Experience With Etanercept, Tocilizumab and Interleukin-1 Inhibitors in Systemic Onset Juvenile Idiopathic Arthritis Patients From the BIKER Registry

Gerd Horneff; Anna Carina Schulz; Jens Klotsche; Anton Hospach; Kirsten Minden; Ivan Foeldvari; Ralf Trauzeddel; Gerd Ganser; Frank Weller-Heinemann; Johannes Perter Haas


Arthritis Res Ther. 2017;19(256) 

In This Article


Within 245 patients with sJIA, 274 treatment approaches using biologics (ETA 143, TOC 71, anakinra 38, canakinumab 22) have been reported. Two patients on ETA had been exposed to a biologic before, compared to 48% in the TOC and 65% in the IL-1i cohort (Table 1). In total, the median age at start of treatment was comparable in all three cohorts (ETA (8.2); TOC (9.6); IL-1i (8.1)) as was the disease duration (3.3, 3.3, 2.8, years, respectively). However, the median disease duration differs between the TOC and IL-1i cohorts when stratifying by the number of previous biological DMARDS (bDMARDs) in the treatment history. Patients had median disease duration of 0.8 years (TOC 0.9, IL-1i 0.6) in first-line use, median disease duration of 5.8 years (TOC 7.4, IL-1i 5.2) in second-line use and median disease duration of 9.0 years in third-line use.

Pre-treatment consisted of systemic steroids in all patients starting biologics (Table 1). Initial concomitant treatment with systemic steroids was significantly less frequent in TOC-treated patients (44%, p < 0.001) and IL-1i-treated patients (45%, p < 0.001) compared to in the ETA cohort (83%). Concomitant MTX was used in the 83% of the TOC cohort, 58% of the IL-1i and in 88% of the ETA cohort.

Systemic manifestations, comprising fever, exanthema, hepatomegaly, splenomegaly or serositis, were present at baseline in 1.4% of patients treated with ETA, in 42.3% of patients treated with TOC and in 63% of patients treated with IL-1i. The mean numbers of active joints (+/- SD) were 9.0 +/- 11.2 with ETA, 5.4 +/- 8.4 with TOC and 2.9 +/- 4.2 with IL-1i. The mean baseline JADAS-10 (+/- SD) was highest in the ETA cohort (20.7 +/- 9.1), followed by the TOC cohort (16.2 +/- 10.6) and the IL-1i cohort (13.4 +/- 9.6).


Response to treatment with ETA, TOC or IL-1i was analysed separately for the presence of systemic symptoms, active arthritis, the JADAS-10 and remission defined as having a JADAS-10 score ≤1 and by the ACR definition of inactive disease.

The number/rate of patients without systemic symptoms increased with TOC (41/58%, 30/86%, 45/94%, 42/93%, 35/95%, 27/96%) and IL-1i (24/37%, 19/68%, 28/78%, 31/79%, 23/74%, 19/83%) at months 0, 3, 6, 12, 18 and 24 months, respectively.

The median (and IQR) of the active joint count and the number (rate) of patients with active joints are given in Table 2, showing a significant decrease in the number and rate of patients with active joints in all cohorts, while the rate of patients with active joints remained higher in the ETA cohort. Among patients with active joints at treatment start, patients treated by TOC had a higher likelihood of controlling the systemic symptoms compared to those treated with IL-1i (beta = 0.14, 95% CI, 0.03–0.60, p = 0.009; TOC, 87%, 97%, 90%, 100%, 95%; IL-1i, 65%, 71%, 77%, 58%, 71%, at months 3, 6, 12, 18, 24, respectively) on follow up. In detail, the difference in the presence of active symptoms between the two groups was significant at month 6 (beta = 0.10, 95% CI, 0.09–1.01, p = 0.051) and month 12 (beta = 0.08, 95% CI, 0.01–0.85, p = 0.036). There was no statistically significant difference between the two cohorts in the response for the control of systemic symptoms (beta = 1.22, 95% CI -0.46–2.91, p = 0.145) in patients without active joints at treatment start.

