Background
Serum biomarkers are frequently used as diagnostic tools as well as for evaluation of disease activity and treatment response in inflammatory rheumatic diseases (IRD). C-reactive protein (CRP) and erythrocyte sedimentation rate (ESR) are routinely measured for disease monitoring. However, less than 50% of patients with axial spondyloarthritis (axSpA) and psoriatic arthritis (PsA) present with an elevated CRP [
1,
2]. Furthermore, anti-interleukin (IL)-6 therapies (such as tocilizumab) have a direct blocking effect on CRP [
3,
4], which precludes its use to assess treatment response or the occurrence of infection. Thus, new biomarkers are required, particularly in rheumatoid arthritis (RA), axSpA, and psoriatic arthritis (PsA).
Calprotectin (or leukocyte protein L1) is a potentially interesting biomarker for a number of IRD. Calprotectin is a heterodimeric complex of two non-covalently associated calcium-binding proteins, S100A8 and S100A9. They belong to the S100 proteins family which regroups 25 members [
5]. These two proteins are also known as myeloid-related protein (MRP) 8 and 14 or calgranulin A and B. Although debated, it seems that they are functionally active when present in their heterodimeric form [
6], forming the calprotectin complex (S100A8/S100A9). Calprotectin is stored in large amounts in the granulocyte cytosol (40–60% of cytosolic protein content) and has both intracellular and extracellular functions. Inside the cells, it regulates calcium homeostasis, interacts with the cytoskeleton and microtubules and plays a role in intracellular trafficking of phagocytes. Its role for leukocyte transmigration has been recently shown in a mouse model [
7]. When released, calprotectin functions as a damage-associated pattern molecules (DAMP) or alarmin, promoting the inflammatory response.
The normal serum levels of calprotectin are estimated to range between 0.1 and 1.6 μg/ml and can be elevated in numerous conditions such as infection, inflammation, or cancer [
8]. More recently, a cut-off above 0.9 μg/ml was proposed to distinguish RA from non-inflammatory arthritis [
9]. Given its low molecular weight (36.5 kDa), calprotectin may diffuse from inflamed tissues to the blood circulation. Indeed, serum concentrations seem to reflect synovial concentrations, with a ratio of 1:2–3 according to some studies [
10,
11]. Both synovial and serum calprotectin levels are elevated in RA, but not in osteoarthritis [
10]. A recent meta-analysis demonstrated that circulating and synovial calprotectin correlates with disease activity in RA [
12]. Calprotectin levels also correlated with ultrasound synovitis, particularly with power Doppler signal which was not the case for either CRP or ESR [
13‐
15]. Furthermore, calprotectin is potentially an independent predictor of radiological progression [
14,
15]. Inconsistent results have been reported regarding the use of calprotectin as a predictor of treatment response [
16‐
19]. Very little is known about calprotectin and IL-6 inhibition therapies. A single cross-sectional study investigated calprotectin and IL-6 inhibitors. In this cohort of 33 RA patients receiving tocilizumab, serum calprotectin levels seemed to be an accurate biomarker for assessing disease activity [
20]. Serum calprotectin levels are also significantly elevated in axSpA and non-radiographic (nr)-axSpA and correlate with CRP as well as with clinical (BASDAI, ASDAS) and radiological (SPARCC) disease activity scores [
21,
22]. Moreover, some authors reported that calprotectin was predictive for progression of structural damage in the spine of patients with axSpA [
23,
24]. Among patients with PsA, serum calprotectin levels also correlated with disease activity, with higher levels in patients with symmetrical polyarthritis compared to patients with a mono- or oligo-articular presentations [
25]. A recent study could also confirm this correlation in early PsA, as well as in early RA [
26]. In psoriasis, serum calprotectin levels correlate with the Psoriasis Area and Severity Index (PASI) score [
27].
The overall aim of the study was to compare the value of serum calprotectin as a biomarker for disease activity and severity in RA, axSpA, and PsA, in a cohort of patients from the Swiss Clinical Quality Management (SCQM) registry.
Methods
Design and study population
This study is a nested case-control study within the Swiss Clinical Quality Management (SCQM) registry. This national cohort was established in 1996 and includes patients with a confirmed diagnosis of RA, axSpA, and PsA. The diagnosis is established according to the expertise of board-certified rheumatologists. Clinicians participating in the SCQM are office or hospital-based rheumatologists. They provide clinical patient data and ultrasound examination data (for RA patients), on a regular basis. In addition, patients fill out a number of patient-reported outcome (PRO) questionnaires at each visit. This registry has been described in more detail in a previous article [
28].
