Practice and consensus-based strategies in diagnosing and managing systemic juvenile idiopathic arthritis in Germany

The SJIA project group was initiated in October 2015. Altogether, 11 experienced pediatric rheumatologists participated in several online surveys and five telephone conferences. They collected and analyzed the literature, planned the process and drafted different statements. This included the following key steps: 1) Planning and consensus on aims of the project group, 2) retrieval of real-life patient data from 2 registries in Germany (see study populations below), 3) retrieval of evidence from literature on the following topics: Diagnostic/classification criteria or case definitions, diagnostics or differential diagnosis, therapeutic targets, monitoring, and medications, 4) a survey of diagnostic approaches based on clinical case scenario (in addition to members of the core project group, another 21 pediatric rheumatologists with particular experience with SJIA participated, together forming the expert panel), 5) survey analysis by the project group, and formulation of statements based on the aforementioned steps, and 6) discussion and consensus of all statements in a revised form during a face-to-face consensus conference organized in June 2016. Members of the project group were experienced in the management of patients with SJIA and had managed up to 150 patients with SJIA in the last 5 years (median: 15 patients); 6 out of 11 members of the expert panel had previously pursued research in SJIA. The expert panel was formed by additionally recruiting pediatric rheumatologists from the 13 top referring centers to the Autoinflammatory Disease registry (AID-Net) and all centers participating in the inception cohort of newly diagnosed patients with JIA (ICON-JIA).

Data was retrieved from 2 registries and 1 inception cohort regarding the current practice of diagnosing and treating SJIA in Germany. For this study, all patients who entered the registries with a diagnosis of SJIA were analyzed whether ILAR classification criteria were fulfilled or not.

The following parameters regarding the diagnosis of SJIA were retrieved for this study from ICON-JIA and the AID-Net: prevalence of arthritis, fever, rash, lymphadenopathy, hepatosplenomegaly, serositis, pharyngitis, leukocyte count, C-reactive protein (CRP) level, erythrocyte sedimentation rate (ESR) and S100 protein levels. Regarding treatment of SJIA, medications used in the first year of therapy were retrieved from the 3 databases, specifically nonsteroidal anti-inflammatory drugs (NSAIDs), systemic glucocorticoids, intraarticular glucocorticoids, methotrexate, anakinra, canakinumab, tocilizumab, adalimumab, etanercept, and cyclosporin A. Furthermore, where available, information on the disease course (monocyclic, polycyclic, chronic) and the attainment of clinically inactive disease was retrieved [26].

In addition, to better understand current concepts in clinical practice, an online survey depicting 6 different clinical scenarios was developed. It had the goal to reflect a spectrum of clinical findings that may be seen in patients with suspected, probable or definitive SJIA (Additional file 2: Table S2). These clinical scenarios contained information about pertinent history of present illness, family history, travel history, pattern of fever, pattern of rash, physical examination findings (including the presence or absence of arthritis, arthralgia, myalgia, lymphadenopathy, hepatosplenomegaly, serositis) and essential laboratory data (including complete blood count, CRP, sedimentation rate, ferritin, transaminases, LDH, albumin, fibrinogen, triglycerides, prothrombin time, partial thromboplastin time, D-dimers) as well as physician and patient global scores, Questions were then targeted to determine the various diagnostic approaches, terminology used and treatments rendered. The survey was sent out to the 33 members of the expert panel.

Descriptive statistics were used to report the registry data. The various classification criteria and case definition were applied to the AID-Net and ICON datasets. Since there were missing data, the proportion of patients fulfilling the criteria was also calculated by means of extrapolation, defining for each individual criterion that its prevalence over the entire SJIA population is reflected by the prevalence among those patients for whom the criterion was tested. For the extrapolation, it was assumed that the individual criteria were independent from each other.

