Systemic juvenile idiopathic arthritis (SJIA) is a rare and serious autoinflammatory disorder characterized by systemic inflammation (hectic quotidian fevers, typical rash, serositis, hepatosplenomegaly, lymphadenopathy, acute-phase reaction) and variably accompanied or followed by chronic arthritis [1–3].

The hypothesis exists that early effective treatment of SJIA during a “window of opportunity” may fundamentally affect its long-term outcome and, specifically, reduce the risk of a chronic articular course [4–6]. Hence, early diagnosis and treatment of SJIA may be essential in order to avoid long-term complications. However, guidance on establishing an early diagnosis of SJIA is limited. The existing International League of Associations for Rheumatology (ILAR) classification criteria for SJIA have been criticized [7]. Classification criteria for the closely related and presumably identical condition of adult-onset Still’s disease (AOSD) exist (Yamaguchi or Fautrel criteria) but have not been formally validated in children [8, 9]. The various classification criteria are summarized in Additional file 1: Table S1.

Historically, SJIA has been effectively treated with glucocorticoids, however, at the cost of substantial adverse effects [10, 11]. Recently advances in treatment have been made via the introduction of biologic drugs targeting interleukin (IL)-1 and IL-6 [12–14]. The outcomes of patients with SJIA have improved markedly due to the availability of these effective antirheumatic therapies [6, 12–19]. Evidence-based guidelines for the treatment of SJIA exist in Germany but are limited in their scope [20]. More recently, treatment recommendations based on evidence and expert opinion for patients with SJIA have been developed by the American College of Rheumatology (ACR) in 2011 and updated in 2013, covering a wide array of clinical scenarios [15, 17]. The North American Childhood Arthritis & Rheumatology Research Association (CARRA) has developed consensus treatment plans based on the usual clinical practice of providers within their group [16]. By experience, treatment of SJIA is highly variable among different practitioners and may often be delayed and/or inadequate; data from inception cohorts and registries indicate that inactive disease is often reached late, for example, beyond the first year of treatment [21, 22]. While there are cross-sectional data on the treatment and outcomes of SJIA in Germany, these data do not allow a precise analysis of treatment steps taken by pediatric rheumatologists in this country [21]. Furthermore, while treat-to-target and tight-control have been central principles in the care of adult patients with rheumatoid arthritis [23], these principles have not been integrated into available recommendations for the treatment of JIA. However, some authors have suggested developing treat-to-target strategies for pediatric rheumatology as well [24, 25].

The PRO-KIND (PROjekte zur Klassifikation, Überwachung und Therapie in der KINDerrheumatologie; projects for the classification, monitoring and therapy in pediatric rheumatology) initiative is a sub-committee of the German Society for Pediatric Rheumatology (GKJR) and aims to define consensus-based strategies to harmonize diagnostic and treatment approaches in Germany. This initiative was started since it was perceived that children with juvenile rheumatic diseases in Germany are currently often treated too late or not with the most up-to-date therapeutic options. Overall, the long-term goal of this project is to improve the quality of care and outcome for patients with SJIA. To meet this challenge, the goals of the PRO-KIND initiative are to foster the use of harmonized standardized diagnostic and therapeutic strategies with defined targets.

We developed practice- and consensus-based statements guiding the management of new-onset SJIA. There is a consensus in Germany that patients with probable SJIA, i.e. patients with a clinical phenotype similar to definitive SJIA but lacking chronic arthritis, may be treated similarly to patients with definitive SJIA. Furthermore, we developed treat-to-target strategies for the management of both probable and definitive SJIA. We emphasize that these strategies are based on the harmonization of current clinical practice and may not represent the optimal way to treat SJIA.

