Background
Systemic juvenile idiopathic arthritis (SJIA) is a serious disease in children with prominent systemic inflammatory features, which can be accompanied by arthritis. The International League of Associations for Rheumatology (ILAR) classification criteria still consider SJIA a category of juvenile idiopathic arthritis (JIA) [
1]. Phenotypically, it is characterized by fever, an evanescent rash, serositis, reticuloendothelial involvement, and articular manifestations. Today, SJIA is seen as part of the autoinflammatory disease spectrum [
2]. In the initial phase, sterile hyperinflammation is driven by innate immune processes with a prominent role of cytokines such as interleukin (IL)-1 and IL-6. If not stopped during this phase, the disease can progress into a later phase involving complex immune dysregulation also leading to a destructive chronic arthritis [
3,
4]. In addition, patients with SJIA are prone to complications, including life-threatening hyperinflammation (macrophage activation syndrome, SJIA-MAS) and the more recently described SJIA-associated lung disease (SJIA-LD) [
5]. Meanwhile, the treatment arsenal in SJIA has expanded markedly, and state-of-the-art therapeutic approaches include approved biologics targeting IL-1 and IL-6 pathways.
It has been well-documented that starting treatment as early as possible is favorable for the effectiveness of cytokine blockade. Early intervention could potentially prevent the biphasic course with the threatening development of chronic inflammatory processes [
6‐
9]. Nevertheless, many patients are still treated too late or not successfully because the diagnosis is not made in the optimal time window (“window of opportunity”). At onset, patients frequently present with fever of unknown origin and/or with nonspecific signs and symptoms. In this initial phase, and particularly in the absence of arthritis, distinguishing SJIA from other diagnostic alternatives remains highly challenging. Therefore, SJIA is still seen as a diagnosis of exclusion, which can significantly delay final diagnosis and initiation of adequate treatment [
10,
11]. To enable inclusion of patients without arthritis into clinical practice recommendations, treatment plans have been developed for “classic” (ILAR defined) SJIA and “probable” SJIA without arthritis [
11]. Consensus treatment plans developed by the Childhood Arthritis and Rheumatology Research Alliance (CARRA) included a case definition that requires arthritis of any duration, rather than the 6 weeks required per the ILAR criteria [
12]. New expert initiatives are currently working on revised classification criteria [
2]. An ongoing European League of Associations for Rheumatology (EULAR)/Paediatric Rheumatology European Society (PReS) task force reached consensus for a novel concept of Still’s disease as one continuum, including both SJIA and its counterpart with onset in adulthood termed adult-onset Still’s disease (AOSD).
Biomarkers differentiating SJIA in its initial inflammatory phase from the main differential diagnoses, i.e., infections or malignancies, may support early diagnosis and thus a timely start of therapy. Among published biomarkers differentiating SJIA from infections, the heterocomplex S100A8/A9 appears especially promising [
13,
14]. This complex consists of two members of the S100 family of calcium-binding proteins, which are expressed and released by cells of myeloid origin. The proteins are also referred to as myeloid-related proteins 8 and 14, and the complex has been termed MRP8/14 or “serum calprotectin” [
15]. As danger signals they are secreted by activated myeloid cells, phagocyte-specific S100 proteins amplify disease processes in SJIA by activating Toll-like receptor 4. Serum levels are extremely elevated in active SJIA, and the measurement of S100A8/A9 can be used for the diagnosis of SJIA [
16‐
18]. These results were recently validated with a commercially available ELISA [
19].
However, the analytical technique needs further optimization to ensure easy availability and enable large-scale use in routine laboratories. To achieve this objective, we collaborated with the central laboratory at our university hospital to establish an immuno-turbidimetric assay for the measurement of serum calprotectin in serum samples on an automated laboratory instrument. In addition, the inclusion of more patients with hematological malignancies was warranted, as especially leukemias can cause fever and joint pain at initial presentation in children [
20].
