Abstract
Objective. Calprotectin (myeloid-related protein 8/14), a heterodimeric complex of calcium-binding proteins, is expressed in granulocytes and monocytes. Calprotectin levels are high in synovial tissue, particularly in activated cells adjacent to the cartilage-pannus junction. This systematic review evaluates the use of calprotectin as an indicator of disease activity, therapeutic response, and prognosis in rheumatoid arthritis (RA).
Methods. Medline, Scopus, and the Cochrane Library (1970–2013) were searched for studies containing original data from patients with RA in whom calprotectin levels were measured in plasma/serum and/or synovial fluid (SF). We included studies examining associations between calprotectin levels and clinical and laboratory assessments, disease progression, and therapeutic response. There were no restrictions for sample size, disease duration, or length of followup.
Results. We evaluated 17 studies (1988–2013) with 1065 patients enrolled; 11 were cross-sectional and 8 had longitudinal designs with 2 studies reporting cross-sectional and longitudinal data. Systemic and SF levels of calprotectin were raised in RA. There was a wide range of levels and marked interstudy and intrastudy variability. Calprotectin levels were high in active disease and were particularly high in rheumatoid factor (RF)-positive patients. Levels fell with effective treatment. Longitudinal data showed that calprotectin was a significant and independent predictor of erosive progression and therapeutic responses, particularly in patients who received effective biological treatments.
Conclusion. SF calprotectin levels are high, suggesting there is substantial local production by inflamed synovium. Blood calprotectin levels, though highly variable, are elevated in active RA and fall with effective therapy. High baseline calprotectin levels predict future erosive damage.
The optimal laboratory assessments of disease activity in rheumatoid arthritis (RA) remain uncertain1 despite increasing understanding of the pathophysiological drivers of RA and the use of targeted biological drugs2,3. Improving disease assessment is crucial for treat-to-target strategies to achieve the best possible outcomes4,5. Conventional laboratory markers like the erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) levels are nonspecific indicators of inflammation that are sometimes high in clinically inactive RA or low in clinically active RA6. Because current RA management does not always achieve sustained RA control7, there is the potential need for additional laboratory assessments to enhance overall assessment and enable better treatment titration.
Calprotectin is an alternative laboratory biomarker that may be useful in inflammatory disorders including RA8. It is a heterodimeric complex of 2 S100 calcium-binding proteins, myeloid-related protein (MRP)-8 (S100A8) and MRP-14 (S100A9), expressed in granulocytes and monocytes9. Its release at sites of inflammation makes calprotectin a potent acute-phase reactant; it increases more than 100-fold with active inflammation8,10. Together with other members of the S100 protein family, particularly S100A12 and S100A4, calprotectin has gained widespread interest in studies of acute and chronic inflammation and associated diseases. Fecal calprotectin is a sensitive, specific marker of intestinal inflammation in inflammatory bowel disease11. Plasma calprotectin may also be a clinically useful biomarker in several other inflammatory rheumatic diseases, including ankylosing spondylitis, psoriatic arthritis, and systemic lupus erythematosus. In addition, calprotectin, as well as the S100A12 protein, has been shown to predict relapses in patients with juvenile idiopathic arthritis12,13.
Its involvement in RA inflammation has been known for many years14,15. Calprotectin is localized in synovial tissue with high levels of activated cells adjacent to the cartilage-pannus junction16. Its molecular weight of 36.5 kDa17 allows calprotectin to enter the systemic circulation where it can be measured.
We systematically reviewed published research on systemic and synovial fluid (SF) levels of calprotectin for 2 reasons. The first was to examine the value of calprotectin as a disease activity biomarker in RA. The second was to assess the role of calprotectin in monitoring RA treatment responses and predicting erosive progression.
MATERIALS AND METHODS
Search strategy
Literature searches were conducted in Medline (through PubMed), the Cochrane Library, and Scopus. Using both MeSH terms and All Fields, the following search terms (synonyms and combinations) were used: “rheumatoid arthritis” AND “calprotectin”, OR “MRP8/14”, OR “MRP8 MRP14”, OR “S100A8/A9”, OR “S100A8 S100A9”, OR “major leukocyte protein L1” (Appendix 1). We searched publications from January 1970 until December 2013. We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) framework18.