Even though SJIA is currently classified according to the ILAR classification criteria for JIA, its pathogenesis, clinical characteristics and response to therapy are very different from other categories of JIA, and SJIA is tentatively considered an autoinflammatory condition [1, 35]. A presumably identical clinical syndrome has been identified in adults, i.e. AOSD for which classification criteria have also been established, e.g. the Yamaguchi criteria or Fautrel’s criteria [8, 9, 36]. The most important difference between the pediatric and adult criteria is that chronic arthritis is required for the ILAR classification for SJIA, whereas this is not required for the classification of AOSD. In reality, many affected children show all symptoms of SJIA except for arthritis and are nevertheless diagnosed as SJIA [37]. These patients are more resembling of a systemic autoinflammatory disease than of a form of JIA. We support the use of the term “probable SJIA” for these patients; some experts would consider “Still’s disease” more appropriate [7]. It is important to note that important differential diagnoses for patients with suspected SJIA exist, including infectious, malignant or hereditary autoinflammatory diseases. Therefore, before concluding that a patient with merely suspected SJIA has probable SJIA, these differential diagnoses should be considered, and, if necessary, specifically ruled out. Furthermore, our understanding of molecular characteristics of the underlying disease mechanisms is improving [38, 39]. The serum calgranulin proteins S100A8/A9 (calprotectin) and S100A12 are rather sensitive and specific for the detection of active SJIA, and those may be incorporated into the diagnostic approach for SJIA [40–42]. S100 protein testing is currently widely applied in routine clinical practice in Germany. We believe that expert opinion is essential in cases of suspected and probable SJIA.

Our data show that in Germany around 40% of patients receive a diagnosis of SJIA without having chronic arthritis and these numbers exceed those published previously; for example, in another cohort of patients diagnosed with SJIA only around 30% fulfilled the ILAR criteria but 88% had arthritis [37]. We agree with others that patients may be classified as having definitive SJIA even without formally fulfilling the ILAR “entry” criterion of having at least 6 weeks of chronic arthritis [16]. Since the stereotypical disease course is that of a prodromal severe inflammatory phase, variably followed by arthritis, it is unrealistic that patients would manifest 6 weeks of arthritis prior to establishing a diagnosis [2]. Consequently, revised classification criteria for SJIA have been suggested and are in development [7, 43]; our data support such a new classification. Classification criteria are of practical importance in pediatric rheumatology since for the participation in clinical trials classification criteria have to be fulfilled [12, 14]. Therefore, improved classification criteria for SJIA better reflecting the entire spectrum of patients with SJIA are essential [7, 43]. However, classification criteria should not be misused as diagnostic criteria since a delay of diagnosis may lead to delayed treatment and serious complications. Unfortunately, the development of accurate diagnostic criteria is deemed impossible for most rheumatic disorders and, therefore, both the European League against Rheumatism (EULAR) and the American College for Rheumatology (ACR) do not endorse the development of diagnostic criteria [44].

The Yamaguchi criteria and the GKJR case definition may perform better than the ILAR criteria in classifying patients with SJIA. We assume that the GKJR case definition for probable SJIA, requiring laboratory evidence of marked systemic inflammation, may perform better in real life than in the registry data analyzed here. This is based on the notion that the registries did not record inflammatory markers at the time of diagnosis but rather at the time of enrolment, often after treatment had been initiated and inflammatory had already improved.

The optimal treatment for patients with probable SJIA is unclear. Still, based on clinical experience and limited data from published trials in AOSD, it is reasonable to expect that these patients benefit from treatments similar to those for definitive SJIA [7, 33, 45]. In addition, it is an intriguing new concept that the early initiation of an effective therapy may be beneficial by timely rebalancing the immune disturbance that underlies SJIA (within a “window of opportunity”). Therefore, timely treatment may prevent a switch towards the later arthritic phase of the disease [4, 6]. If the concept is correct, prevention of chronic arthritis should be targeted, in addition to treating it.

Effective and proven treatment for SJIA exist and are approved, including glucocorticoids, IL-1 and IL-6 blocking biologicals [12–14]. Some of the treatments discussed in this manuscript relate to non-approved treatment options, for example, anakinra, a recombinant IL-1 receptor antagonist. Even though anakinra is not approved for the treatment of SJIA in Germany, it is frequently used in the initial treatment of SJIA, probably also reflecting current international recommendations and consensus treatment plans [15–17]. We believe that our consensus treatment strategies integrate well with existing international recommendations. It is apparent that most pediatric rheumatologists in Germany and their ways to treat new-onset SJIA are represented in the developed treatment strategies. It is important that the developed strategies do not represent clinical trial protocols, but they rather harmonize variations in typical clinical practice. German physicians should use the consensus treatment plan felt appropriate to use in a given patient and diverge from it whenever this is in the patient’s best interest.