At start of treatment, there were relatively more patients in the ETA cohort (121(85) with at least one active joint than in the TOC cohort (34/59.6%) or in the IL-1i cohort (49/69%). In the subgroup of patients with at least one active joint at treatment start, patients treated with TOC had a more pronounced decline in the active joint count as compared to those in the IL-1i cohort (beta = 0.19, 95% CI 0.06–0.33, p = 0.004) in follow up adjusted for the active joint count at baseline and the propensity score. A mean active joint count of 1.3 +/- 2.8 and 2.5 +/- 7.7 was reported for TOC and IL-1i (beta = -2.2, 95% CI, -4.1 to -0.4, p = 0.020) at 12 months of follow up and 0.8+/- 2.2 and 2.2 +/- 6.7 at 24 months.

Accompanying the improvement in clinical symptoms, a marked reduction in the use of systemic corticosteroids was observed (Table 2). Systemic steroids had been more frequently used in the ETA cohort than in the two others and a higher rate of patients had to remain on steroids despite biologic treatment.

A marked decrease in the mean JADAS-10 was noted in all three cohorts. The median (IQR) JADAS decreased in the ETA cohort from 20.8 (14.0–28.4) to 6.2 (1.1–14.7), in the TOC cohort from 16.9 (8.1–24.8) to 1.5 (0.2–3.8) and in the IL-1i cohort from 13.0 (6.7–20.6) to 0.6 (0.2–2.0) after 6 months of therapy with a further improvement with ongoing therapy (Table 2). At the last observation, the median JADAS-10 was lower in the TOC and the IL-1i cohorts than in the ETA cohort. The TOC and IL-1i cohorts did not significantly differ (beta = 0.12; 95% CI, -0.11 to 0.34; p = 0.321).

Efficacy analysed in the intention-to-treat dataset is shown in Figure 1. The intention-to-treat analysis revealed that JADAS-remission (JADAS-10 ≤ 1) was reached at month 3, 6, 12, 18 and 24 in 10/93, 21/105, 22/109, 16/110 and 18/109 patients in the ETA cohort, in 8/31, 16/44, 15/46, 15/43 and 15/41 in the TOC cohort and in 13/26, 16/31, 18/38, 16/32 and 14/29 in the IL-1i cohort, respectively. Minimal disease activity (MDA, JADAS-10 ≤ 3.8) was reached at month 3, 6, 12, 18 and 24 in 32, 35, 38, 31 and 29 patients in the ETA cohort, in 14, 27, 23, 26 and 18 patients in the TOC cohort and in 17, 21, 26, 20 and 18 patients in the IL-1i cohort, respectively (Figure 1). At month 6, the difference between the rate of patients on treatment with IL-1i and TOC reaching JADAS-10 MDA compared to ETA was significant (ETA:TOC, p < 0.01; ETA:IL-1i, p < 0.001) as was the rate of patients reaching JADAS-10 remission (ETA:TOC, p < 0.05; ETA:IL-1i, p < 0.001). No significant differences were detected in JADAS-10 MDA in patients treated with TOC compared to patients treated with IL-1i (OR = 1.06; 95% CI, 0.96–1.16; p = 0.262) or in JADAS-10 remission (OR = 1.01; 95% CI, 0.94–1.09; p = 0.783) adjusted for the propensity score.

Figure 1.

Juvenile Disease Activity Score 10 (JADAS-10) remission (defined as JADAS ≤1) and JADAS minimal disease activity (defined as JADAS ≤3.8). The number of patient contributing to the calculation is given below the figure. TOC, tocilizumab; ETA, etanercept; IL-1i, interleukin-1 inhibitor; ACR, American College of Rheumatology

ACR-defined inactive disease was reached at month 3, 6, 12, 18 and 24 in 23/98, 29/114, 29/114, 25/117 and 27/113 in the ETA cohort, in 19/49, 29/63, 32/67, 24/53 and 20/60 in the TOC cohort and in 17/34, 28/43, 28/47, 20/38 and 19/34 in the IL-1i cohort, respectively.