Diagnostic and treatment patterns were retrieved from the registry and inception cohort data, online survey and evidence for selected clinical questions was extracted from the literature. Based on these data, statements were developed and judged via an online survey among the 11 experts from the project group; each statement could be accepted without further comment, conditionally accepted (comment provided) or rejected. Following this online survey, the statements were refined and presented at an in-person consensus conference in Münster, Germany, on June 24, 2016. Twenty experts were present during this consensus conference, and the process was guided by a professional moderator and assisted by 3 scientists who were not pediatric rheumatologists and one patient representative/parent. We used nominal group technique for consensus building. For each individual statement, the following procedures were performed: the statement and its background were presented by an individual expert member who had extracted the statement in question. Subsequently, every participant of the consensus conference had 1 min of time available to raise issues with the statement being discussed. These issues were recorded on a flip-chart. Then there was a vote via an electronic voting system (audience response system) to identify the top 3 items needing further discussion, followed by a 15-min open discussion of these items to further improve the statement. Subsequently a final round of anonymous voting took place during which each participant could either accept or reject the respective statement. Consensus was considered to be present if at least 80% of experts supportive a statement. The overall consensus process is outlined in Additional file 3: Figure S1.

We used the Oxford Centre for Evidence-based Medicine levels of evidence and grades of recommendation to further support the individual evidence for the developed statements [27]. Levels of evidence range from 1 to 5, and grades of recommendation from A to D.

Only 59.9 and 57.1% patients diagnosed with SJIA in the AID registry and the ICON-JIA cohort, respectively, had arthritis at any time during the follow-up. In both the AID registry and the ICON-JIA cohort, exact data regarding the length and pattern of fever and the duration of arthritis at time of diagnosis were not available. For the purpose of this analysis, it was assumed that, if fever was documented to be present, the fever pattern was typical, and, if any arthritis was documented, that it met the requirement for ILAR classification. Table 1 demonstrates the patient characteristics in detail. For the AID registry, detailed data regarding age at disease onset, duration of disease and presence or absence of arthritis were available.

An online survey among experts in the diagnosis and treatment of SJIA was circulated among the expert panel. Twenty-eight of 33 addressees replied (85% return rate). The initial question was whether the scenario was representative of SJIA or not, followed by the question regarding the nomenclature in a case like this. Of note, many experts requested additional diagnostic testing, since only 38 to 58% of experts designated a diagnosis of definitive SJIA based on the initially presented data (Additional file 4: Table S3). The following diagnostic tests were considered to be potentially useful (at least 50% consent), in addition to the parameters that had already been presented as part of the case scenario (data given as mean ± standard deviation across all case scenarios): echocardiography 97,8 ± 3,4%, blood cultures 96.8 ± 2.4%, ultrasound of the abdomen 90.4 ± 9.4%, serum S100 proteins 91.4 ± 2.9%, chest X-ray 85.4 ± 3.6%, serum immunoglobulins 79.9 ± 3.7%, ECG 75.1 ± 8.2%, urine studies 75.7 ± 3.3%, autoantibodies 70.5 ± 5.7%, joint ultrasound 69.9 ± 3.6%, peripheral blood smear 68.6 ± 4.4%, serum complement levels 61.9 ± 6.0%, bone marrow aspiration 55.4 ± 9.4%, serum procalcitonin 50.3 ± 6.2%. Other tests were considered to be essential by fewer than 50% of participant (listed in order of descending frequency): rheumatoid factor, soluble IL-2 receptor levels, serum amyloid A, plasma cytokines, lymphocyte subpopulations, lumbar puncture, NK cell degranulation, pulmonary function testing, NK cell function, whole body MRI, MRI of painful joints or PET-CT.

Data on the treatment and disease course of SJIA was retrieved from the national pediatric rheumatology database, the ICON-JIA cohort and the AID-Net registry (Table 2).

In the aforementioned online survey, in addition to diagnostic procedures, we also queried about treatment decisions. These data are represented in Additional file 5: Table S4. It is apparent that systemic glucocorticoids are a preferred treatment modality for SJIA among experts in Germany since between 58 and 91% experts would suggest using glucocorticoids in the various scenarios. Non-steroidal anti-inflammatory drugs were suggested by 48 to 71% of experts, depending on the scenario, but only as adjunctive therapy. However, biologics (anakinra, canakinumab or tocilizumab) are also preferred frequently as an initial therapy, depending on the scenario by 29 to 50% of experts.

