Isotype distribution of ANTI–CYCLIC citrullinated peptide antibodies in undifferentiated arthritis and rheumatoid arthritis reflects an ongoing immune response

Authors


Abstract

Objective

The evolution of the rheumatoid arthritis (RA)–specific anti–cyclic citrullinated peptide (anti-CCP) antibody response, as measured by the isotypes of anti-CCP, has not been described. This study was undertaken to determine anti-CCP isotype usage in patients with undifferentiated arthritis (UA), patients with recent-onset RA, and patients with RA of long duration.

Methods

IgA, IgM, and IgG subclasses of anti-CCP were measured by enzyme-linked immunosorbent assay in serum samples that were obtained from IgG anti-CCP antibody–positive patients with UA (n = 110) and IgG anti-CCP antibody–positive patients with RA (n = 152) early after the onset of arthritis. Patients with UA in whom RA developed within 1 year (UA→RA) were compared with patients with UA in whom RA did not develop within 1 year (UA→UA). In addition, baseline serum samples obtained from a subset of patients with RA (n = 64) were compared with sera obtained from the same patients a median of 7 years later.

Results

IgM anti-CCP was present in early samples from both patients with UA and patients with RA and in followup samples from patients with RA. Several IgG anti-CCP antibody–positive patients who did not have IgM anti-CCP early after disease onset did display IgM anti-CCP later in the course of the arthritis. A diverse pattern of isotype usage was detected in early samples, with a trend toward lower frequencies of all isotypes of anti-CCP in patients with UA compared with patients with RA and in UA→UA patients compared with UA→RA patients. Levels of all isotypes except IgG1 had decreased after 7 years.

Conclusion

These data indicate development of the anti-CCP isotype repertoire into full usage early in the course of arthritis. The sustained presence of IgM anti-CCP indicates ongoing recruitment of new B cells into the anti-CCP response, reflecting a continuous (re)activation of the RA-specific anti-CCP response during the course of anti-CCP–positive arthritis.

Antibodies against cyclic citrullinated peptide (CCP) are highly specific for rheumatoid arthritis (RA) (1), are predictive of the development of RA in patients with undifferentiated arthritis (UA) (2), and are associated with the extent of joint destruction (3). Furthermore, anti-CCP antibodies have been shown to enhance disease severity in mice with experimental arthritis (4). Taken together, these findings point to a pivotal role of anti-CCP antibodies in the progression of RA. At present, little information is available regarding isotype usage of the anti-CCP antibody response, because total levels of IgG anti-CCP are commonly measured. Given the possible contribution of these antibodies to the progression of RA, more detailed analyses of the anti-CCP response are valuable, because the results of such analyses could provide insight into the nature of the antibody response.

Naive B lymphocytes express 2 classes of membrane-bound antibodies, IgM and IgD, which function as the receptors for antigens. Activation of mature naive B cells requires signals delivered through their antigen receptors and, in the case of T cell–dependent antigens, additional signals that are provided after interaction with an antigen-specific helper T cell.

Activation of naive B cells upon the first antigen encounter results in proliferation and differentiation into IgM antibody–secreting cells. During their differentiation, upon further contact with T cells, some B cells start to produce antibodies of other heavy-chain classes (isotype class switching), and the affinity of the produced antibodies matures. Eventually, this will lead to expanded populations of class-switched high-affinity antibody-secreting plasma cells and to the generation of memory B cells that will differentiate into plasma cells after antigen reencounter.

Upon repeated antigen exposure, the IgM response is usually absent or relatively low as compared with the primary antibody response (5). Since immunoglobulin isotype switching is the result of a recombination process in the genetic region that encodes for the different heavy-chain classes of the corresponding isotypes, with deletion of the intervening DNA, switched B cells are not able to return to IgM production. Furthermore, IgM has a relatively short half-life of ∼5 days (6), and long-lived plasma cells producing IgM or IgM memory B cells against T cell–dependent antigens have not been described. The continuous presence of IgM against T cell–dependent antigens, therefore, points to continuous triggering of newly generated B cells.

