Efficacy and Safety of Tocilizumab in a Real-Life Observational Cohort of Patients With Polyarticular Juvenile Idiopathic Arthritis

Minna-Maija Grönlund; Terhi Remes-Pakarinen; Liisa Kröger; Kati Markula-Patjas; Maria Backström; Anne Putto-Laurila; Kristiina Aalto; Paula Vähäsalo


Rheumatology. 2020;59(4):732-741. 

In This Article


The current recommendations suggest TNF inhibitors as the first choice of treatment in children with pJIA if sDMARDs have failed; thereafter another TNF inhibitor or abatacept is recommended for the second-line biological treatment.[7] However, the evaluation underpinning these recommendations did not include studies of tocilizumab for pJIA as no such studies were available at that time. Recent German pJIA therapy algorithms include tocilizumab in the first-line use of biological agents,[9] as do the Childhood Arthritis and Rheumatology Research Alliance (CARRA) Consensus Treatment Plans.[8] The pathways of using biological agents in clinical practice are constantly changing[10,20] and vary considerably in different countries due to local regulations and availability of the biological agents.[11,12,20–22] Since there is rather short experience of tocilizumab in the treatment of pJIA, only few register and observational reports have been published on the safety and efficacy of tocilizumab so far.[11,12,16]

This study evaluated a nationwide cohort of 56 consecutive Finnish pJIA patients starting tocilizumab treatment between 2013 and 2017. Tocilizumab was on average a third-line biological treatment, the median disease duration was 5.2 years and the median JADAS-10 was 7.6 at tocilizumab commencement, depicting a cohort with treatment-resistant polyarthritis. Taking this into account, drug survival at 12 months (82%) was high. It was higher than reported on TNF inhibitors as second-line or third-line biological therapy at 12 months in non-systemic JIA (50–54%).[23] A higher survival of tocilizumab compared with TNF inhibitors was also detected in the BiKeR registry, a study that included patients in exactly the same JIA categories as ours and mostly using tocilizumab as at least second-line biological agent.[12] Nevertheless, a steady decline in the survival rate of tocilizumab was noted throughout the follow-up in our study, ending at 64% at 24 months. Tocilizumab was discontinued solely due to inadequate treatment effect in half of the cases, whereas AEs were associated with discontinuation in the other half. Horneff et al.[12] reported discontinuation of tocilizumab due to inefficacy in 39% and intolerance in 9% of the cases, while in 43% of the cases the reason was not specified. In the study by Otten et al.[23] a second-line biological agent was discontinued due to inefficacy in 64% and to AE in 23% of the cases; however, these results are not readily comparable to ours since their study included 50% of patients with systemic JIA.

It is widely recognized that sDMARDs are not well tolerated in JIA patients.[24] Accordingly, the rate of tocilizumab monotherapy increased in this study from 16% at tocilizumab commencement to 38% at 24 months. The rate of tocilizumab monotherapy in our study was still lower than the 54% rate found in the BiKeR registry.[12] In the original placebo-controlled withdrawal study of 188 pJIA patients (the CHERISH study) the response rates were higher in the group of JIA patients receiving a combination of tocilizumab and MTX compared with tocilizumab monotherapy. However, tocilizumab was very efficacious even as monotherapy; at 40 weeks the ACRpedi70 response was achieved in 67% and 53% of patients with and without MTX, respectively.[25] Conversely, we found no difference in the efficacy of tocilizumab (assessed by JADAS-10 scores) at 12 months of follow-up between patients with monotherapy and those with tocilizumab and sDMARD combination. Still, this study may have been too small to discover subgroup differences. Interestingly, it was found in the BiKeR registry that the survival rate of biological treatment (etanercept, adalimumab and tocilizumab) at 5 years' follow-up was the same in pJIA patients with biological monotherapy and with MTX combination therapy.[12]

According to the recent recommendations on the treatment goal of JIA, LDA is considered a valid target for patients with longstanding, persistently active pJIA.[26] This resembles the disease situation in our patients at tocilizumab commencement. There are only few studies reporting treatment outcomes for this kind of patient. Using the recent Finnish JADAS-10 cut-off values[18] a large proportion of the patients remaining on tocilizumab reached LDA at 12 months (58%) and 24 months (84%). If the cut-off values of Consolaro et al.[27] are used, the treatment responses are the same or somewhat higher in our cohort (data not shown) compared with the pJIA patients in the BiKeR registry using tocilizumab; at 12 months LDA was reached in 58% in our cohort vs 61% in the BiKeR study and remission in 28 vs 21%, respectively, and at 24 months LDA in 84 vs 62% and remission in 48 vs 23%, respectively.[12] Clinically inactive disease at 12 months was reached in 19% (by JADAS-10 ≤0.7) to 28% (by Wallace's preliminary criteria) of our patients, which is higher than that previously reported in patients switching biological agents (7–17%).[23,28] The improvement in the clinical outcome of our study patients was also reflected in the decreased usage of different sDMARDs, systemic and intra-articular glucocorticoids, and in the lower pain-VAS.