The present study includes all participants in the SCQM registry with a blood sample available between March 2011 and April 2013. Asymptomatic first-degree relatives (FDRs) of RA patients, from the SCREEM-RA cohort [
29], were used as healthy controls (HC). Participants from this Swiss multicenter cohort study of RA FDRs were matched to the RA population in terms of age and sex. They had no signs of autoimmunity, defined as the absence of anti-citrullinated peptide antibodies (ACPAs), rheumatoid factor (RF), and without shared epitope and no joint complaints. This study was approved by the Ethics Committee of the University Hospital of Geneva, and all individuals signed an informed consent form prior to enrolment, in accordance with the Declaration of Helsinki.
Exposure of interest and outcome parameters
Serum calprotectin levels were measured using the QUANTA Lite Calprotectin ELISA (Inova Diagnostics, San Diego, Research Use Only for serum/plasma). All values are expressed in micrograms per milliliter. Levels of CRP (mg/l) and ESR (mm/h) were also documented in each disease group. Differences in calprotectin levels in each disease group were compared with the Wilcoxon test. As the distribution of serum calprotectin in our cohort is non-normal (Shapiro-Wilk normality test w = 0.815, p value < 0.001) and as the relationships between calprotectin and the outcome parameters are non-linear, we chose to categorize calprotectin levels into quartiles for each disease group. Comparison of clinical outcomes by calprotectin quartile levels was then performed using the Kruskal-Wallis tests for continuous outcomes or trend tests for categorical outcomes.
We examined the cut-off for the serum calprotectin level as marker for disease activity with a receiver operating characteristic (ROC) analysis.
For RA patients, outcome measures included clinically assessed scores such as the swollen joint count (SJC), tender joint count (TJC), and self-reported scores such as the Rheumatoid Arthritis Disease Activity Index (RADAI) and the health assessment questionnaire (HAQ) disability index. We also used composite scores such as the Clinical Disease Activity Index (CDAI) and Disease Activity Score (DAS28). Hand and feet radiographs were assessed regularly over time (until 2016) with a validated scoring method, the Ratingen score [
30]. Radiographs are evaluated prospectively by an assessor blinded to the clinical information. Multivariable analyses were corrected for age, sex, smoking status, disease duration, disease activity (DAS28), number of prior biologics, and calendar year of biosampling.
For musculoskeletal ultrasound assessments in RA patients, we used a standardized semiquantitative (0 to 3) scoring system for grayscale (GS) mode and Power Doppler (PD). This score was developed by the Swiss Sonography in Arthritis and Rheumatism (SONAR) group, based on the recommendations from the OMERACT group [
31], and has demonstrated good correlation with clinical disease activity and sensitivity to change in an observational cohort study [
32]. The SONAR score includes 22 joints (the same joints as the DAS28, but excluding the 2 joints in the thumbs and the shoulders). A total GS-mode score of at least 10 (out of 66) or a total PD-score of at least 1 (out of 66) was defined as a positive SONAR score in this study, suggesting active inflammatory disease. Multivariable regression models were used to compare the association of CRP and calprotectin with USPD, and the proportions of explained variance were estimated using
R2.
In a subgroup analysis, we performed the ROC and Kruskal-Wallis analyses among patients treated with tocilizumab during at least 1 month (n = 92) to compare the capacity of calprotectin and CRP to detect a DAS28-ESR score equal or superior to 3.2 (corresponding to moderate or severe disease activity). We restricted the analysis to the subjects who had provided a blood sample within 30 days of clinical examination.
For axSpA patients, our main outcomes were clinical disease activity scores focusing on axial involvement and disability, such as the Bath Ankylosing Spondylitis Disease Activity Index (BASDAI), the Bath Ankylosing Spondylitis Functional Index (BASFI), and the Ankylosing Spondylitis Disease Activity Score (ASDAS) [
33]. We also examined the physician global disease activity scale (graded from 0 to 10). Clinical signs of peripheral disease were also investigated, such as the SJC, coxitis, enthesitis, and dactylitis.
In PsA, the TJC, the SJC, and the Disease Activity Index for PSoriatic Arthritis (DAPSA) were the main outcomes for articular involvement. For skin evaluation, we dichotomized the reported extent of skin involvement into absent (no lesion and almost no lesion) or present (mild, moderate or severe involvement). The Dermatology Life Quality Index (DLQI) self-reported score was also included.
All analyses were performed using R v3.5.1 (R foundation, Vienna).
Discussion
This is the largest study of serum calprotectin in patients with RA, axSpA, and PsA including cross-sectional as well as prospective analyses, with supporting evidence of the role of calprotectin as a potential prognostic biomarker for joint-related rheumatic diseases. In our study, serum calprotectin levels in the HC group were similar to levels found in the literature, with a median level of 1.2 μg/ml. Consistent with previous reports, we found higher levels of serum calprotectin in RA, axSpA, and PsA patients compared to the healthy control group. In RA, serum calprotectin levels were associated with disease activity (SJC and CDAI) and disease severity (HAQ and Ratingen score). A significant association between calprotectin quartiles levels and articular inflammation on ultrasound PD was also observed. Despite the low sensitivity of calprotectin, it has a good specificity to differentiate RA patients from healthy controls (even in this established disease cohort where the majority of patients were treated and well controlled). In association to other biomarkers (mainly ACPA and RF), serum calprotectin may be helpful to detect inflammatory disease in patients complaining of arthralgia. However, further studies are required to confirm this hypothesis [
9].
