Introduction
Anti-citrullinated protein/peptide antibodies (ACPAs) are considered the most specific serologic markers of rheumatoid arthritis (RA) [
1]. Although the sensitivity is similar to that of rheumatoid factor (RF)--the only antibody included in the American College of Rheumatology (ACR) classification criteria--ACPAs have a higher specificity [
2]. ACPAs recognize proteins or peptides with arginine residues converted to citrulline by a posttranslational modification and have diagnostic and prognostic significance [
3]. Recent studies have demonstrated that ACPAs are the strongest serum marker associated with future RA progression in patients with recent-onset arthritis [
4,
5] and radiographic progression in ACPA-positive patients with early RA is greater than that in ACPA-negative patients [
6‐
8]. ACPAs also have been linked to certain genetic and epidemiologic characteristics such as the HLADRB04 genotype [
9] and smoking [
10].
ACPAs can be detected by using enzyme-linked immunosorbent assays (ELISAs) with different citrullinated protein or peptide substrates. The most widely used in clinical practice is the cyclic citrullinated peptide 2 assay (CCP2) [
1‐
3]. The cyclic peptide(s) in the CCP2 have no homology with known proteins [
1] and improved the sensitivity of the first test to become available based on a synthetic citrullinated peptide derived from filaggrin (CCP1) [
11]. More recently, other ELISA tests using citrullinated mutated vimentin [
12], citrullinated human fibrinogen [
13], or third-generation citrullinated peptides (CCP3) [
14] with useful diagnostic and prognostic properties have been developed. Although most ACPA-based tests seem to have similar diagnostic yields, discrepancies may arise, probably because of the different antigen sources used [
15].
Although original assays used citrullinated peptides derived from human filaggrin, this epithelial protein is not expressed in synovial tissues and so is probably not the
in vivo target of these autoantibodies. Several proteins present in inflamed rheumatoid synovium, such as type I collagen, vimentin, α-enolase, and fibrin(ogen), can be citrullinated [
16]. Citrullinated fibrin has been identified as a potential synovial target for ACPA [
17], and two citrullinated fibrin-derived peptides in the α and β chains of human fibrin have been found to be the main antigen substrates for these antibodies [
18]. We previously demonstrated the special ability and specificity of a citrullinated peptide sequence of α-fibrin ([Cit
621,627,630]α-fibrin(617-631)) to recognize autoantibodies in RA sera [
19]. Three cyclic citrullinated peptides derived from these regions of α-fibrin were found to recognize serum autoantibodies in RA in a subsequent study [
20]. Further interesting results were obtained with a chimeric fibrin/filaggrin citrullinated peptide (CFFCP1) containing an α-fibrin peptide and the cyclic filaggrin peptide, cfc-1cyc, which forms the basis of the commercial CCP1 test. The noncommercial ELISA test using this chimeric citrullinated synthetic peptide was more sensitive than the commercial one using CCP1 (82% vs. 65.8%) in an RA population and reacted with some CCP2-negative sera [
20].
The aim of the present study was to evaluate the diagnostic yield of three ELISA tests based on this CFFCP1 peptide and two new synthetic chimeric fibrin/filaggrin peptides (CFFCP2, CFFCP3) and to compare their sensitivity and specificity in RA and other disease groups with the commercial CCP2 test. The prognostic value of these chimeric ACPAs also was studied in a cohort of patients with early RA.
Discussion
This study analyzed the diagnostic and prognostic properties of three new ELISA tests for RA to detect ACPA. These noncommercial assays, which use synthetic chimeric cyclic citrullinated peptides from filaggrin and the α chain of fibrin as antigenic substrates, showed a high sensitivity and specificity for RA in comparison with healthy controls and those with other chronic diseases. Furthermore, in a group of early RA patients, these antibodies were associated with poor radiographic outcome.
Synovitis in RA is characterized by excessive generation and breakdown of fibrinogen [
26]. Citrullinated fibrinogen may participate in the inflammatory process in RA, affecting the balance between coagulation and fibrinolysis [
27] and acting as an important autoantigen [
28]. Recent data on the arthritogenic properties of citrullinated fibrinogen in a mouse model of arthritis [
29] and the presence of immune complexes containing citrullinated fibrinogen in the sera and synovium of RA point to a direct role in the pathogenesis of rheumatoid synovitis [
30]. An
in vitro human model demonstrated the inflammatory potential of ACPA-containing immune complexes of citrullinated fibrin, inducing TNF-α secretion by macrophages [
31]. Together, these findings reinforce the role of citrullinated fibrinogen in RA pathogenesis and suggest that this protein could be an important
in vivo target of the ACPAs present in RA sera and therefore an optimal antigenic substrate for the determination of ACPAs.
