Disease evolution in systemic juvenile idiopathic arthritis: an international, observational cohort study through JIRcohort

Systemic juvenile idiopathic arthritis (systemic JIA) is a pediatric inflammatory condition classified within the spectrum of juvenile idiopathic arthritis (JIA). Systemic JIA is distinguished from other JIA subtypes due to its particularity of having a variety of systemic symptoms: persistent (> two weeks) undulating fever with a paroxysmal pattern, rash, lymphadenopathy, serositis (i.e. pericarditis, pleuritis or peritonitis) or organomegaly (hepato- and/or splenomegaly) [1]. Patients are also at risk of developing potentially fatal systemic inflammatory complications, such as macrophage activation syndrome, which has resulted in systemic JIA having the highest mortality rate (3.9/1000 person years) of all JIA subtypes [2, 3].

Despite recent medical advances, systemic JIA diagnosis is often challenging as it remains an exclusion diagnosis. Its heterogeneous and nonspecific clinical presentation, as well as the lack of specific laboratory markers, create difficulties for a rapid and accurate diagnosis. Systemic JIA reportedly accounts for 10% of all JIA cases in Europe, however its impact is likely under-represented due to the current classification criteria which have been criticized [4, 5]. New preliminary classification criteria have been recently proposed by an international consensus [6].

Clinical presentation of systemic JIA was studied by Behrens et al. in 136 children. The most frequent symptoms at first presentation were fever followed by arthritis and cutaneous rash, all diagnosed in over 75% of the children included. Lymphadenopathy, organomegaly and pericarditis were seen in much fewer patients. Articular involvement was oligo- or polyarticular in over 80% of cases, whereas the most frequent joints involved were the wrists, knees and ankles [7].

Disease evolution may be variable in systemic JIA. Three disease evolutive patterns were first described by Lomater et al. as either monocyclic, intermittent or persistent [8]. Singh-Grewal et al. are one of the few groups to examine the predictors of the disease evolution in systemic JIA. According to them, the only factor at diagnosis that is predictive of a non-monophasic (i.e. intermittent or persistent) evolution was the presence of a polyarticular arthritis. They described additional predictors of a non-monophasic course at three months (i.e. fever and active arthritis) and six months post diagnosis (i.e. erythrocyte sedimentation rate, ESR > 26 mm/hour and the persistent use of corticosteroids) [9]. Spiegel et al. identified thrombocytosis and persistent active systemic disease (six months after onset) as strong predictors for a poor functional outcome [10]. Persistent active systemic disease has proven to be a particularly important prognostic marker [8, 11]. So far, our knowledge of the early predictors in systemic JIA disease evolution is based on very limited data, which is fractured across groups with varying diagnostic definitions.

This international cohort study aimed to identify predictors of systemic JIA disease evolution in a real-life condition. We provided detailed diagnostic phenotypes of 65 enrolled patients including clinical, laboratory and treatment variables. We then examined univariate and multivariable correlations of these variables with the disease progression.

Systemic juvenile idiopathic arthritis is a chronic disease that can remain active over several years resulting in significant morbidity for young adults with an impact on their social, professional and financial future. Although it has been established that there are three different types of disease evolution, information on the predictors of the disease course is limited. The present study was designed to describe the occurrence and clinical course of the different disease evolution types as well as to investigate the prognostic capacity of clinical features to better predict a chronic evolution.

First, these results show that our systemic JIA population presented all the three previously described disease evolution types: monophasic, polyphasic and persistent evolution. More than half of the patients were classified as having persistent disease (55.4%), followed by the monophasic group (35.4%), with only a small portion having a polyphasic evolution (9.2%). These findings are in line with the results from a prospective observational study that analyzed 45 patients treated in an academic rheumatology center in Canada between 1996 and 2000 [9], but differ from a more recent retrospective study including 76 patients treated in Boston Children’s Hospital between 1996 and 2015 were lower percentage of persistent disease course was reported [15]. Because of the low number of patients in the polyphasic group, this group was combined with the monophasic one (non-persistent group) for the assessment of potential predictors for a persistent disease course.

Demographic features, such as age at disease onset, sex and time to follow-up, are consistent with findings in previous studies [7, 11]. Concerning clinical presentation, most children in this cohort presented initially with fever and/or arthritis, while other systemic symptoms such as cutaneous rash, lymphadenopathy, organomegaly or serositis were less frequent. These results are in agreement with a retrospective observation study that included 136 patients diagnosed with systemic from three tertiary rheumatology referral centers in the United States of America between 1990 and 2005, as well as with the previously reported retrospective study including 76 patients treated in Boston Children’s Hospital besides rash that was among the most common symptoms following fever and arthralgia in this systemic JIA cohort [7, 15].