Four subclasses of IgG (IgG1, IgG2, IgG3, and IgG4) can be distinguished in humans. The relative contribution of the different IgG subclasses to an antibody response depends on the nature of the antigen eliciting the response, repeated exposure to the antigen, as well as the cytokines produced in the vicinity of the B cells. Likewise, the route of entry, the dose of antigen, and the host's genotype can play a decisive role in the isotype usage of B cell responses (7).

IgA, IgM, and IgG subclasses display substantial differences in the ability to mediate effector responses. For example, immunoglobulin isotypes differ considerably in their ability to activate the complement system, with IgM and IgG3 being the most potent complement activators. A recent study demonstrated that the difference in recruiting cellular effector functions is a consequence of differential affinities of IgG subclasses for specific activating IgG Fc receptors compared with their affinities for the inhibitory IgG Fc receptor (8). Moreover, the different IgG Fc receptors are expressed on different effector cells, adding further to the differential ability of IgG subclasses to mediate effector responses (9).

The concepts described above (with the increase in antibody affinity and the occurrence of isotype switching being crucial for more efficient binding and neutralization of a pathogen and, thus, ultimately for the survival of the host) have been obtained mainly by studying responses against viral or bacterial antigens. It is unknown whether the principles applying to specific humoral immunity to infection also apply to the emergence of autoantibody responses. The dynamics of autoantigen-specific responses in relation to antibody isotype usage and levels have scarcely been studied, and little information is available regarding the evolution of the anti-CCP antibody response in patients with arthritis. For example, it is not known whether one “initial hit” is responsible for the continuous production of these antibodies or whether novel autoantigen-specific antibody-producing cells are continuously activated from newly activated mature B cells.

Because information on isotype distribution of the anti-CCP response could contribute to an understanding of the effector functions of these antibodies, and because the analysis of isotypes of anti-CCP antibodies early and later in the course of arthritis could provide insight into the underlying immune reaction, we investigated the presence and levels of IgM, IgA, and subclasses of IgG anti-CCP in patients with RA and patients with UA.

PATIENTS AND METHODS

Study population and serum samples.

Patients with UA or RA were selected from among patients in the Leiden Early Arthritis Clinic (EAC), which is an inception cohort of patients with arthritis of recent onset (symptom duration <2 years). The EAC was established at the Department of Rheumatology of the Leiden University Medical Center in 1993 and has been described in detail by van Aken et al (10). For all patients, a diagnosis was registered 2 weeks after the first visit. RA was diagnosed according to the American College of Rheumatology (ACR; formerly, the American Rheumatism Association) 1987 revised criteria for the classification of RA (11). Patients who could not be properly classified according to one of the ACR criteria were categorized as having UA.

After 1 year of followup, the disease status of all IgG anti-CCP–positive patients with UA was examined in order to determine whether RA (as defined according to the ACR criteria) had developed. Baseline serum samples were drawn at the first visit to the EAC. Additional serum samples obtained after 6–9 years (median 7 years) of followup were available for 64 IgG anti-CCP–positive patients in whom RA was diagnosed within 1 year after their first visit to the EAC. Informed consent was obtained, and the study was approved by the local medical ethics review board.

Anti-CCP autoantibodies.

Total IgG anti-CCP was assessed in baseline serum samples from patients with UA and patients with RA, by enzyme-linked immunosorbent assay (ELISA) (Immunoscan RA Mark 2; Euro-Diagnostica, Arnhem, The Netherlands). The cutoff level for IgG anti-CCP positivity was set at 25 units/ml, according to the manufacturer's instructions.

Measurement of anti-CCP antibody isotypes.

The levels of IgG1, IgG2, IgG3, IgG4, IgA, and IgM anti-CCP were measured in baseline serum samples obtained from 262 IgG anti-CCP–positive patients (152 with RA and 110 with UA) and in followup samples obtained from 64 patients with RA in whom the diagnosis was made within the first year. Levels of IgA and IgM anti-CCP were also determined in a group of 80 IgG anti-CCP–negative patients with RA.