Study selection
Two reviewers performed the search and screened the initial selection of titles and abstracts for relevance and to exclude duplicates. Relevant studies were retrieved in full text and assessed in relation to predefined inclusion and exclusion criteria. References from eligible studies were manually scanned for potentially relevant articles missed in the electronic databases.
Studies included were (1) written in English, (2) contained original data from patients with RA19,20,21 whose levels of calprotectin had been measured in plasma/serum and/or SF, and (3) examined associations between calprotectin and clinical and/or laboratory assessments of disease activity or reported on the risk of disease progression or therapeutic response in relation to measured calprotectin values. Excluded studies were (1) solely referring to the monomeric proteins MRP8 (S100A8) and MRP14 (S100A9), (2) contained mixed patient samples with various inflammatory arthritides, and (3) were review articles, editorials, or case reports. No restrictions were made on the grounds of methodological standards, sample size, participant’s age, disease duration and severity, drug treatment, duration of followup, or publication year. Any disagreements between reviewers were discussed and resolved by consensus after referring to the protocol.
Data extraction
Two reviewers retrieved study design and results. The following data items were extracted from the included studies: (1) general information: title, journal, year, name of first author, and study design; (2) study population, number of patients, and diagnostic criteria; (3) baseline characteristics: age, sex, disease duration and severity, and concurrent medication; (4) baseline calprotectin levels, method of detection, and sample site; (5) details of controls and reference values; (6) clinical reference standard tests; (7) associations between calprotectin and clinical and laboratory variables; and (8) effect of drug treatment on calprotectin levels.
Method of synthesis
Because the research in this field involves a wide range of studies that measure calprotectin levels in different ways and in different circumstances, we have only undertaken a descriptive review; the studies were not suitable for any formal metaanalysis. Concentrations of calprotectin were converted to µg/l. When studies contained data from different study groups (cross-sectional vs longitudinal), the data were presented individually. To estimate the intrastudy and interstudy variability of the calprotectin concentration, the coefficient of variation (CV) was calculated for each study. When studies reported median and range, the median was used as an estimator of the mean and the SD was estimated by the range ÷ 422. When studies presented median and interquartile ranges, the CV could not be derived23; such studies were not included in the estimation of intrastudy and interstudy variations.
RESULTS
Eligible studies
The original search identified 270 studies; 17 met the inclusion criteria24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40 (Figure 1). A manual search of reference lists from eligible studies did not identify any additional articles.
Study characteristics
The 17 included studies, published from 1988 to 2013, had 1065 patients enrolled (Table 1). Eleven studies were cross-sectional and 8 had longitudinal designs, with 2 reporting cross-sectional and longitudinal datasets, giving a total of 19 investigations.
Most investigations had limited sample sizes (median 43, range 11–170). Sixteen reported age (median 57, range 19–87) and sex (70% women). Fifteen studies reported RA treatments, but none reported other medications. None reported extraarticular features or comorbidities. Thirteen investigations described their population source and 13 reported disease duration. Eleven studies reported disease activity levels; only 4 included patients with varying disease activities.
More studies measured blood levels than SF levels: 8 in blood and 5 in SF. All studies stated their assay method (ELISA in all investigations). Thirteen reported methods of preservation/storage and 12 provided detailed protocols for their assay methods. Reference values of calprotectin were included in 13 studies. Information on whether samples were analyzed prospectively or retrospectively was rarely provided. Eleven investigations included control groups; 5 were healthy controls and 6 were disease controls.
Calprotectin in plasma and serum
Sixteen studies (1988–2013) reported blood levels of calprotectin in patients with RA24,25,26,27,29,30,31,32,33,34,35,36,37,38,39,40. Three studies reported more than 1 patient population30,39,40. Calprotectin was measured in 23 different groups of patients (n = 1051). Twelve studies measured plasma calprotectin (n = 701) and 11 measured serum calprotectin (n = 350). Table 2 summarizes the results.
