Juvenile idiopathic arthritis (JIA) is a heterogeneous group of chronic arthritides with clinically distinct disease courses and long-term outcomes.1 Approximately 20–30% of all JIA patients, mostly those with a highly active polyarthritis, a systemic disease, or a vision-threatening uveitis may respond inadequately to conventional disease-modifying anti-rheumatic drugs (DMARD).2^–4 During the past 10 years, biotechnology has brought several immunomodulatory molecules for the treatment of such patients.

Anti-tumour necrosis factor (TNF) treatment, etanercept, a soluble TNFα receptor,5 6 and infliximab, a chimeric monoclonal TNFα antibody,7 8 are effective in JIA. A third anti-TNF agent, a humanised monoclonal TNFα antibody, adalimumab, has recently become available, yet no studies on JIA patients have been published. In a long-term study on rheumatoid arthritis (RA), adalimumab has shown sustained efficacy.9 In addition, infliximab and adalimumab seem to improve the outcome of childhood uveitis.10^–14

All JIA patients do not respond to anti-TNF agents, or the response is insufficient or not sustained. During infliximab treatment adverse drug reactions and loss of efficacy are associated at least partly with the development of antichimeric antibodies.8 In long-term studies, the efficacy of etanercept was sustained in 81% of 43 and 94% of 32 JIA patients still on drug at 2 years6 and at 4 years,15 respectively. In the German etanercept registry, American College of Rheumatology paediatric 30% improvement was achieved at 6 months in 83% of 322 JIA patients and at 30 months in more than 50% of non-systemic JIA patients.16 In a recent study on infliximab, 70% of 112 patients at one year had American College of Rheumatology paediatric 50% improvement.8 So far there are no data available on the long-term efficacy of infliximab, on a long-term comparison between etanercept and infliximab therapies or on the switch between these agents.

We report the drug survival (ie, continuation) rates of anti-TNF agents in 209 consecutive JIA patients and the outcome after the switch from one to another agent.

PATIENTS AND METHODS

Patients

JIA patients on anti-TNF therapy since 1999 for at least one year, and who were younger than 16 years at baseline, were included in a multicentre retrospective chart review. JIA subtypes were classified according to revised International League of Associations for Rheumatology criteria.17 The Finnish Ministry of Social Affairs and Health gave permission for the retrospective study.

Drug therapy

Before 2003 the supply of etanercept was limited and patients started infliximab off-label. The decision to initiate anti-TNF therapy on top of concomitant DMARD, or to switch from one biological agent to another was at the discretion of the paediatric rheumatologist. During clinical remission combination DMARD were tapered and TNF antagonists were withdrawn before discontinuing the last single DMARD (usually methotrexate).

Assessment of disease activity

All patients were unresponsive to previous therapy and had increased disease activity based on the number of active joints, inflammation markers and/or active uveitis.18 Patients were divided into two categories according to their baseline disease activity parameters: high or moderate (table 1).18 Inactive disease was defined as in Wallace et al.19

Statistical analysis

Differences between baseline characteristics of patients taking etanercept or infliximab were compared using Fisher’s exact test for category variables, χ² for dichotomous variables and Mann–Whitney U test or Student’s t test for continuous variables. Two separate analyses were performed, the first on all patients with their first anti-TNF agent and the second focusing on patients during the second anti-TNF agent. Kaplan–Meier analyses illustrated drug survival, ie, continuation rates (%) on a particular drug. The comparison of differences between the first-line anti-TNF agent were based on log-rank statistics (non-adjusted, gender and disease activity-adjusted models), and the comparison between the first and second course of treatments were based on McNemar’s test. The predictors for treatment discontinuation were assessed by proportional hazards model (Cox’s regression) using 95% CI for hazard ratios (HR). The factors included in Cox models were the baseline anti-TNF agent, age, duration of JIA, gender, subtype of JIA, rheumatoid factor, antinuclear antibody (ANA), human leucocyte antigen B27 (HLA-B27), uveitis, C-reactive protein and erythrocyte sedimentation rate, number of DMARD, number of active joints, prednisolone/kg, and in the analyses of the second-line anti-TNF agent, also the first anti-TNF agent and the reason to discontinue the first agent.