The levels of IgG1, IgG2, IgG3, IgG4, IgA, and IgM anti-CCP were determined using the sandwich ELISA technique. Microtiter plates coated with CCP (Immunoscan RA Mark 2; Euro-Diagnostica) were incubated for 2 hours with serum samples, 100 μl/well, at a dilution of 1:50. This and each subsequent incubation step was performed at 37°C in a humidified atmosphere, followed by washing steps with washing buffer for the Immunoscan RA Mark 2 system. All samples and reagents were diluted in dilution buffer for Immunoscan RA Mark 2. For the detection of IgM and (total) IgA, the plates were incubated for 2 hours with either 100 μl/well goat anti-human IgM (1:1,000 dilution) or goat anti-human IgA (1:1,000 dilution) conjugated with alkaline phosphatase (catalog nos. AHI0605 and AHI0105; BioSource International, Camarillo, CA).

For detection of IgG subclasses, the plates were incubated for 2 hours with monoclonal mouse anti-human IgG subclass–specific antibodies, in a 1:10,000 dilution for IgG1 (antibody HP6188; Sanquin, Amsterdam, The Netherlands) and IgG3 (antibody HP6080; Nordic, Tilburg, The Netherlands), in a 1:1,000 dilution for IgG2 (antibody HP6002; SBA, Birmingham, UK), and in a 1:15,000 dilution for IgG4 (antibody HP6206; Nordic). All of the monoclonal antibodies that were used had been evaluated for their specificity in an International Union of Immunological Societies/World Health Organization collaborative study (12, 13). After incubation with the monoclonal antibodies, the plates were incubated overnight at room temperature, with 100 μl/well rabbit anti-mouse immunoglobulin (1:750 dilution) conjugated with alkaline phosphatase (Dako, Glostrup, Denmark). The presence of CCP-specific antibodies was detected using 4-nitrophenyl phosphate disodium salt (Sigma-Aldrich, Steinheim, Germany) as substrate, as previously described (5).

A series of successive dilutions of pooled patient sera that were positive for all isotypes of anti-CCP was used as a reference standard in all plates. Distinct dilutions of this standard (1:25 for IgG2, IgG3, IgA, and IgM; 1:50 for IgG4; 1:200 for IgG1) were defined as containing 1,000 arbitrary units (AU) per milliliter. The number of AUs per milliliter for one isotype is not comparable with the number of AUs per milliliter for other isotypes.

To control for the possibility that IgM rheumatoid factor (RF) interferes with the detection of anti-CCP of the IgM isotype, we depleted RF antibodies in a set of IgM- RF–positive, IgM anti-CCP–positive, IgG anti-CCP–positive sera, using IgG-coated CNBr-activated Sepharose beads. This did not result in a reduction of IgM anti-CCP levels. As an additional control, we mixed sera that were highly positive for IgM-RF and negative for IgM anti-CCP and IgG anti-CCP with sera that were IgM RF negative, IgM anti-CCP negative, and IgG anti-CCP positive, in order to analyze whether IgM anti-CCP would now be detected. This was not the case (data not shown). Moreover, several IgM-RF–positive sera were negative for IgM anti-CCP in the presence of IgG anti-CCP, and several IgM-RF–negative sera were positive for IgM anti-CCP, further indicating that IgM-RF did not lead to false-positive results for the detection of IgM anti-CCP.

Cutoff values and specificity control.

Cutoff values for the presence of IgG subclasses of anti-CCP antibodies were defined as the mean plus 2 SD for serum samples obtained from a group of 50 IgG anti-CCP–negative control subjects who did not have a diagnosis of RA or UA. This definition resulted in cutoff values for positivity of 2, 20, 52, and 0.1 AU/ml for IgG1, IgG2, IgG3, and IgG4, respectively. Microtiter plates coated with the same amount of the uncitrullinated control peptide were provided by the manufacturer (Euro-Diagnostica) and were used as a control for citrulline specificity of the anti-CCP antibodies. IgG subclass antibodies against the control peptide were not detected, as measured in 101 anti-CCP antibody–positive sera (data not shown).