Importantly, early disease onset (< 6 years old) was significantly more frequently observed in the persistent group and was identified as a presumed predictor of persistent disease evolution, in line with the results of Flato et al. indicating early disease onset as a risk factor for poor outcome in JIA [16]. However, after adjusting for potential confounders this association was no longer statistically significant, highlighting the need of larger studies to address this question.

Furthermore, we observed a higher prevalence of polyarthritis in the persistent group during the first 6 months of disease evolution, with the patients among this group being more likely to have active at 12 months. In addition, the visual analogue scale (VAS) score at 12 months evaluated by patient and the physician’s global assessment (PGA) was significantly higher in the persistent group, reflecting the increased disease activity and severity in the persistent disease evolution group. Regarding the systemic features, our data show a higher trend of rash and splenomegaly in the non-persistent group at diagnosis, with, on the other hand, a higher frequency of fever, lymphadenopathy and rash in the persistent group at 12 months, but without any statistical significance.

The duration of elevated inflammatory markers during the first year of disease evolution differed according to the disease course, with patients in the persistent group having significant higher CRP levels at 12 months.

Moreover, use of sDMARD was identified as a predictor of persistent disease evolution in our systemic JIA cohort. In line with that, Beukelman et al. reported, in 2012, a higher use of methotrexate and ciclosporin in systemic JIA patients with polyarthritis and children with radiologic damage, respectively [14]. Therefore, an important use of sDMARDs in patients with persistent disease course could retrospectively be interpreted as an indirect sign of a greater articular involvement which is expected in this group. The long treatment duration in the persistent group is interlinked with the definition of the group but should also be seen as an evidence of how difficult treatment of these chronically affected children is. The long treatment duration in the polyphasic group is due to mostly several relapses of the disease resulting in a therapeutic intervention. We also provide evidence that therapeutic strategies remain heterogeneous and are based on exacerbation if the initial essay is not effective. Furthermore, we show that glucocorticoids remain an important baseline therapy; however, biologic agents are faster and more frequently used in the therapeutic procedure due to important scientific evidence of their efficacy and safety in systemic JIA [17‐20]. Baris et al. showed that the duration of steroid treatment correlated with the duration from disease onset to first bDMARDs initiation, mostly related with anakinra and canakinumab [15]. Anti-IL-1 agents (anakinra, n = 32; canakinumab, n = 25) were the most prevalent therapeutic group in the bDMARD category in our study. However, tocilizumab (anti-IL-6R monoclonal antibody, n = 33) was highly used as well and was significantly more often prescribed in the persistent group. Without drawing any conclusions about the efficacy of Tocilizumab, one could hypothesize, that one single target therapy may not be enough to achieve inactive disease, as an established disease may result from complex immunological pathways and a multitude of cytokines (e.g. IL-1β, IL-6, IL-18, IL-17) implicated in the inflammatory process of systemic JIA. Therefore, it may be the high complexity of the aberrant innate immune system in this pathology rendering therapy and prediction of disease evolution difficult [21‐23].

There are several limitations in this study. One main limitation is the non-standardized retrospective data collection due to limited time and resources within the framework of this study. Therefore, patients included in this study, despite the complete information on disease activity, had nonetheless missing information on medication use, such as non-reported glucocorticoid or Non-steroidal anti-inflammatory drugs (NSAIDs) therapy, missing information on laboratory values as well as missing PGA and VAS evaluations by patients. Another main limitation is that excluded patients may differ from the included population leading to a potential selection bias, even if the basic socio-demographic and clinical data from baseline were comparable in both included and excluded patients (Table 1). More cases with available information would have increased the statistical power of the multivariable models. However, our univariate findings, with the significantly higher frequency of arthritis and the significantly higher rate of elevated inflammatory markers during the first year post-diagnosis in the chronic disease evolution group, are tendencies that mirror identified predictors by a multiple regression model in previous studies [8, 9, 11, 21].

In our model parameters like young age at disease onset (< 6 years) and active arthritis at 12 months, were found to potentially have an influence on the risk of a persistent disease evolution during the first 24 months after diagnosis, even if the association was no longer statistically significant after adjusting for confounding factors. The need of sDMARD use may also predict a persistent disease course, reflecting a more severe phenotype. Further research should concentrate on standardized definitions of inactive disease and remission, which would enable combining harmonized data in meta-analyses. Moreover, future JIA research needs to be conducted on a larger sample with stringently prospectively collected data in order to generate enough statistical power to confirm or reject the presented tendencies within this study.