In axSpA, an association between serum calprotectin and physician global disease activity and ASDAS scores was found. Serum calprotectin does not seem to be a good marker for axial disease activity (BASDAI, BASFI). Nevertheless, it correlated more strongly with peripheral articular involvement, and a significant association between serum calprotectin quartiles levels and hip involvement was observed. To our knowledge, this is the first study showing an association of serum calprotectin level with large joint involvement in axSpA.
Surprisingly, we did not find any associations between disease activity and calprotectin levels in PsA patients. Other studies demonstrated a correlation between serum calprotectin and disease activity in a polyarticular PsA population [
34] or in early PsA [
26]. One explanation could be the low SJC in axSpA and PsA compared to RA patients (Table
1). Another explanation could be the distribution of joint involvement. As already mentioned, calprotectin levels were found to be higher in patients with polyarticular PsA [
25]. In a previous work focusing on PsA subtypes in the SCQM cohort, Stekhoven et al. showed that only a minority of patients had a polyarticular pattern (16 out of 957) in this Swiss cohort [
35]. Finally, an interesting study observed a different distribution of calprotectin expression in the synovial tissue of patients with PsA compared to RA and SpA [
11]. This might also suggest a different role of calprotectin in the physiopathology in PsA.
The serum calprotectin levels are influenced by local inflammation and synovial fluid calprotectin levels [
10,
36]. Thus, we hypothesize that serum calprotectin levels reflect the amount of synovial inflammation, which is higher in polyarticular disease or in large joint involvement such as the hip. This observation is further supported by a correlation between calprotectin and the USPD score, with calprotectin outperforming CRP. The swollen joint count was also significantly different across calprotectin quartiles, even in the subgroup of patients with normal CRP levels. Recently, a prospective study of RA patients with normal CRP levels confirmed an association between calprotectin levels and disease activity in multivariate analyses [
37]. Similarly, in a cross-sectional study of 87 patients receiving a TNF inhibitor, calprotectin seemed to be more accurate than the acute-phase reactants to discriminate disease activity [
38].
Our results also suggest that calprotectin could be a useful biomarker for monitoring disease activity in RA patients and may add additional information to that provided by CRP. This observation is further underlined by the superiority of calprotectin over CRP in patients treated with tocilizumab. Indeed, as opposed to CRP, calprotectin is not produced by hepatocytes in response to inflammatory cytokines. It is mainly released via passive mechanisms when tissues are damaged (through necrotic cells and formation of neutrophil extra-cellular trap). Calprotectin is also a good marker of neutrophil activation and can therefore be more useful than CRP in some diseases, like RA, where neutrophil activation is present [
39]. Therefore, calprotectin is not considered as a bona fide acute phase protein and could represent a good alternative when determination of acute-phase proteins cannot be reliably used such as in patients on anti-IL-6 therapies.
Interestingly, higher calprotectin levels were also associated with a higher prevalence of rheumatoid nodules, RF seropositivity, radiographic damage, and HAQ. These observations might suggest that serum calprotectin could also be a biomarker of disease severity and prognosis in RA, but this has to be further evaluated in a longitudinal study. Other authors have demonstrated that calprotectin is associated with radiographic progression [
14] and cardiovascular risk [
40]. Biologically, it has been shown that the principal extracellular effect of calprotectin is to amplify the inflammatory process through several mechanisms which may contribute to joint damage. Acting as a DAMP, it facilitates the activation of the TLR-4 signaling [
41,
42], leading to secretion of proinflammatory cytokines such as IL-1, IL-6, and TNF, via the activation of NF-kB and p38 MPAK pathways [
43].
This study has several limitations. Firstly, no longitudinal samples were collected, and consequently, we were unable to study the evolution of the calprotectin levels over time. Secondly, most of the patients had long-standing disease, and practically, all patients were on treatment with 40.8 to 57.5% of participants on a biologic, agent depending on the diagnosis. This may reduce the probability of demonstrating a positive association with the serum calprotectin level. Nonetheless, we demonstrated several significant associations between calprotectin and disease activity in both RA and axSpA. Despite the availability of detailed clinical characteristics, some characteristics such as the PASI skin score were not available. Another potential limitation is that the blood samples were not always performed on the same day as the clinical evaluation, although the time difference was minimal and unlikely to have influenced the results.
The strengths of this study include the large cohort size that is representative of the general population with inflammatory joint disease, including patients from outside tertiary hospitals, as well as the availability of detailed clinical, biological, and imaging data.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.