There were no differences between calprotectin levels in plasma or serum. Calprotectin levels were higher in patients with RA than controls. Plasma means ranged from 1923–15,516 µg/l (median 607–12,185 µg/l). Serum means ranged from 4700–38,900 µg/l (median 730–2650 µg/l). There were considerable variations of calprotectin levels in each study. The expected intraassay CV using calprotectin ELISA is typically 5%36,37,38. However, the observed CV varied by substantially greater amounts (Appendix 2). In plasma assays, CV varied from 70% to over 200%. In serum assays, CV varied from 15% to over 200%. Although differences in ELISA assays and protocols might contribute to some variation, intrastudy heterogeneity is most likely to reflect variability in patient values indicating differences in disease activities. The CV were lowest in studies using sera, suggesting that these would be of most value in clinical practice.
Calprotectin in SF
Five studies measured SF calprotectin levels in 112 patients with RA25,28,30,31,32; all had a cross-sectional design. All reported high calprotectin concentrations in SF (Table 3). There was substantial interstudy and intrastudy variability. Some studies had high CV (over 400%) while others had low CV (13% and 59%; Appendix 2). All studies showed higher SF levels compared to plasma and serum levels; 3 studies reported significant correlations between SF and blood calprotectin levels25,30,31 (Table 3). Such correlations were not seen in disease control patients with osteoarthritis and spondyloarthropathy. SF calprotectin was higher in patients with RA than patients with osteoarthritis in all 4 studies comparing these patients (Table 3); all reported highly significant differences25,28,30,32. It was also substantially higher in RA than spondyloarthropathy31.
Association with clinical and laboratory assessments
The majority of studies included correlation analyses between calprotectin and laboratory and clinical variables of RA disease activity. A comprehensive summary is provided in Appendix 3. Correlations with serological markers were generally higher than with clinical variables. Of the laboratory markers, calprotectin levels were most positively correlated with CRP and ESR (r = 0.80, p < 0.000132; r = 0.70, p < 0.00131). One study showed that the 1-year averaged level of calprotectin in 61 patients with very early arthritis correlated significantly with mean levels of serological markers [CRP r = 0.68; ESR r = 0.55; RF and anticyclic citrullinated peptide antibodies (anti-CCP) r = 0.33, p < 0.05]35. Multiple regression analyses, adjusted for sex, age, and disease duration, showed that calprotectin was associated with immunoglobulin M (IgM)-RF (p = 0.003) and anticitrullinated protein antibodies (ACPA; p = 0.045), but not IgA-RF. However, another smaller study in recent-onset RA found no association between calprotectin and IgM-RF or anti-CCP levels37.
Several clinical indices, including the Disease Activity Score at 28 joints (DAS28) and swollen joint counts (SJC), were used to assess the clinical interrelationships of calprotectin (Appendix 3). Strong correlations (r = 0.60 and 0.55, p < 0.001) were seen with DAS2830,33. One study found that calprotectin was the only serological marker to have significant correlations with SJC (r = 0.24), grip strength (r = −0.22), proximal interphalangeal joint circumferences (r = 0.33), and a combined global assessment score (r = 0.24), all with p < 0.05. CRP, ESR, and RF had no significant correlations26.
Five studies specifically related calprotectin levels to RF status (Table 4)24,26,27,33,36. RF-positive patients had higher calprotectin levels than RF-negative individuals. The strongest association was found in the largest study (p < 0.0001), a cross-sectional analysis of 145 patients that included 96 RF-positive patients based on ELISA analyses for IgM-RF33. A 10-year followup study reported that patients positive for IgM-RF, IgA-RF, or ACPA had higher calprotectin levels at baseline and at followup than patients negative for these serological markers (p < 0.001)36.