The cutoff value for anti-CCP reactivity of IgA antibodies was set at 25 AU/ml. In a high proportion of sera (63%) with less than 25 AU/ml, the reactivity against the citrullinated peptide could not be distinguished from that against the uncitrullinated control peptide. Only 4 (2%) of the 162 IgA anti-CCP–positive sera reacted against the control peptide with similar optical density values. These 4 patients were considered negative for the presence of IgA anti-CCP antibodies (Table 1).

Table 1. Anti-CCP isotypes in IgG anti-CCP–positive patients with UA and IgG anti-CCP–negative patients with RA*
Anti-CCP isotypeUA (n = 110)RA (n = 152)OR (95% CI)P
  • *

    Values are the number (%). Anti-CCP = anti–cyclic citrullinated peptide; UA = undifferentiated arthritis; RA = rheumatoid arthritis; OR = odds ratio; 95% CI = 95% confidence interval.

  • By chi-square test.

IgA    
 Positive64 (58)94 (62)1.2 (0.7–2.0)0.55
 Negative46 (42)58 (38)  
IgM    
 Positive52 (47)93 (61)1.8 (1.04–3.0)0.03
 Negative58 (53)59 (39)  
IgG1    
 Positive107 (97)150 (99)2.10 (0.28–25)0.65
 Negative3 (3)2 (1)  
IgG2    
 Positive74 (67)125 (82)2.3 (1.2–4.2)0.005
 Negative36 (33)27 (18)  
IgG3    
 Positive43 (39)91 (60)2.3 (1.4–4.0)<0.001
 Negative67 (61)61 (40)  
IgG4    
 Positive104 (95)149 (98)2.9 (0.62–18.)0.17
 Negative6 (5)3 (2)  

For IgM, a cutoff value of 32 AU/ml units was determined by using the definition of the mean value plus 2 SD for the IgG anti-CCP–negative control population. Using this cutoff value, sera from 8 (5%) of the 153 IgM anti-CCP–positive patients reacted against the uncitrullinated control peptide as well. These 8 patients were considered to be negative for IgM anti-CCP, because no specific response was detectable (Table 1).

Statistical analysis.

Chi-square tests were performed to compare the proportions of individuals in the different groups who were positive for the various anti-CCP antibody isotypes. If one of the cells in the cross-table contained fewer than 6 subjects, P values were calculated using Fisher's exact test. Odds ratios were calculated in a case–control setting, in which the number of diagnoses of RA and the number of diagnoses of UA in patients with and those without the different isotypes of anti-CCP were compared. Relative risks for the development of RA were calculated based on the presence of the different anti-CCP isotypes in patients with UA. A t-test was used to compare the differences in levels of anti-CCP isotypes between the groups of patients studied. Differences in the frequencies and levels of anti-CCP isotypes between baseline and followup were tested using McNemar's test and a paired-samples t-test, respectively. Because the levels of all isotypes were not normally distributed, log transformation was performed first to normalize the data.

RESULTS

Isotypes of anti-CCP at baseline in patients with UA and patients with RA.

Different classes and IgG subclasses of anti-CCP isotypes were measured in baseline serum samples obtained from IgG anti-CCP–positive patients with UA (n = 110) and IgG anti-CCP–positive patients with RA (n = 152) (Table 1). No IgA anti-CCP or IgM anti-CCP was detected in 80 IgG anti-CCP–negative patients with RA, indicating that the occurrence of IgA and IgM anti-CCP is confined to IgG anti-CCP–positive patients. Among IgG anti-CCP–positive patients, those with RA more frequently displayed IgM (P = 0.03), IgG2 (P = 0.005), and IgG3 (P < 0.001) anti-CCP antibodies compared with patients with UA (Table 1). Among patients with UA, a median of 4 isotypes were used in the anti-CCP antibody response, compared with a median of 5 among patients with RA (P = 0.007).