The JIA-ACR responses, especially the high JIA-ACR70 and JIA-ACR90 were reached more often in the TOC and IL-1i cohort compared to the historical comparator ETA. A JIA-ACR70 response at month 3, 6, 12, 18 and 24 was only reached by 32%, 35%, 36%, 33%and 34% in the ETA cohort, by 44%, 46%, 40%, 44% and 40% in the TOC cohort and by 47%, 45%, 44%, 59% and 47% in the IL-1i cohort, respectively. The JIA-ACR90 response at month 3, 6, 12, 18 and 24 was reached by 16%, 19%, 22%, 18% and 19% in the ETA cohort, by 31%, 31%, 27%, 34% and 35% in the TOC cohort and by 34%, 36%, 35%, 51% and 41% in the IL-1i cohort, respectively (Additional File 1: Figure S1).

Discontinuations caused by inefficacy were significantly more frequent within the ETA cohort (43.4%) than within the TOC (8.5%) or IL-1i (21.7%) cohorts. However, TOC was more often discontinued due to intolerance compared to ETA or IL-1i (Table 3). No patients in the ETA cohort discontinued biologic therapy because of reaching remission whereas remission was reported to be the reason for discontinuation in 25.4% of the TOC and 18.3% of the IL-1i cohorts.

Early Effectiveness

Intention-to-treat analysis revealed a significant advantage of TOC and IL-1i over ETA (Figure 1). At month 6, significantly more patients reached JADAS-remission (JADAS-10 ≤ 1) with TOC (36.4%; p = 0.05) or IL-1i (52%; p < 0.001) than with ETA (20%). Furthermore, significantly more patients reached a JADAS-MDA (JADAS-10 ≤ 3.8) with TOC (61%; p < 0.01) or with IL-1i (68%; p < 0.001) than with ETA (33%). There was no statistically significant difference between the TOC and IL-1i cohort in JADAS-remission (OR = 1.18; 95% CI, 0.39–3.57; p = 0.766) or in JADAS-MDA (OR = 1.73; 95% CI, 0.55–5.44; p = 0.349). ETA was more often discontinued due to inefficacy before month 6 (85.7%) than TOC (45.5%) or IL-1i (45.5%).

Effectiveness in Early Disease

There is a trend towards earlier treatment with biologics in sJIA. Therefore the efficacy of TOC and IL-1i in patients with sJIA who were treated within 12 months of diagnosis was compared to those with initiation of biologics later in the course of disease. There were 26 treatments (42%) started with TOC and 22 treatments started with IL-1i (36%) within the first 12 months of the disease. Earlier treatment led to a JADAS-10 ≤ 1 (JADAS-remission) at the last observation in 18 patients (75%) in the TOC cohort and 16 patients (80%) in the IL-Ii cohort. Fewer patients with sJIA who started later treatment with TOC or IL-1i reached JADAS remission (20 (44%) in the TOC cohort and 14 (38%) in the IL-1i cohort). These differences were significant for both TOC (OR 3.9 (95% CI, 1.3–11.6; p = 0.012) and IL-1i (OR 6.6 (95% CI, 1.8–23.7); p = 0.002) (Figure 2a and Figure 2b).

Figure 2.

Juvenile Disease Activity Score (JADAS)-remission (JADAS-10 ≤ 1) and American College of Rheumatology (ACR) inactive disease (Wallace criteria) in patients in the tocilizumab cohort (a) and the interleukin-1 inhibitor (IL-1i) cohort (b) split according to disease duration <1 year and >1 year. At the last report in the early-treated cohort, 18 patients (75%) on tocilizumab and 16 (80%) on Il-1-inhibitors reached JADAS-remission significantly more frequently (OR 3.9 (95% CI, 1.3–11.6); p = 0.012 for tocilizumab and OR 6.6 (95% CI, 1.8–23.7; p = 0.002 for IL-1i) compared to 20 (44%) and 14 (38%), respectively. No difference was noted in the rate of patients reaching ACR inactive disease. Data are expressed as observed. The number of patients contributing to the calculation is given below the figure

There was no correlation between disease duration and the rate of patients reaching ACR inactive disease. A total of 17 patients, 6 in the IL-1i cohort and 11 in the TOC cohort, who had reached JADAS remission, did not satisfy ACR inactive disease criteria due to a physician global assessment indicating values >0 (but below 10 mm on the 100 mm VAS) in 15 patients and a skin rash observed in 2 patients.