RESULTS

Patient characteristics

Of 258 consecutive patients taking anti-TNF treatment, 209 met the inclusion criteria: 104 patients taking infliximab and 105 taking etanercept. The patients on infliximab more often had JIA-associated uveitis and persistent oligoarthritis, whereas those on etanercept had more seronegative polyarthritis or systemic JIA and they were younger at JIA onset (table 2). The mean time on the first anti-TNF agent was 30 months (SD 21, range 1–97), no statistical difference existed between the two treatment groups. At baseline, the number of patients in treatment groups with high or moderate disease activity did not differ significantly (p = 0.155). The mean follow-up of all patients on/off therapy was 46 months (range 12–103).

Males had seronegative polyarthritis less often (8% vs 42%, p<0.001), more enthesitis-related arthritis (ERA) (10% vs 0%, p<0.001) and ANA positivity less often (9% vs 34%, p = 0.022) than females. In males, the onset of JIA occurred later (6.2 vs 4.6 years, p = 0.018) and the duration of JIA was shorter (4.2 vs 5.4 years, p = 0.007). No difference existed between males and females in baseline disease activity parameters.

At baseline, 205 (98%) patients were on DMARD; 57 (27%) on single and 135 (64%) on combination therapy. Of the 205, 166 (85%) were on methotrexate, 71 (36%) on hydroxychloroquine, 60 (31%) on cyclosporine, 51 (26%) on sulfasalazine, 30 (15%) on azathioprine and 137 (65%) on prednisolone (mean daily dose 8 mg or 0.3 mg/kg). Specific data on DMARD in 14 patients were missing.

Drug survival of the first-line anti-TNF agent

During the 48-month follow-up, no difference in drug survival rates was observed between infliximab and etanercept (p = 0.194 in log-rank test, fig 1A). This was true also in gender and disease activity-adjusted models (p = 0.092 and p = 0.262, respectively). The proportion of patients continuing with etanercept or infliximab treatment was 83% (95% CI 76% to 90%) or 80% (95% CI 72% to 87%) at 12 months, 68% (95% CI 59% to 78%) or 68% (95% CI 58% to 77%) at 24 months, 64% (95% CI 54% to 74%) or 53% (95% CI 42% to 64%) at 36 months, 61% (95% CI 50% to 71%) or 48% (95% CI 36% to 59%) at 48 months and 53% (95% CI 41% to 65%) or 45% (95% CI 33% to 57%) at 60 months, respectively. In this analysis, withdrawal from anti-TNF agents due to inactive disease was not considered as treatment discontinuation.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Drug survival (%) for juvenile idiopathic arthritis patients treated with the first biological therapy, etanercept (n  =  105) or infliximab (n  =  104). Treatment discontinuation during a 48-month follow-up due to all reasons except inactive disease (A), adverse events (B), inefficacy (C) and inactive disease (D).

Predictors for discontinuation of anti-TNF agents were female gender, systemic JIA and infliximab as the first therapy. The risk of discontinuing anti-TNF agents was threefold in girls compared with boys (HR 2.8, 95% CI 1.3 to 5.8, p = 0.006). Infliximab doubled the risk (HR 2.0, 95% CI 1.2 to 3.3, p = 0.004) compared with etanercept and the risk was eightfold in systemic JIA (HR 7.8, 95% CI 1.7 to 34.9, p = 0.007) compared with non-systemic JIA.