A trend toward higher levels of anti-CCP antibodies in RA compared with UA was detected for all isotypes, when all samples were considered. The differences in levels were highest for IgM (P = 0.03), IgG2 (P = 0.002), and IgG3 (P < 0.001) (Figure 1A). However, the exclusion of samples that were negative for the respective isotypes revealed no differences in levels of anti-CCP isotypes between patients with RA and those with UA (P = 0.27–0.75), indicating that the level of these isotypes did not differ between patients with RA and patients with UA, but rather, that the number of antibody-positive patients was higher in the RA group than in the UA group.

Figure 1.

Dot plots showing baseline levels of anti–cyclic citrullinated peptide (anti-CCP) isotypes in A, 110 IgG anti-CCP–positive patients with undifferentiated arthritis (UA) and 152 IgG anti-CCP–positive patients with rheumatoid arthritis (RA) and B, 29 IgG anti-CCP–positive patients with UA who still had a diagnosis of UA after 1 year (UA–UA) and 74 IgG anti-CCP–positive patients with UA in whom RA developed within 1 year (UA–RA). P values for the group differences in mean log-transformed values (arbitrary units/ml) were calculated. When only patients who were positive for the respective isotypes of anti-CCP were analyzed, the P values for RA versus UA were 0.75 for IgA, 0.67 for IgM, 0.39 for IgG1, 0.27 for IgG2, 0.36 for IgG3, and 0.31 for IgG4; for UA–UA versus UA–RA, the respective P values were 0.47, 0.63, 0.23, 0.06, 0.90, and 0.01. The number of arbitrary units per milliliter for one isotype is not comparable with the number of arbitrary units per milliliter for another isotype. Bars show the geometric means and 95% confidence intervals. Broken lines indicate the cutoff values.

Thus, among IgG anti-CCP–positive patients, those with RA displayed a more diverse pattern of anti-CCP antibodies, as determined by the presence of different isotypes. No differences in mean levels were detected between the groups of patients who tested positive for the respective isotypes.

Anti-CCP isotypes at baseline in patients with UA in whom RA developed within 1 year (UA→RA) and in patients with UA in whom RA did not develop within 1 year (UA→UA).

Of the 110 anti-CCP–positive patients who had a diagnosis of UA at baseline, 74 had fulfilled the ACR criteria for RA after 1 year of followup, whereas 29 still had a diagnosis of UA. In 7 patients, other diseases had developed by this point in time. Inspired by our observation that patients with RA display a more extensive usage of anti-CCP antibody isotypes compared with patients with UA, we sought to determine whether the anti-CCP response in UA→RA patients differed from that in UA→UA patients.

IgA, IgM, IgG2, and IgG3 anti-CCP were present at higher frequencies in the UA→RA patients than in the UA→UA patients (P = 0.03, P = 0.01, P = 0.03, and P = 0.01, respectively) (Table 2). Among UA→UA patients, a median of 3 isotypes were used in the anti-CCP response, compared with a median of 5 among UA→RA patients (P = 0.004); this result served as another indication of more extensive anti-CCP isotype usage in UA→RA patients. A higher risk for the development of RA within 1 year of followup was observed in patients with UA who were positive for IgA anti-CCP (RR 1.3, 95% confidence interval [95% CI] 1.00–1.7), IgM anti-CCP (RR 1.4, 95% CI 1.1–1.8), or IgG3 anti-CCP (RR 1.4, 95% CI 1.11–1.8).

Table 2. Anti-CCP isotypes in IgG anti-CCP–positive patients with UA, according to the development of RA within 1 year*
Anti-CCP isotypeUA→UA (n = 29)UA→RA (n = 74)RR (95% CI)P
  • *

    Values are the number (%). Anti-CCP = anti–cyclic citrullinated peptide; UA = undifferentiated arthritis; RA = rheumatoid arthritis; UA→UA = patients with UA in whom RA did not develop within 1 year; UA→RA = patients with UA in whom RA developed within 1 year; RR = relative risk; 95% CI = 95% confidence interval.