Patients and methods
Patients
We included patients with fever of unknown origin, signs of unprovoked inflammation, or relevant differential diagnoses of Still’s disease, who were undergoing a diagnostic checkup for suspected inflammatory or malignant diseases. By definition, there was no definite SJIA diagnosis in any of these patients at the time of sampling. In these patients, the diagnosis used in the statistical analyses was confirmed post hoc by the referring physicians as previously reported [
19]. The later confirmed diagnoses of the patients were SJIA (
n = 99), nonsystemic JIA (nsJIA,
n = 169), systemic febrile infections (
n = 51), other systemic inflammatory diseases (SIDs,
n = 126), and acute lymphoblastic leukemia (ALL,
n = 147). The patients with fever of unknown origin were collected at the University of Muenster as reported before [
19]. The NOPHO Leukemia BioBank (NLBB) in Uppsala, Sweden, provided the serum samples of consecutive ALL patients from Aalborg and Aarhus University Hospitals, Denmark [
21]. Patients with nsJIA were from the Nordic JIA cohort as previously described in detail [
22,
23]. In addition, samples from 23 healthy controls were included at the University of Muenster. They were from children who had presented to the pediatric endocrinology unit with concern for short or tall stature but for whom an inflammatory condition had been ruled out. Thus, in total, 615 samples were available for analysis.
Sample handling
Calprotectin is expressed and stored in huge amounts in leukocytes. To avoid unspecific release of the biomarker into collected samples, centrifugation and separation within 4 h after blood drawing were performed. Serum supernatants were stored at − 80 °C until processing. Calprotectin complexes are stable for at least 3 days at room and for months below − 20 °C, even with freeze/thaw cycles 3 times or more [
19].
Biomarker analyses
To evaluate the accuracy in identifying children with SJIA, the performance of an immuno-turbidimetric assay for measurements of serum calprotectin (BÜHLMANN sCAL® turbo) implemented into the automated laboratory instrument (Roche cobas® c502) at the central clinical laboratory of the University Hospital Muenster. The BÜHLMANN sCAL turbo test is a particle-enhanced immuno-turbidimetric assay (PETIA) with a detection range of 230–15,000 ng/ml (extended range up to 225,000 ng/ml by dilution of 1:15) in sample volumes of only 2–3 µl. The upper limit of serum levels in 160 healthy adults was 2740 ng/ml (90% confidence interval), according to the manufacturer’s information. In addition, an established MRP8/14 ELISA from BÜHLMANN (EK-MRP8/14) was used according to the manufacturer’s protocol to compare formerly reported values to the results obtained with the new immuno-turbidimetric assays. Samples were diluted 1:100 initially; for results below or above the assay range, a lower or higher dilution was applied.
Statistics
The statistical analysis of rank differences was performed using two-sided ANOVA and the Kruskal–Wallis test with Dunn’s correction for multiple comparisons. Receiver operating characteristic (ROC) curves were plotted to determine the accuracy of inflammation marker measurements as a diagnostic test and for calculation of cut-off values. The performance of the models was evaluated using the area under the receiver operating characteristic curve (AUC), sensitivity, specificity, positive or negative likelihood ratio, and accuracy using a threshold of p < 0.05. Spearman’s p correlations were calculated. SPSS 25.0 for Microsoft Windows (SPSS, Chicago, IL, USA) and GraphPad Prism version 8.0 for Mac (GraphPad Software, La Jolla, CA, USA) were used for statistical analyses. Unless stated otherwise, data are expressed as mean with 95% CI.
Discussion
Here, we show that an immuno-turbidimetric assay for detection of serum calprotectin on an automated laboratory instrument can be implemented in clinical laboratory settings to facilitate its use as a diagnostic routine test in clinical practice. The sCAL turbo assay on an automated instrument can provide results within 1 h. The combination of an automated process with less labor burden and time consumption also helps to limit resources. In a realistic estimation, the future cost efficiency could be comparable to other routine laboratory parameters. Of note, values generated with the formerly used MRP8/14 ELISA are higher than with the sCAL turbo assay (mean 2.5-fold difference). This has to be taken into account when comparing present values to those of previous studies. The mean values and cutoffs separating SJIA from infections or other diseases are now lower than reported for the commercial MRP8/14 ELISA [
19] but remain comparable to previous data demonstrating performance of an experimental ELISA system [
24,
25].