While the existing ACR recommendations and CARRA consensus treatment plans imply a treat-to-target idea, we explicitly embrace a treat-to-target and tight-control approach [16, 17]. Treat-to-target requires the formulation of treatment targets, close monitoring of disease activity and adjustment of treatment if targets are not reached. It has been demonstrated in the management of RA that a treat-to-target strategy improves outcome irrespective of which specific treatments were used [46, 47], essentially indicating that strategy may be more important than individual medications. The ACR recommendations indicate milestones based on the treating physician’s estimation of global disease activity and the active joint count but do not specify milestones earlier than after 2 weeks of treatment changes and beyond 1 month after treatment changes [17].

There is consensus among SJIA experts in Germany that another goal of SJIA treatment is to minimize glucocorticoid exposure and side effects, previously frequently seen in patients with SJIA [11]. The explicit consensus goal is to achieve glucocorticoid-free clinical inactive disease (CID) within 6 to 12 months after initiation of treatment with CID being defined according to the Wallace criteria [26]. Some participants argued for a shorter time frame of 3 months for this goal but there was no consensus for this opinion. The stated goals appear reasonable based on data from clinical trials and outcome data from inception cohorts [21, 22, 48].

Ideally, all patients with SJIA should be offered the opportunity to participate in disease registries allowing the collection of outcome data so that the disease courses and factors affecting the disease course may be better understood.

Furthermore, the consensus treatment strategies we developed should complement existing or future treatment guidelines. Existing treatment guidelines for SJIA are limited by the fact that they are strictly evidence-based; clinically relevant issues, for example, the specific steps when initiating or escalating therapy, are often not addressed [20]. To improve the outcome in patients with SJIA, efforts in harmonizing treatment approaches for SJIA (and other diseases) should be complemented by the collection of outcome data, so that in the long-term the outcome of different treatment strategies may be compared by means of comparative effectiveness research to further optimize treatment strategies. This approach is embodied by the plan-do-study-act improvement cycle used in quality improvement [49].

There are important limitations to our work: These strategies do not address patients with long-standing refractory disease but only patients with new-onset disease. The patient data available from registries were partially incomplete. The data is limited by the fact that temporal resolution is poor for some of the databases. For example, for the national pediatric rheumatology database, data are obtained annually, for ICON-JIA at most quarterly and for the AID-Net irregularly (usually quarterly). For that reason, not all important changes in the clinical parameters and treatment are reflected in the collected data. Therefore, the precise sequence of treatments rendered cannot be deduced from the data available. The available laboratory data in the AID-Net and ICON in most cases do not represent those that occurred during the peak inflammatory phase of the disease onset but rather at the time of enrollment which, in the case of the ICON-JIA cohort, may have been up to 12 months after initial diagnosis. The statements are consensus-based and rarely evidence-based. The outcome of patients with probable SJIA, i.e. an autoinflammatory illness without arthritis, is less well known, especially since they were not included in the landmark clinical trials of SJIA [12–14]. Some of the laboratory parameters discussed are not globally available yet, such as the S100 proteins. The different parameters used in the statistical model of meeting the various classifications are not independent. Furthermore, these consensus statements do not specifically address the management of macrophage activation syndrome (MAS), a life-threatening complication of SJIA. In the future, a separate consensus process including also hematologists and immunologists is planned to develop strategies to manage MAS.

We emphasize that our statements and deductions regarding the diagnosis of probable or definitive SJIA are evidence-informed and consensus-based and may not represent the optimal way to diagnose (and treat) SJIA. However, we believe that harmonization is important in to compare and improve different approaches. The available registries need to be improved so that detailed and relevant information regarding patient outcomes can be extracted. Ideally, once outcome data become available, the strategies can be further optimized based on ongoing circles of quality improvement [50]. Additionally, it is important that the approach considered here is compatible with international recommendations or plans on the diagnosis and treatment of SJIA [15–17].

In summary, we developed consensus-based statements and strategies on the management of SJIA by harmonization of existing practice patterns among experts in Germany which should aid clinicians in the work-up of patients with suspected JIA and the management of patients with probable and definitive SJIA.