  • Positive versus negative for the different anti-CCP isotypes.

  • By chi-square test.

IgA    
 Positive13 (45)50 (68)1.3 (1.00–1.7)0.03
 Negative16 (55)24 (32)  
IgM    
 Positive9 (31)43 (58)1.4 (1.1–1.8)0.01
 Negative20 (69)31 (42)  
IgG1    
 Positive27 (93)74 (100)0.08
 Negative2 (7)0 (0)  
IgG2    
 Positive16 (55)57 (77)1.4 (0.98–1.9)0.03
 Negative13 (45)17 (23)  
IgG3    
 Positive6 (21)37 (50)1.4 (1.11–1.8)0.01
 Negative23 (79)37 (50)  
IgG4    
 Positive27 (93)71 (96)1.2 (0.58–2.5)0.62
 Negative2 (7)3 (4)  

A trend toward higher levels of all isotypes of anti-CCP except IgG1 was observed in UA→RA patients compared with UA→UA patients, when all samples were taken into consideration (Figure 1B). When only those patients who were positive for a respective isotype were considered, only the levels of IgG4 anti-CCP were higher in UA→RA patients (P = 0.007).

Taken together, these results show that at the population level, the anti-CCP response in anti-CCP–positive patients with UA in whom RA was not diagnosed within 1 year was less diverse with respect to isotype usage compared with the response in patients in whom RA did develop, and that levels of most isotypes of anti-CCP were similar in both patient groups.

Changes in anti-CCP isotypes in patients with RA, after years of followup.

It has been shown for several antigens that repeated antigen exposure results in higher levels of antibodies of the IgG4 subclass (14). Because RA is a chronic disease and the autoantigens are expected to be present throughout the disease process (15), we hypothesized that after years of arthritic episodes, higher levels of IgG4 anti-CCP would have developed in patients with RA. Therefore, we next investigated whether the pattern of isotype usage changed during disease progression or whether the presence and/or levels of different anti-CCP isotypes remained relatively stable over time. To this end, IgA, IgM, and IgG subclasses of anti-CCP in 64 IgG anti-CCP–positive patients with RA were determined at baseline and after a median of 7 years of followup.

At the time of followup, the proportions of patients who were positive for IgA and IgG3 had decreased (P = 0.012 and P = 0.007, respectively) (Table 3), whereas no relevant changes were detected for IgM, IgG1, IgG2, and IgG4. A median of 5 isotypes were used at baseline, and a median of 4 were used at followup (P = 0.003). Among patients who were positive for a specific isotype of anti-CCP, a trend toward a decreased level of that isotype was observed after followup, for all isotypes except IgG1 (Figure 2 and Table 3). Thus, isotype usage in general (and IgG4 anti-CCP in particular) had not further increased during a median followup period of 7 years. Instead, the levels and extensiveness of isotype usage had declined during this period.

Table 3. Changes in the presence and levels of anti-CCP isotypes in paired serum samples obtained from 64 patients with RA at baseline and after a mean followup of 7 years*
IsotypeAnti-CCP isotype status at baseline/followupMedian level in anti-CCP– positive patients, AU/ml
+/++/−−/−−/+PBaselineFollowupP
  • *

    Except where indicated otherwise, values are the number of patients. Anti-CCP = anti–cyclic citrullinated peptide; RA = rheumatoid arthritis; AU = arbitrary units.

  • By McNemar's test, baseline versus followup.

  • By Wilcoxon's signed rank test, baseline versus followup.

IgA26161840.01267470.002
IgM28131580.27556460.015
IgG1603010.6251341650.530
IgG23614770.1899840<0.001
IgG318192250.00714449<0.001
IgG4584110.37526180.059
Figure 2.