In the initial phase of disease when arthritis is absent, SJIA presents with nonspecific signs and symptoms (fever, rash, arthralgia, elevation of inflammatory markers), which can render definite diagnosis complicated and potentially delay targeted treatment [
11]. Therefore, an early diagnosis is often challenging since many differential diagnoses need to be considered, including infections, malignancy, and other autoinflammatory or autoimmune diseases [
19,
26,
27]. Biomarkers capable of distinguishing SJIA from its differential diagnoses can aid in this process if they are applicable in routine clinical practice, reproducible with high sensitivity and specificity, and ideally have a clear link to disease pathogenesis [
13].
Up to date, the alarmins S100A8 and S100A9 (as complex also called myeloid-related protein 8 and 14, MRP8/14 or known as serum calprotectin) provide the only validated biomarker robustly discriminating between SJIA from differential diagnoses in routine clinical settings [
19]. In contrast to other biomarker candidates, serum calprotectin analyses have been validated as a helpful tool supporting the diagnosis of SJIA in children with prolonged fever or inflammatory disease [
20]. While our university hospital previously offered this test to centers within Germany, the aim is now to broaden access and establish uniform quantification standards.
Our study has relevant limitations. Firstly, SJIA and other included diagnoses are rather rare diseases; thus, a newly designed prospective study with prespecified aim to test the performance of various biomarkers in FUO was not achievable. Therefore, we rely on a collection of samples used in previous biomarker studies and thus need to accept inhomogeneous characteristics of the patient groups. However, all samples come from carefully designed biorepositories. Age and gender do not impact the sCAL results, at least not beyond infancy. Sample handling errors are very unlikely, and the stability of the analyzed protein complex also facilitates the post hoc use as in the present study. As another limitation, while the collection of FUO patients was prospective and already used for the validation of the MRP8/14 ELISA, the samples from nsJIA and ALL patients were included retrospectively. Those were run on the new sCAL turbo assay exclusively. Thus, a comparison with historic biomarker results was not possible for these samples. Lastly, we were able to include patient samples from the most important differential diagnoses of SJIA, but it was not feasible to include all possible causes of FUO and/or unclear signs of inflammation. It was demonstrated that S100 proteins differentiate well between SJIA even from many other hyperinflammatory conditions [
28], but undoubtedly, further conditions including other malignancies than ALL, specific infections not represented by our collection, or further systemic inflammatory diseases such as multisystem inflammatory syndrome in children (MIS-C) may also mimic SJIA. This needs to be considered when interpreting our results. However, sCAL can specifically help to exclude ALL as a differential diagnosis, also due to the fact that neutrophil activity leading to its release into the serum is much lower in ALL, especially in cases with lower neutrophil counts (Supplemental Figs.
1 and
2).
In an individual patient with fever of unknown origin or unprovoked signs of inflammation, the differential diagnosis is impacted by a combination of medical history, clinical symptoms, laboratory features, and imaging findings. Depending on the level of suspicion, the demand for an exclusion of differential diagnoses with near certainty may vary before a definite diagnosis of SJIA is made. This warrants careful considerations, and the diagnostic process will always rely on an evaluation made with caution. While sCAL turbo levels above 10,500 ng/ml put SJIA on the top of the differential diagnoses list (with a specificity against infection of 82% and against ALL of 100%), other conditions cannot be ruled out completely. If the sCAL turbo levels are higher than 15,000 ng/ml, SJIA becomes even more likely (specificity ruling out infections 94%, ruling out ALL 100%), and in cases above 20,000 ng/ml, infections or leukemias can be assumed as highly unlikely. In clinical practice, such a gradual interpretation of sCAL turbo results in combination with other signs and symptoms can steer the workup and facilitate therapeutic decisions.
Conclusions
An immunoturbidimetric assay for serum-calprotectin (MRP8/14, S100A8/A9) analyses could be demonstrated as a helpful tool supporting the diagnosis of SJIA in children with prolonged fever or inflammatory disease. The implications are profound as making the correct diagnosis of SJIA can be challenging. Our findings provide evidence that S100A8/A9 are not only mechanistic biomarkers that help to distinguish SJIA from its differential diagnoses with high sensitivity and specificity but moreover can be analyzed in a routine laboratory using the sCAL turbo assay on automated standard instruments. In clinical practice, this enables clinicians to obtain reliable result within a few hours, thus facilitating the diagnostic workup. Rapid diagnosis is of high relevance for clinicians and patients, enabling start of targeted therapies to stop the immune activation underlying SJIA as early as possible.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.