Antibodies against cyclic citrullinated peptide are associated with HLA–DR4 in simplex and multiplex polyarticular-onset juvenile rheumatoid arthritis




Anti–cyclic citrullinated peptide (anti-CCP) antibodies have been detected in patients with juvenile rheumatoid arthritis (JRA), particularly in those with polyarticular, rheumatoid factor (RF)-positive JRA. Our objectives were to determine whether anti-CCP antibodies are associated with HLA–DR4 in children with polyarticular JRA, whether anti-CCP antibodies are associated with clinical features of disease, and whether affected sibling pairs (ASPs) with JRA are concordant for this antibody.


Stored serum samples obtained from 230 HLA-typed patients with JRA (77 with polyarticular-onset disease and 153 with pauciarticular- or systemic-onset disease), 100 JRA ASPs, and 688 healthy children were tested for anti-CCP antibodies and RF.


Thirteen percent of the patients with polyarticular-onset JRA and 2% of the other JRA patients exhibited anti-CCP antibodies, compared with only 0.6% of the controls. Fifty-seven percent of RF-positive patients with polyarticular-onset JRA had anti-CCP antibodies. HLA–DR4–positive patients with polyarticular-onset JRA were more likely to have anti-CCP antibodies than were those without HLA–DR4 alleles (odds ratio [OR] 5.20, 95% confidence interval [95% CI] 1.30–20.9). Anti-CCP antibodies were associated with polyarticular onset (OR 7.46, 95% CI 1.99–28.0), a polyarticular disease course (OR 9.78, 95% CI 1.25–76.7), and erosive disease (OR 14.3, 95% CI 3.01–67.9). Concordance rates for anti-CCP antibodies among ASPs were statistically significant.


These data demonstrate increased anti-CCP antibody formation in HLA–DR4–positive patients with polyarticular-onset JRA. The overall prevalence of anti-CCP antibodies in JRA is low, but a substantial proportion of RF-positive patients with polyarticular-onset JRA have these antibodies. Anti-CCP antibodies in JRA are associated with polyarticular onset, a polyarticular course, and erosive disease.

Juvenile arthritis, whether classified as juvenile rheumatoid arthritis (JRA) according to the American College of Rheumatology (ACR) criteria (1) or as juvenile idiopathic arthritis (JIA) according to the International League of Associations for Rheumatology (formerly, the International League Against Rheumatism) criteria (2), is the most common chronic autoimmune rheumatic disease of childhood. It is characterized by the presence of chronic arthritis before the age of 16 years that lasts for at least 6 weeks, and represents a heterogeneous group of patients with different disease subtypes. Each subtype exhibits its own clinical, immunologic, and genetic characteristics. Using the ACR criteria, JRA comprises 3 different types of onset and 3 different types of disease course, which makes JRA largely different from adult-onset RA. The exception is that ∼5% of patients with JRA experience disease onset at or after puberty and are positive for both IgM rheumatoid factor (RF) and HLA–DR4, representing the childhood equivalent of adult-onset RA. In contrast, in some younger patients with JRA, especially those with pauciarticular-onset disease, HLA–DR4 haplotypes have a protective effect (3–5).

Recently, antibodies to citrullinated proteins have received a great deal of attention in terms of their use as a novel and more specific test that can aid in the diagnosis of adult RA and in the determination of disease prognosis (6–8). This autoantibody system includes antiperinuclear factor, antikeratin antibodies, antifilaggrin antibodies, anticitrullinated fibrin antibodies, and anti-Sa (citrullinated vimentin) antibodies (9–14). Based on the development of a synthetic cyclic citrullinated peptide (CCP) by Schellekens et al, for which serum reactivity can be assessed by enzyme-linked immunosorbent assay (ELISA) (15), the antibodies are now commonly referred to as anti-CCP antibodies. Anti-CCP antibody ELISA testing has shown a specificity of 98% in sera from adults with established RA and 96% in sera from adults with early RA (15). The sensitivity of this anti-CCP antibody ELISA was 68% and 48% in sera from patients with established RA and early RA, respectively (15). Using the second-generation anti-CCP assay (CCP-2) in patients with established RA, the sensitivity of the test improved to 82%, with a specificity of 98.5% (16).