We analyzed the sensitivity and specificity of CFFCP1 in a large group of RA patients and other well-defined control groups. We also included two new synthetic chimeric peptides also derived from filaggrin and the α chain of fibrin (CFFCP2 and CFFCP3). The ELISA assays for chimeric citrullinated peptides showed a high sensitivity (71.4% to 78.0%) for RA diagnosis, similar to that seen for the CCP2-based test (73.9%), by using cutoff values yielding 98% specificity with respect to healthy blood donors. The sensitivity of CFFCP is similar in patients with early RA and in those with established disease, and lower in RF-negative patients, as previously reported [
32].
As expected, the specificity of anti-CFFCP antibodies was not as high when compared with other chronic diseases, especially SLE, in which up to 9.2% of sera yielded positive results, as observed in our population and in other studies [
33,
34]. However, antibody levels were low in most SLE patients with positive results compared with RA patients. The different anti-CFFCP antibodies were also detected in a small number of patients with PsA or HCV infection.
As previously described by our group [
19,
20], not all citrullinated fibrin-derived peptides are equally immunoreactive with ACPA, reflecting the importance of the aminoacyl environment of citrulline residues and the correct balance of Arg/Cit residues within the synthetic molecule. This probably influences the adoption of a preferred conformation with enhanced capacity for binding to autoantibodies present in RA patients. Of all the α-fibrin-related peptides we analyzed, those spanning the region from 617 to 631, with a citrullyl residue at the 630 position, were clearly the most reactive [
20]. Thus, in agreement with the results reported by Sebagg
et al. [
18], the Cit
630-bearing peptides from α-fibrin could represent a good molecular mimic of the ACPA epitopes targeted
in vivo. Particularly good results were obtained by combining two peptide units of different citrullinated proteins. In particular, CFFCP1, CFFCP2, and CFFCP3, the three 15-mer Cit
630-bearing peptides from α-fibrin covalently coupled to the cyclic filaggrin peptide, which constitutes the CCP1 test [
35], have been shown to be the most reactive ACPA epitopes [
20].
The CCP2-based test was very specific and highly sensitive for RA, with a higher sensitivity than that of CCP1 [
36]. Our results showed a similar sensitivity, specificity, and predictive values for the diagnosis of RA compared with the commercial CCP2 test. A similar diagnostic yield was described for CCP2 and anti-citrullinated fibrinogen in a study with an ELISA test containing human fibrinogen as the antigenic epitope [
37]. A very strong correlation was found between the three anti-CFFCP antibodies and between anti-CFFCPs and anti-CCP2 antibodies, although a significant proportion of sera showed discrepancies, mainly in patients with antibodies near the cut-off values. Our studies also found discordant results when different commercial ELISA tests were compared, with the main explanation being the different types of antigenic peptide/proteins used [
15,
38]. The sensitivity of the presence of any of the three anti-CFFCP antibodies increased to 80.5%, 6.6% higher than the commercial CCP2-based test, with no loss of specificity.
Studies have shown that ACPAs are associated with a poor disease outcome. Most of these studies used the commercial CCP1- or CCP2-based tests [
6‐
8], but some also used antibodies against citrullinated human fibrin(ogen) [
13,
39], and mutated vimentin [
12,
40] has been associated with radiographic damage. In our cohort with early RA, modified Larsen scores at baseline and 2 years after starting DMARD therapy were higher in patients with positive baseline values of anti-CFFCPs, with significant differences at both time points for anti-CFFCP1 antibodies. A clear but nonsignificant trend also was observed for anti-CFFCP2, anti-CFFCP3, and anti-CCP2 status. Patients with positive anti-CFFCP and negative anti-CCP2 status had a radiographic progression similar to that observed in patients positive for both autoantibodies, indicating that these autoantibodies may be a markers of poor radiographic outcome in a subgroup of patients who are anti-CCP2 negative; however, the subgroups were too small to draw definitive conclusions.
Despite radiographic progression, patients with and without baseline anti-CFFCP antibodies and CCP2 followed a similar clinical course, according to DAS28. Short follow-up could explain why anti-CFFCP status at baseline does not correlate with disease activity, because some studies found a more-severe disease course in patients with ACPA, primarily after 2 years of follow-up [
41,
42].
As observed with anti-CCP2 antibodies [
43] and antibodies against mutated vimentin in early RA [
40], reduced anti-CFFCP antibody titers were observed in our cohort during the follow-up. We found no correlation between changes in CFFCP antibody titers and changes in most parameters of disease activity during follow-up. Our results are very similar to those observed with anti-CCP2 antibodies in this and other studies [
42‐
44], but differ from those observed with anti-citrullinated vimentin antibodies, in which an association between clinical improvement and change in antibody titers was reported in early RA [
40]. We do not know whether this reflects different antibody properties or differences in treatment strategies with DMARDs.
Competing interests
The CFFCP test is currently in process for a patent. We have not received any reimbursements, fees, funding, or salary from any organization.
Authors' contributions
RS had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. RS, JDC, and IH designed the study. RS, EG, JAG-P, MLP, and MJG acquired the data. RS, EG, IH, OV, GE, JG, ML, and AB analyzed and interpreted the data. RS, IH, ML, JAG-P, JDC, and AB prepared the manuscript. RS and JG performed statistical analyses.