Treatment survival between etanercept and infliximab was comparable in patients with seronegative polyarthritis (p = 0.080), seropositive polyarthritis (p = 0.578), ERA (p = 0.548) and systemic JIA (p = 0.433). Infliximab was discontinued more often in patients with extended oligoarthritis (p = 0.019). In the remaining subtypes, cases were too few for meaningful statistical comparisons. Anti-TNF treatment survival rates in JIA subtypes at 24 and 48 months were 92% and 92% in patients with oligoarthritis, 91% and 78% in ERA, 75% and 65% in extended oligoarthritis, 65% and 52% in seropositive polyarthritis, 62% and 47% in seronegative polyarthritis and 54% and 24% in systemic JIA, respectively.

Discontinuation of the first anti-TNF treatment

During a 48-month follow-up, the first anti-TNF therapy was discontinued in 43% (46/105) of patients taking etanercept and in 59% (61/104) taking infliximab (table 3).

Discontinuation due to adverse events

Side-effects were the most common (93%) adverse events. Two patients discontinued the therapy due to fear of injections. Treatment discontinuation due to adverse events occurred more frequently during infliximab than etanercept (table 2, fig 1B, p = 0.002), also in the gender and disease activity-adjusted models (p = 0.001 and p = 0.002). Predictors for discontinuation due to adverse events were infliximab therapy (HR 4.6, 95% CI 1.8 to 11.8, p = 0.002) and female gender (HR 5.4, 95% CI 1.1 to 26.9, p = 0.039). In addition, the presence of HLA-B27 (HR 0.19, 95% CI 0.05 to 0.67, p = 0.009) predicted lower risk, ie, HLA-B27-positive patients had an 81% lower risk of discontinuing anti-TNF agents due to adverse events. However, when 64 males, of whom 38 were on infliximab, were analysed independently, discontinuation rates due to adverse events between infliximab and etanercept treatment groups were not different (p = 0.638 in log-rank test).

Discontinuation due to inefficacy

Treatment discontinuation due to inefficacy between etanercept and infliximab was comparable (fig 1C, p = 0.445) and also in gender and disease activity-adjusted models (p = 0.600 and p = 0.350). Systemic JIA (HR 4.2, 95% CI 1.3 to 14.1, p = 0.020) and seronegative polyarthritis (HR 2.2, 95% CI 1.0 to 4.8, p = 0.048) but not other subtypes predicted discontinuation due to inefficacy.

Discontinuation due to inactive disease

The rate of discontinuation because of inactive disease or remission seemed to be slightly higher during infliximab than during etanercept therapy. This difference was, however, not significant in non-adjusted (fig 1D, p = 0.068) or gender-adjusted (p = 0.059) or disease activity-adjusted (p = 0.092) models. The only predictor of the discontinuation of the anti-TNF agent due to inactive disease was the duration of JIA at the onset of the anti-TNF agent (HR 0.77, 95% CI 0.64 to 0.94, p = 0.008). In other words, the shorter the interval between the onset of JIA and the initiation of the anti-TNF agent, the more frequently the therapy was discontinued due to inactive disease. Of 27 patients discontinuing the first anti-TNF agent due to inactive disease, eight of 10 (80%) with previous infliximab and seven of 17 (41%) with previous etanercept treatment relapsed.

Switch to another biological agent

During a mean follow-up time of 53 months (SD 22), altogether 73 (35%) patients switched therapy. The proportion of females was higher (p<0.001) among those who switched therapies (88% of 73) than in non-switchers (60% of 136). The majority (77% of the 73 patients) switched from etanercept to infliximab or vice versa, 15 patients switched to adalimumab and two patients to anakinra. Drug survival of etanercept, infliximab or adalimumab as the second biological agent was 84% (95% CI 70% to 98%) versus 58% (95% CI 39% to 76%) versus 73% (95% CI 51% to 96%) at 6 months and 60% (95% CI 41% to 79%) versus 58% (95% CI 39% to 76%) versus 66% (95% CI 42% to 90%) at 12 months, respectively. Drug survival of etanercept or infliximab was 48% (95% CI 26% to 70%) versus 32% (95% CI 14% to 50%) at 24 months and 38% (95% CI 14% to 65%) versus 26% (95% CI 7% to 44%) at 36 months, respectively. Treatment survival of etanercept, infliximab or adalimumab as the second anti-TNF agent was comparable (p = 0.196 in log-rank test, fig 2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Drug survival (%) during the first course (etanercept or infliximab, solid line) and the second course (dashed line) of biological agents in 73 juvenile idiopathic arthritis patients who switched therapy.