Individual statements regarding the diagnosis and management of SJIA were discussed and consented during the final consensus conference. The agreed upon statements, the consensus level and the level of evidence are shown in Table 3. For statement 1 (case definitions for PRO-KIND strategies), the current SJIA and AOSD classification criteria, and Childhood Arthritis & Rheumatology Research Association (CARRA) case definition were identified and analyzed (see Additional file 1: Table S1). For statement 2 (diagnostics and differential diagnosis), a literature search was initiated regarding various blood biomarkers routinely available in Germany (see Additional file 6: Table S5). Several reports were identified discussing the occurrence of malignancy mimicking SJIA, therefore supporting statement 3A [28‐31]. Multiple infections may mimic some or most of the clinical manifestations of SJIA. Since the probability for various infections strongly depends on clinical circumstances, exposures and risk factors, the working group concluded that testing for specific infections was not warranted under all circumstances and that, rather, a reasonable clinical approach regarding testing for infections should be used as outlined in the German guideline in respect to fever of unknown origin [32]. There was insufficient data to represent the utility of various imaging studies in the diagnosis of SJIA and statement 3C is based on expert opinion. Since most patients with SJIA demonstrate an excellent response to glucocorticoid, anti-IL-1 or anti-IL-6 therapy, the lack of such a response should prompt reconsideration of the diagnosis of SJIA (statement 3D) [12‐14, 33]. Statement 4 defines the treatment targets. Essentially, the treatment targets should be reached in succession, indicating that the priority in SJIA therapy is control of systemic inflammation followed by overall well-being and control of arthritis. If a treatment goal is not reached, changes in treatment are required. For this purpose, the initial treatment target focuses on the absence of fever and marked reduction in CRP (by at least 50%) and the interim target includes the physician global assessment and the count of joints with active arthritis and/or the juvenile arthritis disease activity score (JADAS)-10 which has been validated for different categories of JIA, including SJIA [34]. Statement 5 addresses the available treatment options for patients with SJIA, indicating strong evidence for the efficacy of using glucocorticoids, IL-1 and IL-6 blockade. Statement 6 addresses the treatment of patients with probable SJIA, including a preference for treatment with high-dose glucocorticoids and/or IL-1 blockade. Statement 7 tackles the treatment of patients with definitive SJIA, also addressing specific issues, such as the use of methotrexate and alternative biologics, including tumor necrosis factor (TNF) blockers and abatacept. Table 5 summarizes our case definitions for patients with probable and definitive SJIA. Of note, our case definition for definitive SJIA is similar to the CARRA case definition, i.e. the presence of any arthritis for any length of time would satisfy the entry criteria. The ILAR criteria, formally requiring 6 weeks of arthritis are felt to be unrealistic in the real world since most patients require treatment much earlier.

The different classification criteria, i.e. the ILAR and Yamaguchi criteria and the GKJR case definitions were applied to the individual patient data in the AID registry and the ICON-JIA cohort (Additional file 7: Table S6). The probability of patients within each registry to fulfill the various criteria and case definition was calculated, indicating that the sensitivity of the ILAR classification was rather poor in the AID registry and the ICON-JIA cohort, identifying only 47.8 and 54.3% when applying the criteria de facto, respectively (Additional file 7: Table S6). In contrast, the Yamaguchi classification criteria identified 55.1 and 77.1% of patients correctly, respectively, and the GKJR case definition identified 62.3 and 65.7%, respectively. The sensitivity improved when applying the criteria to patients with laboratory data obtained within the first month after diagnosis in the ICON-JIA cohort and when accounting for missing data by extrapolation (Table 4).

Treatment strategies for patients with probable/suspected or definitive SJIA were derived from the agreed upon statements. Figure 1 visualizes the treatment targets identified. Figure 2 visualizes the treat-to-target concept and the various ramifications during the first year of treatment of new-onset probable SJIA. The central treatment options for probable SJIA include high-dose systemic glucocorticoids and/or interleukin-1 blockade (anakinra). Figure 3 indicates the options for patients with definitive SJIA. The initial treatment options include high-dose glucocorticoids, anakinra, canakinumab or tocilizumab. In addition, systemic glucocorticoids, intraarticular glucocorticoids, NSAIDs, and/or MTX may be used complementary during the initial treatment. During the course of the disease, for patients who have persistent polyarthritis without systemic inflammation, treatment with tumor necrosis factor-alpha (TNF) blockade or abatacept is an option. Methotrexate is commonly used as an adjunctive therapy in case of polyarthritis and intraarticular glucocorticoid injections may be used in case of persistent arthritis. These treatment strategies represent a harmonization of current SJIA treatment among experts in Germany which is independent of the current approval status of the medications addressed. However, the strategies do not address the management of patients beyond the first year of therapy. Especially in case of refractory SJIA, advice by experts in the management of SJIA should be sought.