Effectiveness in Biologic-naïve Patients Compared to Those Pre-treated

Patients in the TOC cohort, the combined IL-1i cohort and the separate canakinumab and anakinra-cohort were sub-analysed by pre-treatment with ETA, IL-6 or IL-1i (Figure 3). The reported number of patients with no active joints, no fever and either JADAS-remission (JADAS10 ≤ 1) and ACR inactive disease at the last report were used for comparison of biologic-naïve to pre-exposed patients. For the patients on TOC (Figure 3a) and the total patients on IL-1i (Figure 3b), comparison of remission rates (either ACR-defined inactive disease or JADAS-remission) assessed by the chi-square test there was no statistically significant difference between biologic-naïve and pretreated patients. Patients pretreated with anakinra significantly less frequently (p = 0.02) reached JADAS-remission and ACR-defined inactive disease upon treatment with canakinumab (Figure 3c). There were only four patients who were switched from TOC to anakinra. None of them reached ACR-defined inactive disease on anakinra (Figure 3d). Compared to the rate of 0.44 in biologic-naïve patients treated with anakinra, this observation was statistically significant (p = 0.02). No other significant influence of previous failure of biologic treatment was noted.

Figure 3.

Last documented response for the tociizumab (TOC) cohort (a), combined IL-1 inhibitor (IL-1i) cohort (b), canakinumab cohort (c) and anakinra (ANA) cohort (d). The rate of patients with no active joints, no fever, Juvenile Disease Activity Score (JADAS)-remission, American College of Rheumatology (ACR)-defined inactive disease, ACR-remission is given, respectively, in the total cohort and for biologic-naive and pre-exposed patients. Differences in rates were calculated using the chi-square-test. Pearson's p value is outlined if < 0.05. ETA, etanercept


There were 71 adverse events (AE) in the ETA cohort, 118 in the TOC and 81 in the IL-1i cohort (Table 4). Rates of AE (per patient-year) were significantly higher with TOC (risk ratio (RR) 5.3; p < 0.0001) compared to ETA and serious AE were also more frequent with TOC (RR 2.5; p = 0.01) and IL-1i (2.9; p < 0.01) compared to ETA. There were 11 reports of MAS, 3 occurring in the ETA cohort, 5 in the TOC and 3 in the IL-1i cohorts. Rates among the cohorts were not different. Of 110 infections, 11 were reported as serious. 3 were observed in the ETA cohort (2 patients with sepsis, one with gastroenteritis), 2 in the TOC cohort (1 patient with herpes zoster and 1 with pneumonia), 6 in the IL-1i cohort (1 patient with pneumonia, 1 with tonsillitis, 1 with enteritis and 1 with the common cold and 2 with bronchitis). Infection rates were significantly increased in both the TOC (RR 11.0; p < 0.00001) and IL-1i-cohorts (RR 7.4; p < 0.0001) compared to the ETA cohort. While serious infections were rare, rates were numerically higher in the IL-1i cohort. Each of one case of Hodgkin's lymphoma, one of Crohn's disease and one of demyelination occurred in the ETA cohort and one case of depression with suicidal thoughts occurred in the IL-1i cohort. There were two patients in the ETA cohort who died, both at the age of 16 years, due to septic shock and MAS. The events were considered as not related to biologics. One patient died 2 months after MAS was diagnosed and 6 months after discontinuation of ETA due to inefficacy. The other patient succumbed to septic shock and gastrointenstinal haemorrhage while on treatment with ETA. Before, the patient had been heavily pretreated with methotrexate, cyclophosphamide, chlorambucil and mycophenolate mofetil and corticosteroids. The large proportion of neurological signs and symptoms upon TOC were related to a single patient with complex seizure disorder with eight reports of seizures (three were classified as serious as the patient was admitted) but with onset of seizures before the start of TOC.