In these 73 patients, the discontinuation rate due to adverse events or inefficacy during the first course of biological therapy was 96% and during the second course of therapy it was 53% (p = 0.001 in McNemar’s test). The mean treatment time was 18.4 months (95% CI 14 to 22) during the first and 16.3 months (95% CI 13 to 20) during the second course of therapy (p = ns). The second biological agent was discontinued in 39 (53%) patients: in 13 patients due to adverse events and in 26 due to inefficacy (table 4). Furthermore, four patients discontinued the second biological agent because of inactive disease.

Predictors for discontinuation (due to adverse events or inefficacy) of the second course of biological agents were systemic JIA (HR 4.5, 95% CI 1.8 to 11.3, p = 0.002) and failure of the first course with etanercept (HR 2.2, 95% CI 1.4 to 4.2, p = 0.019). In addition, the risk of discontinuing the second biological agent due to adverse events was higher in patients who had also discontinued the first therapy due to adverse events (HR 6.8, 95% CI 1.6 to 28.7, p = 0.009) and in those who failed the first course of etanercept (HR 12.6, 95% CI 2.5 to 64.9, p = 0.002). Predictive factors for discontinuing the second biological agent due to inefficacy were systemic JIA (HR 9.4, 95% CI 3.1 to 28.5, p<0.001) and negative ANA (HR 2.8, 95% CI 1.2 to 6.5, p = 0.019).

Re-start of the first biological agent

In 31 of 209 (15%) patients, the first anti-TNF treatment was re-started after a disease flare following an earlier treatment termination (table 4). The re-started therapy was considered successful if the patients continued it during the follow-up. The success rate was similar between etanercept and infliximab (p = 0.120).

Infliximab doses

Infliximab was discontinued due to inactive disease on a mean dose of 3.6 mg/kg (range 2.6–5.8) and an infusion interval of 9.3 weeks (range 8–12). Infusion reactions were observed in 22 patients at a mean time of 9.6 months (range 0.5–35) from onset with a dose of 3.1 mg/kg (range 2.2–5.0) and a dose interval of 6.2 weeks (range 4–8). Compared with those with inactive disease, the patients with infusion reactions received a similar dose (p = 0.147) with a shorter dose interval (p = 0.001). During the first or second course of infliximab in 65 patients without hypersensitivity reactions the mean dose interval was 7.3 weeks (range 4–12) and the mean dose 3.9 mg/kg (range 1.9–11.1). The latter was 0.8 mg/kg (95% CI 0.2 to 1.0, p = 0.025) higher compared with the doses of those with infusion reactions. The only predictor for infusion reactions was female gender (HR 12.7, 95% CI 1.7 to 94, p = 0.103).

DISCUSSION

In a real-life setting, the 4-year anti-TNF treatment survival of etanercept and infliximab in JIA patients was comparable. During infliximab therapy, however, the discontinuation rate due to adverse events, but not due to inefficacy or inactive disease, was higher than during etanercept. Although our patient series reflected the heterogeneous population in clinical practice, the drug survival rates for the first anti-TNF therapy were in line with those seen in prospective efficacy studies on etanercept15 and infliximab.8 Compared with the prospective series of Ruperto et al,8 treatment discontinuation due to adverse events during infliximab treatment in our patient series occurred more often and earlier. Between 1999 and 2004, we used the recommended 3 mg/kg infliximab doses, which are associated with infusion reactions more frequently than 6 mg/kg doses.8 This may at least partly explain the difference in discontinuation rates.