The relationship of calprotectin to radiographic damage was evaluated in a cross-sectional study of 145 patients. It reported significant associations with the van der Heijde modified Sharp score (r = 0.43, p < 0.001) and the Rheumatoid Arthritis Articular Damage (RAAD) score (r = 0.40, p < 0.001)33. After adjusting for CRP, ESR, RF, DAS28, sex, and age, multiple regression analysis showed that calprotectin was associated with the modified Sharp score (p = 0.018) and RAAD score (p = 0.04). Neither CRP nor ESR had independent associations with joint damage in corresponding analyses. When patients were divided into quartiles based on calprotectin concentrations, there were associations with the modified Sharp and RAAD scores (p < 0.001)33.
Predictive and prognostic potential and treatment effects
Eight studies evaluated calprotectin as a predictive marker of structural damage and/or as a surrogate measure of medication efficacy and treatment response29,30,31,36,37,38,39,40 (Table 5). A 10-year followup study of 124 patients found correlations between plasma calprotectin and disease progression in the modified Sharp and RAAD scores36. When patients were grouped by baseline calprotectin levels, the Sharp progression scores and RAAD scores were different between groups (p < 0.001). Baseline calprotectin levels remained associated with the Sharp progression score (p = 0.045) and RAAD score (p = 0.012) when multiple regression analyses were adjusted for baseline CRP, ESR, and anti-CCP, as well as sex, age, and disease duration36. In contrast, a prospective investigation over 5 years of 56 patients did not identify calprotectin as a predictor of radiographic damage when using the Larsen score as an outcome measure29. However, median disease duration was long in that study [7.8 years (2.3–19.4)], creating bias toward less radiographic progression. Cross-sectional correlations were reported between calprotectin and ultrasonography scores (B-mode/power Doppler) from a comprehensive investigation over a 12-month period of treatment with adalimumab in 20 patients with RA38. Calprotectin had the highest correlation coefficients compared with CRP, ESR, and serum amyloid A. Regression analyses showed calprotectin was independently associated with total ultrasonographic sum scores. In addition, calprotectin was shown to have a higher response to change from biologic treatment than CRP, ESR, and serum amyloid A38.
Circulating calprotectin levels decreased with effective treatment (Table 5). Initiation of conventional treatment in patients naive for disease-modifying antirheumatic drugs/glucocorticoid resulted in the near normalization of calprotectin levels after 3 months37. Levels were unrelated to doses of glucocorticoids and/or methotrexate. Changes in serum calprotectin positively correlated with changes in serum CRP (r = 0.48, p = 0.002), DAS28 (r = 0.39, p = 0.01), and SJC (r = 0.54, p < 0.001). Decreases in calprotectin, but not CRP, were associated with improvements in the total number of swollen joints over time37.
Studies reported reduced calprotectin levels when patients received biologics (Table 5). This change was only significant in treatment responders30,37,39,40. In 170 patients with active RA, calprotectin levels decreased significantly in responders to adalimumab and infliximab40, but showed no changes in nonresponders. Treatment with rituximab also decreased calprotectin in responders, with nonresponders showing increased levels40. When addressing the incremental predictive value, baseline calprotectin was the only statistically significant independent determinant of therapy response in full multivariate analysis with adjustments for DAS28 and 68–tender joint counts40.
DISCUSSION
Our systematic review shows that there are high calprotectin SF levels in RA, suggesting that it is produced locally by the inflamed synovium. Blood calprotectin levels, though highly variable, are elevated in active RA; they are particularly high in RF-positive patients, and fall with effective therapy. Further, high baselines calprotectin levels predict future erosive damage.
Calprotectin differs from many other laboratory biomarkers by its local production and release from the inflamed synovium. In contrast, the acute-phase reactants CRP and ESR are primarily hepatocyte-dependent after induction by interleukins released during inflammation, and can be strongly influenced by genetic factors41. As a consequence, systemic calprotectin levels may more accurately reflect the number of activated leukocytes in the inflamed joints. However, although significant correlations were reported between plasma and SF levels of calprotectin, direct evidence of synovial origin does not exist. To our knowledge, none of the published studies took into account other factors known to affect calprotectin levels, such as the presence of cardiovascular disease and obesity42,43,44. Thus, the specificity of calprotectin as a marker of active RA is not fully resolved and the possible role of comorbidities on elevated calprotectin levels needs further investigation.