Dot plots showing the levels of anti-CCP isotypes in paired serum samples obtained from 64 IgG anti-CCP–positive patients with RA at baseline (BL) and after a median of 7 years of followup (FU). P values were calculated for paired-sample t-tests on log-transformed values (arbitrary units/ml). The number of arbitrary units per milliliter for one isotype is not comparable with the number of arbitrary units per milliliter for another isotype. Bars show the geometric means and 95% confidence intervals. See Figure 1 for other definitions.

Triggering of new B cells after followup, as indicated by the appearance of IgM anti-CCP.

The presence of IgM anti-CCP in the setting of UA, early RA, and established RA suggests a constant recruitment of new anti-CCP–producing B cells from naive precursors. To substantiate this notion more accurately, we investigated whether IgG anti-CCP–positive patients who were negative for IgM anti-CCP at baseline became IgM positive at followup. Although the presence or absence of IgM anti-CCP seemed to be a relatively stable phenotype in two-thirds of the patients, 8 of the 23 patients who did not have IgM anti-CCP at baseline did display IgM anti-CCP at followup (Table 3 and Figure 3). Because switched B cells are unable to return to IgM production, the data showed that new B cells can be recruited into the anti-CCP response, further indicating that the immune reaction responsible for the production of anti-CCP is still ongoing in patients with established RA. Similar results were observed for IgA, IgG2, IgG3, and IgG4 anti-CCP, for which 4 patients, 7 patients, 5 patients, and 1 patient, respectively, were negative at baseline and positive at followup (Table 3 and Figure 3), again indicating an ongoing immune response in these patients.

Figure 3.

Levels of isotypes in IgG anti–cyclic citrullinated peptide (anti-CCP)–positive patients who were negative at baseline (BL) and positive after a median of 7 years of followup (FU) for IgA, IgM, IgG2, and/or IgG3 anti-CCP. For both IgG1 and IgG4 anti-CCP, only 1 patient changed from negative to positive (results not shown).

DISCUSSION

In this study, we analyzed the presence and levels of IgM, IgA, and subclasses of IgG anti-CCP in patients with RA and patients with UA, in order to determine whether the RA-specific anti-CCP response is stable, early and later in the course of RA. Furthermore, we attempted to determine whether the anti-CCP response reflects an ongoing immune reaction in which newly activated B cells are recruited.

The reported results indicate a more diverse pattern of isotype usage of anti-CCP antibodies in patients with RA compared with patients with UA (median 5 versus 4 isotypes; P = 0.007), with a higher prevalence of IgM, IgG2, and IgG3 anti-CCP in serum samples obtained from patients with RA compared with baseline serum samples obtained from patients with UA. Patients with UA in whom RA developed within 1 year of followup (UA→RA) displayed a more diverse pattern of isotypes of anti-CCP than did patients with UA in whom RA did not develop within 1 year (UA→UA) (median 5 versus 3 isotypes; P = 0.004). These results suggest that early during the course of disease progression from UA to RA, isotype switching is occurring. Alternatively, these data suggest that disease that is more severe at the time of onset, as reflected by earlier fulfillment of more of the ACR criteria for RA, is accompanied by a more diverse pattern of anti-CCP antibody isotypes.

IgM responses against T cell–dependent antigens are, in general, not continuously present. In the setting of vaccinations, for example, levels of antigen-specific IgM increased and decreased during the weeks after a primary or a booster immunization against rabies (5). Similarly, 10 weeks after vaccination, the proportion of measles-specific IgM–positive individuals dropped to <10% (16). On the basis of the nature of the antigen (i.e., protein) and the association between the presence of anti-CCP antibodies with HLA (17), it is likely that the anti-CCP B cell response is T cell dependent. Intriguingly, even though IgG anti-CCP has been detected years before the first symptoms of arthritis (18, 19), our data showed the presence of IgM anti-CCP in a considerably large proportion of early samples from IgG anti-CCP–positive patients with UA or RA as well as in followup samples from IgG anti-CCP–positive patients with RA. Additional regression analyses did not reveal a correlation between the duration of symptoms at baseline or the duration of followup and the presence or absence of IgM anti-CCP at either or both time points (data not shown), indicating that these time intervals did not influence the (change in) presence of IgM anti-CCP.