Anti-CCP antibodies have been detected in the serum of children with chronic arthritis, particularly those with polyarticular-onset, RF-positive disease (17, 18). In adults, the presence of anti-CCP antibodies has been associated with specific HLA alleles constituting the shared epitope (SE) (19, 20). This association of anti-CCP antibodies with the SE has not been evaluated in JRA.

In this study, we addressed 2 questions, as follows. First, are there any associations between HLA haplotypes (specifically HLA–DR) and the presence of anti-CCP antibodies? Second, do anti-CCP antibodies in a JRA cohort including simplex and multiplex families have any associations with clinical phenotypes? To address the latter question, we tested for associations between anti-CCP antibodies and several clinical manifestations and laboratory findings of the disease, including the presence of erosions, antinuclear antibodies (ANA), and iritis. We further assessed the relationship between anti-CCP antibodies and both onset type and disease course and evaluated multiplex families for the degree of concordance of anti-CCP antibodies.


Study population.

A total of 330 patients with JRA were included in this study: 117 simplex (only 1 affected child per family) patients recruited from among a local cohort (Cincinnati) and 213 multiplex patients from the National Institute of Arthritis and Musculoskeletal and Skin Diseases–supported JRA Affected Sibpair Registry, who were recruited from throughout North America. Only 1 affected child per sibpair family (113 total) was included for comparisons between JRA patients and controls. The sibling to be included in the study was selected randomly, using random number assignment. The ASP registry and the local cohort populations differed only in the extent of joint involvement but not in terms of other disease characteristics (21, 22). For determinations of concordance among sibpairs, 100 sibpairs (200 individuals) were evaluated, including those who were excluded from the original analyses.

Diagnoses were validated using the ACR criteria (1) relative to disease subtype and course. Charts were reviewed, and patients were classified as having pauciarticular, polyarticular, or systemic onset according to the number of affected joints and the presence or absence of systemic features. The type of disease course was defined by the number of joints that were affected 6 months after disease onset. Because of the existence of age-related clinical phenotypes and previous HLA-related findings (23), the patients were further subdivided into early-onset and late-onset types based on the development of disease before or after the sixth birthday.

The diagnosis of iritis was determined from available ophthalmology information in the medical chart. Each patient's ANA status was retrieved from the medical record. Radiology reports of joint space narrowing and/or the existence of bone erosions provided the basis for labeling the patient's disease as erosive.

Controls for these analyses were selected from the prospective Diabetes and Autoimmunity Study in the Young (DAISY), which follows up children who are at increased risk for type 1 diabetes mellitus, as defined by possession of HLA–DR4 and/or DR3 alleles (24). The most recent serum samples obtained from these children were used for analysis. No parent indicated during the visit for sample collection, or on previous visits, that the child exhibited signs or symptoms of arthritis, and no controls had a known diagnosis of JRA.

HLA typing.

Low-resolution HLA–DRB1 typing was performed using allele-specific polymerase chain reaction (GenoVision, West Chester, PA). This test includes 13 distinguishable DRB1 alleles. For the controls, HLA typing was done at Roche Molecular Systems (Alameda, CA) (24).

Autoantibody testing.

RF was measured in thawed sera obtained from study subjects, using nephelometry according to the manufacturer's specifications (Dade-Behring, Newark, DE). A positive test was defined by a level of >15 IU/ml. A specific ELISA (anti–CCP-2) was used to measure IgG antibodies to CCP (Diastat; Axis-Shield Diagnostics, Dundee, Scotland, UK). A positive test result was defined by a level of >5 units/ml.

Statistical analysis.

Categorical data were analyzed using the chi-square test or Fisher's exact test for proportions, as appropriate. Odds ratios (ORs) were calculated and reported with 95% confidence intervals (95% CIs). For comparisons of means, either a t-test or a Wilcoxon's rank sum test was used, as appropriate. Concordance rates were evaluated using the kappa statistic. P values less than 0.05 were considered significant. Statistical analysis was performed using SAS software, version 8.2 (SAS Institute, Cary NC). The figures were created using Sigma Plot 8.02 software (SPSS, Chicago, IL).


Study populations.