Tightly controlling RA disease activity, which involves frequent disease activity measurement and treatment adjustment, improves RA clinical outcomes4,5. Identifying ideal laboratory biomarkers is challenging because limitations of current indices of disease status in RA45 and many confounding factors can influence correlations between biomarkers and disease activity assessments. Although no gold standard exists for disease activity assessment in RA, multibiomarker disease activity (MBDA) testing using a serum protein panel provides a reliable and objective assessment46,47. Calprotectin was evaluated as a potential biomarker when MBDA testing was developed; although calprotectin had several benefits as part of this system, methodological issues in the assay resulted in its exclusion from the final biomarker panel48. Improved measurement methods could change this perspective. Analytical performance of an assay is of particular importance in RA, where the presence of heterophilic antibodies (e.g., RF) may interfere with the identification and/or detection antibodies of the immunoassays49. Heterophilic Ig may further develop as a result of treatment with certain biologics attached to mouse (or humanized) monoclonal antibodies. To our knowledge, these issues have never been addressed in any research of calprotectin.
Identifying patients with potentially aggressive disease courses or those likely to respond to specific therapeutic strategies avoids overtreatment and reduces side effects and costs40,50. Calprotectin has the potential as a biomarker of treatment efficacy and response and prognosis prediction. Its potential role predicting clinical and radiographic joint damage is of interest, particularly because the MBDA test does not currently predict erosive progression44. Calprotectin is relatively stable and can be measured without the need for cold storage, making it a feasible biomarker in multicenter studies40. However, given the marked variation in levels among patients with RA, choosing a cutoff level may limit its sensitivity and/or specificity.
Our systematic review has several limitations. First, the reporting of results was incomplete, making data extraction and interpretation challenging. Second, many studies did not report key methodological features, such as study setting, the inclusion and exclusion criteria, or disease severity. The lack of information reduced transparency of the methods and results and made it difficult to exclude bias. Third, many studies had small sample sizes and did not include power calculations, suggesting they were performed without prespecified hypotheses. In heterogeneous diseases such as RA, small sample sizes restrict generalizability, may overlook important associations, and limit the number of variables that can be included in multivariate analyses. Fourth, we did not set quality thresholds for including studies in our review and some of the studies have methodological limitations. Fifth, because the studies evaluated divergent patient populations and often did not include prospective hypotheses about calprotectin levels, we did not undertake any metaanalyses and our analyses are only descriptive. Finally, although our literature search was extensive and conducted in 3 databases, some published studies may have been overlooked.
Patients with RA have raised systemic levels of calprotectin with marked interpatient variability. High calprotectin levels are found in active disease. Levels fall with effective treatment. Calprotectin predicts RA outcomes and therapeutic responses. If used in conjunction with other biomarkers, measuring calprotectin might help optimize treatment strategies. However, given that the methodological strength of the literature is low, the level of evidence is still insufficient to provide definitive recommendations for routine practice. Future well-designed studies using large populations of patients with RA, controlling or adjusting for confounding variables in appropriate statistical multivariate models, as well as further standardization of the laboratory test, are needed to fully validate calprotectin as an RA biomarker.
APPENDIX 1.
APPENDIX 2.
APPENDIX 3.
Footnotes
-
Supported by the UK National Institute for Health Research (NIHR) Programme Grants for Applied Research (Grant Reference Number: RP-PG-0610-10066. Title: Treatment Intensities and Targets in Rheumatoid Arthritis Therapy: Integrating Patients’ and Clinicians’ Views – The TITRATE Programme). Additional support from the NIHR Biomedical Research Centre at Guy’s and St. Thomas’s National Health Service (NHS) Foundation Trust, in partnership with King’s College London. The views expressed are those of the authors and not necessarily those of the NHS, the NIHR, or the UK Department of Health.
- Accepted for publication January 12, 2015.