We found a higher rate of SAEs (12.9/100 PYs) for tocilizumab than that reported in a recent European retrospective cohort of paediatric patients with inflammatory rheumatic diseases (8.2/100 PYs, total exposure 245 PYs)[29] and that of pJIA patients in the BiKeR registry (4.1/100 PYs, total exposure 72.5 PYs).[12] The largest observational safety report on tocilizumab in JIA has been that of 435 PYs from the BiKeR registry, where the rate of SAEs was 17.3/100 PYs.[16] However, this congress abstract probably also includes patients with systemic JIA, which would explain the high number of SAEs.[16,30] Nevertheless, the rate of SAEs in our study was comparable to that of another retrospective observational cohort of Finnish JIA patients treated with biological agents (11.4/100 PYs), of whom 39% were biological agent switchers.[31] As pointed out by Tarkiainen and colleagues, data obtained from patient charts seem to be a more reliable source of information on AEs than register data.[31] This may be due to a combination of more frequent data collection available from all hospital visits and the public health care system in Finland, where all visits that require hospital treatment are very likely to be handled by the same tertiary hospital taking care of these JIA patients. Actually, the rate of SAEs in our study was comparable to the CHERISH study (12.5/100 PYs), but the rate of AEs was clearly lower than in CHERISH (480/100 PYs).[25] The extended safety data of 2 years' follow-up in the CHERISH study found no change in the safety profile of tocilizumab,[32] neither did the extension data of Polish and Russian JIA patients included in the CHERISH study.[33] Previous studies on tocilizumab safety in European JIA patients have not reported any cases of malignancies, tuberculosis, demyelinating disorders or deaths,[12,25,33] which was also the case in our study. One exception to this has been published recently by Horneff et al.,[16] who reported one malignancy.

It could be hypothesized that the use of multiple, consecutive biological agents could have caused the higher rate of SAEs and AEs in this study. However, previous reports on JIA do not support this notion. No difference was found in the rate of AEs or SAEs between the JIA patients on first-line biological agents and those switching biological agents in the BiKeR registry; in contrast, more infectious AEs were found in the first-line biologic group compared with the biologic-switcher group.[28] In addition, the rate of AEs and SAEs were comparable during the first to the third course of biological agents in JIA patients of the Dutch ABC Register.[23]

The rate of specific AEs is difficult to compare between studies of biological treatments due to the differences in the classification of AEs and in the type of the study (i.e. a single-centre, registry or placebo-controlled study, or a long-term extension cohort).[29,34] Furthermore, the rates of AEs associated with tocilizumab differ depending on the indication (i.e. systemic JIA or pJIA).[25,30,35] However, one typical AE for tocilizumab is neutropenia.[35] We detected a higher rate of grade 3 neutropenia (≤1.0×109/l) in 18% of our patients compared with 5.9% reported in the CHERISH study patients during 2 years of follow-up.[35] Pardeo and colleagues found no association between development of neutropenia and tocilizumab trough levels or concomitant MTX use, and neutropenia was not associated with the development of infections.[35] The difference in the rate of neutropenia might be explained by the fact that many of our patients used concomitant sDMARDs other than MTX as well as combinations of sDMARDs, while only MTX and glucocorticoids were allowed in the CHERISH study.[25] sDMARDs other than MTX during biological treatment have been associated with higher rates of AEs and SAEs.[29]

No new-onset uveitis was found in our patients during tocilizumab treatment. Three patients with severe recalcitrant uveitis with two to three previous pre-tocilizumab biological treatments had active uveitis throughout the follow-up. A recent retrospective study by Calvo-Río and colleagues reported a complete remission of severe uveitis that was refractory to conventional DMARDs and at least one anti-TNF agent in 19 of 25 patients after 12 months of tocilizumab treatment.[36] The first prospective study on tocilizumab efficacy in JIA-associated anti-TNF agent refractory uveitis is being conducted at present (the APTITUDE trial).[37]

There are several limitations to our study. As this was an observational study there were some missing data and the follow-up was shorter than 24 months for some patients. However, most of data regarding the JIA core set variables were available since they were collected prospectively to the JIA databases. In addition, the study population was rather small, 56 patients and 85.1 PYs, and included only one seropositive polyarthritis patient. It is therefore not possible to generalize our results to this or other missing categories of JIA. Nevertheless, this is one of the largest real-life studies reporting on the efficacy and safety of tocilizumab in pJIA patients.

In conclusion, our study showed that survival of tocilizumab was good and a large proportion of the hard-to-treat pJIA patients reached LDA at 12 months of treatment. The rate of LDA continued to increase among those remaining on tocilizumab throughout 24 months. Patients needed significantly fewer synthetic DMARDs, and systemic or intra-articular glucocorticoids during the 24-month follow-up. The rate of AEs and SAEs was higher than that in previous register studies but lower than that of the original progenitor trial of tocilizumab in pJIA.