history of infliximab treatment, questioning the often held assumption that the patients most likely to relapse while taking infliximab would be the borderline/lower (ACR20) responders. Why this seems to be the case is uncertain, with no obvious explanation on analysis.
The response achieved on switching to etanercept was impressive, with 61% achieving either a moderate or good EULAR response (67% and 56% for primary and secondary infliximab nonresponders, respectively). This result, however, was associated with still moderate to high overall DAS28 scores. The elevated DAS28 scores may be partly explained by longstanding disease duration and the likely associated secondary damage that was sustained, contributing to persistence in notable DAS28 scores. It is important to note that the REFLEX Randomized Evaluation of Long-Term Efficacy of Rituximab in RA study (14), assessing efficacy of rituximab in patients who failed to respond to TNF antagonists, also demonstrated comparable EULAR and DAS28 scores (see below). This highlights the need to use clinical judgment in addition to the specific criteria used in determining response.
The practice of switching TNF-blocking therapies has come under focus with the introduction of newer biologic agents with different mechanisms of action. The costimulation-blocking agent abatacept was approved for license in the US (and is under review in Europe), as was the B cell–targeted therapy rituximab (approved in the US and Europe for TNF blockade failures). Both of these treatments have demonstrated efficacy comparable with the TNF antagonists in a biologic-naive cohort and in patients who failed to respond to TNF blockade. Following 6 months of treatment with abatacept (n = 223) in a placebo-controlled randomized study (n = 258; 2:1 randomization of abatacept to placebo [the Abatacept Trial in Treatment of Anti-TNF Inadequate Responders study]) (15), a significantly greater amount of patients in the treatment arm achieved an ACR20 response compared with the placebo arm: 50% versus 20% (20% versus 3% achieved ACR50 response and 10% versus 1.5% achieved ACR70 response). Impressive quality of life results were also observed in this resistant cohort of TNF antagonist failures. A phase II randomized study of 161 methotrexate-refractory patients demonstrated significantly higher ACR50 responses at week 24 (50%) in the rituximab group compared with the placebo group (16). The more recent phase III randomized, placebo-controlled study (REFLEX study) specifically evaluated the efficacy of rituximab in patients who failed to respond to ≥1 TNF antagonists (14). At 24 weeks, 51% achieved ACR20 responses (27% achieved ACR50 and 12% achieved ACR70). Notably, although 65% achieved either moderate or good EULAR responses, the mean DAS28 score after 24 weeks was still high at 5.1 (reduction of 1.83 from 6.9 at baseline).
Why a proportion of patients fail to respond to one TNF antagonist and respond to the alternative TNF-blocking agent is unclear; however, failure to achieve universal response raises questions on exactly how pivotal TNFα is in all patients with RA. We have previously addressed whether TNF antagonist failures represent an interleukin-1 (IL-1)–driven group, with the convincing clinical observation of a uniform lack of response to anakinra suggesting otherwise (16). These results are further emphasized by comparable levels of IL-1α and IL-1β expression observed at baseline in infliximab responders and nonresponders (Buch MH: unpublished observations).
Differences between infliximab and etanercept may provide further clues as to mechanisms for nonresponse. Infliximab, a chimeric monoclonal antibody, binds to both soluble and membrane-bound TNFα, whereas etanercept also binds to LTα. A previous observation by our group as to the presence of synovial LTα in a patient resistant to infliximab but responsive to etanercept (9) certainly raises the possibility of a role for LTα. In a subgroup of our patients (n = 5) demonstrating total failure to respond to infliximab but a measurable clinical response to etanercept, no significant difference in expression of synovial LTα at baseline (prebiologic) was observed compared with infliximab ACR50/ACR70 responders. Larger synovial studies may be of merit in definitively answering this hypothesis.
In summary, this is the largest study to date evaluating patients who failed to respond to infliximab who switched to etanercept. This comprehensive study provides substantial evidence for the efficacy of switching TNF antagonist therapies to etanercept in the context of infliximab failures. It also provides data for comparison with the newer biologic agents. Greater success rates may be achieved if therapy switching is undertaken in the context of certain profiles of nonresponse to the initial agent. This may also provide insight into reasons for treatment failure and therefore pathogenesis of disease. It is clear that switching is of clinical importance in the management of TNF antagonist failures, especially because there is now evidence that even in nonresponder patients, structural damage is inhibited (17). It remains to be seen whether the strategy of switching from one TNF antagonist to another will provide benefits that are comparable with or indeed superior to other strategies that target alternative mechanisms.
Dr. Emery had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study design. Buch, Emery.
Acquisition of data. Buch, Bingham, Bejarano, Bryer, Quinn, Reece, White.
Analysis and interpretation of data. Buch, Emery.
Manuscript preparation. Buch, Emery.
Statistical analysis. Buch.
- 5Adalimumab (humira®) is effective in patients who have previously been treated with TNF-antagonists (etanercept and/or infliximab) in widespread clinical practice: 12-week outcomes in the ReAct trial. Ann Rheum Dis 2005; 64: S0047., , , , , , et al.
- 7WHO Collaborating Centre consensus meeting on anti-cytokine therapy in rheumatoid arthritis: guidelines for prescribing TNF-α blockers in adults with RA. London: British Society for Rheumatology. June 2001.
- 13Disease characteristics of patients with rheumatoid arthritis that failed to respond to infliximab and the response to etanercept therapy: preliminary data from the EMBARK Study [abstract]. Ann Rheum Dis 2005; 64 Suppl 3: S172., , , , , .
- 14REFLEX trial Group. Rituximab for rheumatoid arthritis refractory to anti-tumor necrosis factor therapy: results of a multicenter, randomized, double-blind, placebo-controlled, phase III trial evaluating primary efficacy and safety at twenty-four weeks. Arthritis Rheum 2006; 54: 2793–806., , , , , , et al, and the
- 17Active-Controlled Study of Patients Receiving Infliximab for the Treatment of Rheumatoid Arthritis of Early Onset (ASPIRE) Study Group. Predictors of joint damage in patients with early rheumatoid arthritis treated with high-dose methotrexate with or without concomitant infliximab: results from the ASPIRE trial. Arthritis Rheum 2006; 54: 702–10., , , , , , et al, and the