To the best of our knowledge, our study may be the first to evaluate a switch between anti-TNF therapies in juvenile arthritides. Of all JIA patients who discontinued the first anti-TNF agent due to adverse events or inefficacy, approximately half were able to continue the second anti-TNF agent until the end of the follow-up. In a UK register study, 73% of RA patients switching biological agents continued the second anti-TNF agent for a mean of 6 months,20 consistent with our findings. Although adverse events in our series were different from those in the UK register,20 discontinuation of the first anti-TNF agent due to adverse events was related to discontinuation of the second agent due to adverse events in both series. The inefficacy of the first anti-TNF agent did not seem to predict inefficacy of the second anti-TNF agent in JIA, whereas in RA discontinuation rates due to inefficacy during the first and second course of anti-TNF agents were associated.20 This suggests that in JIA switching from one anti-TNF agent to another can be considered a reasonable therapeutic option.

Of the JIA subtypes, treatment termination due to inefficacy during both the first and second anti-TNF therapy was strongly associated with systemic onset JIA. This is not surprising, because the response of biological therapy in JIA seems to be subtype specific. In systemic JIA only a few studies have evaluated the efficacy of etanercept, but little is known about the efficacy of other anti-TNF agents. Kimura et al21 reported a poor or fair response to etanercept in more than 50% of 45 systemic JIA patients. Other studies on systemic JIA documented discontinuation rates of etanercept of up to 42%,6 16 which in our cohort were even higher: during the first course 69% and during the second course of anti-TNF agents 100% (data not shown).

In patients with ERA, of whom 56% were HLA-B27 positive, the discontinuation rate was markedly low, only 14%. This concurs with previous findings of 19% in 26 patients with ERA in the German etanercept registry,16 with sustained efficacy of etanercept over 2 years in eight ERA patients (seven HLA-B27 positive) in the study of Henrickson and Reiff,22 and with sustained efficacy of anti-TNF therapy over one year in 10 HLA-B27-positive patients with juvenile spondylarthropathy.23 In our cohort, HLA-B27 positivity and male gender, but not ERA, were protective factors against treatment discontinuation due to adverse events. This finding may be partly explained by the higher proportion of females among the less well-responding and HLA-B27-negative patients with systemic JIA or highly active polyarthritis, but may also be an independent factor and thus warrants further investigation. In RA, during the use of biological agents women are less likely to achieve remission.20 In JIA, female gender has been suggested as a predictor of disability24 and continued disease activity.25

Some preliminary evidence in adults exists about the benefits of early aggressive anti-TNF treatment to induce remission. In the BeSt study, compared with more conservative treatment strategies the combination of initial infliximab and methotrexate led to a higher remission rate and more often to discontinuation of all antirheumatic drugs.26 Interestingly, in the present study the initiation of anti-TNF treatment early in the course of JIA was associated with inactive disease, which suggests indirectly that early anti-TNF treatment may induce clinical remission.

Our series represents a real-life cohort of anti-TNF therapy users, with the limitations being those of many retrospective and register studies. The high proportion of patients with JIA-associated uveitis adds heterogeneity to the present series, and must be taken into consideration in the evaluation of drug survival rates, because infliximab has shown higher efficacy in childhood uveitis than etanercept.10 14

During the 4-year follow-up the overall drug survival of etanercept and infliximab as first-line anti-TNF agents was comparable and satisfactory even as second-line anti-TNF agents. Among the subtypes of JIA, the risk of treatment discontinuation was highest in systemic arthritis. Female gender was an independent risk factor for treatment discontinuation. Based on our results, we recommend a switch to another anti-TNF agent for non-systemic JIA patients who fail the first anti-TNF agent. In future, other biological compounds may provide therapeutic options for JIA patients refractory to anti-TNF agents.