Because IgM antibodies have a half-life of only ∼5 days (6), and long-lived plasma cells producing IgM antibodies or IgM memory B cells against T cell–dependent antigens have not been described, the presence of IgM most likely reflects the presence of recently activated IgM-producing B cells. We additionally observed that in some IgG anti-CCP–positive patients with RA in whom no IgM anti-CCP was detectable at baseline, IgM anti-CCP was present 7 years later. Taken together, these results are important, because they indicate that novel IgM-producing B cells are continuously recruited to the anti-CCP response, demonstrating that the anti-CCP response is continuously reactivated during the course of arthritis.

Although the presence of IgG4 anti-CCP has been described previously (20), the observation that IgG4 anti-CCP antibodies were detected at a lower frequency after long-term followup of patients with RA is of interest, because it was hypothesized that patients with RA of long duration would demonstrate a higher frequency of IgG4 anti-CCP antibodies. IgG4 is expressed predominantly under conditions of long-term exposure to protein antigens; this is well illustrated by the longitudinal analysis of the antibody response to bee venom in beekeepers (14) and by the hyposensitization protocols performed in patients with allergies (for review, see ref.21). Consistent with these observations, the finding that the frequency of IgG4-positive patients with UA or RA was relatively high at the first visit to the EAC could indicate that long-term exposure of autoreactive B cells to citrullinated antigens already occurred early in the course of symptomatic disease.

After 7 years of followup, not only had the presence and levels of IgG4 anti-CCP decreased, but the levels of the other isotypes (except IgG1) also had decreased. A possible cause of this limited anti-CCP isotype usage at a later time point could be treatment with immunosuppressive medication. However, given the fact that all patients with RA were receiving treatment when the followup serum samples were obtained, the impact of treatment in the present study cohort is difficult to determine and will be the subject of further investigation.

As in other autoimmune diseases, the isotypes of autoantibodies may be of prognostic value. For example, IgG1 and IgG3 isotypes of islet cell autoantibodies in prediabetic children have been associated with progression to type 1 diabetes mellitus (for review, see ref.22), while IgG4 and IgE autoantibodies have been associated with protection against type 1 diabetes mellitus (23, 24). Similar associations with progression from UA to RA and the presence of 1 or 2 particular isotypes of anti-CCP in patients with UA were not observed in this study (Table 1), although patients with UA who were harboring ≥4 different isotypes displayed a 1.4-fold higher risk for the development of RA within 1 year in comparison with patients who harbored ≤3 isotypes (95% CI 1.01–1.80; data not shown).

Considering the number of hypotheses tested, it can be argued that a correction for multiple testing should be performed. The presence and levels of the different isotypes, however, are not independent phenotypes (data not shown), and the hypotheses tested are far from independent. Determining the correct adjustment strategy is, therefore, not straightforward. Because the main conclusions in this study are drawn from a collection of observations rather than from single hypotheses tested, we chose to mention P values without adding a specific label to the level of significance, and we wish to mention that a P value less than 0.05 may not be statistically significant in the context of a single observation.

In conclusion, the presence of IgM anti-CCP in early serum samples obtained from both patients with UA and patients with RA and in followup samples obtained from patients with RA suggests an ongoing activation of new clones of anti-CCP–producing B cells. This notion is further supported by the observation that IgG anti-CCP–positive patients who do not display IgM anti-CCP can convert to IgM anti-CCP positivity later in the course of arthritis. Furthermore, relatively extensive isotype usage in the anti-CCP response was detected in patients with recent-onset arthritis. Taken together, these data indicate that full usage of the isotype repertoire occurs early in the course of arthritis, and that a continuous (re)activation of the RA-specific anti-CCP antibody response occurs during the disease course.

Acknowledgements

We thank Euro-Diagnostica (Arnhem, The Netherlands) for supplying the microtiter plates coated with uncitrullinated control peptide.

Ancillary