Stored sera from 330 patients with JRA and 688 controls were tested for anti-CCP antibodies and RF. The 330 study subjects included 117 simplex JRA patients and 113 multiplex JRA patients who were used in the comparisons between patients with JRA (n = 230) and controls. An additional 100 multiplex subjects were used only to establish concordance rates among ASPs. The characteristics of the 230 patients with JRA and 688 controls are shown in Table 1. The polyarticular-onset group and the pauciarticular-onset group were composed of predominantly female patients. The mean age at the time of serum sample collection was younger in controls (6.3 years) than in patients with JRA (13.2–17.4 years) (P < 0.0001). The mean disease duration at the time of serum sample collection was 9.69 years in the overall JRA cohort (range of 8.8 years to 11.1 years among groups representing the different onset types). The rate of conversion from pauciarticular-onset disease to a polyarticular course of JRA was 29%, while 50% of the patients with systemic-onset disease progressed to a polyarticular disease course.

Table 1.  Characteristics of the study groups*
 JRA onset typeControls (n = 688)
Polyarticular (n = 77)Pauciarticular (n = 139)Systemic (n = 14)
  • *

    JRA = juvenile rheumatoid arthritis; NA = not applicable.

No. (%) female58 (75)105 (76)7 (50)322 (47)
Mean age at time of sample collection, years17.413.214.66.3
Mean age at disease onset, years6.44.34.4NA
Mean disease duration at time of sample collection, years11.18.810.3NA
Disease course, no.    

Association of anti-CCP antibodies and RF with JRA onset type and disease course.

The results of autoantibody testing, presented according to the type of onset of JRA, are summarized in Table 2 and are depicted graphically in Figure 1. Anti-CCP antibodies were present in 13 (5.6%) of 230 JRA patients and in 4 (0.58%) of 688 controls. Anti-CCP antibodies were more likely to be positive in patients with polyarticular-onset JRA than in those with either pauciarticular-onset or systemic-onset JRA (13% versus 2%; OR 7.46, 95% CI 1.99–28.0). RF was positive in 25 patients with JRA (11%), including 18% of those with polyarticular-onset JRA, versus 3.3% of controls. Anti-CCP antibodies were more likely to be present in patients with polyarticular-onset, RF-positive disease than in patients with all other JRA subtypes (8 [57%] of 14 versus 5 [2.3%] of 216; P < 0.0001). In 9 patients with JRA (3.9%), both anti-CCP antibodies and RF were detected. Of these 9 patients, 8 had polyarticular-onset disease, and the remaining patient had pauciarticular-onset JRA with a subsequent polyarticular disease course.

Table 2.  Anti-CCP and RF levels in patients with JRA and controls*
 JRA onset typeControls (n = 688)
Polyarticular (n = 77)Pauciarticular (n = 139)Systemic (n = 14)
  • *

    Anti–cyclic citrullinated peptide (anti-CCP) positivity was defined as >5 units/ml; rheumatoid factor (RF) positivity was defined as >15 IU/ml. JRA = juvenile rheumatoid arthritis.

 Mean (median) units/ml9.72 (0.32)1.20 (0.32)0.58 (0.35)0.37 (0.24)
 No. (%) positive10 (13)3 (2.2)0 (0)4 (0.58)
 Mean (median) IU/ml48.6 (11)13.2 (11)34.5 (10.2)9.23 (8.4)
 No. (%) positive14 (18)9 (6.5)2 (14)23 (3.3)
No. (%) positive for anti-CCP and RF8 (10)1 (0.72)0 (0)0 (0)
Figure 1.

Anti–cyclic citrullinated peptide (anti-CCP) levels according to juvenile rheumatoid arthritis (JRA) onset type. The maximum anti-CCP antibody level was >100. For graphic purposes, all values >100 are plotted as 100.

Data on the course of disease were available for 209 subjects (91%). Among patients for whom information on disease course was available, anti-CCP antibodies were significantly associated with a polyarticular disease course compared with either a pauciarticular or systemic course (OR 9.78, 95% CI 1.25–76.7). RF was also associated with a polyarticular disease course (OR 3.16, 95% CI 1.13–8.82). Figure 2 graphically depicts individual anti-CCP antibody test results according to disease course and shows that anti-CCP antibody levels were higher in those patients with a polyarticular course.

Figure 2.

Anti–cyclic citrullinated peptide (anti-CCP) levels according to the course of juvenile rheumatoid arthritis (JRA). The maximum anti-CCP antibody level was >100. For graphic purposes, all values >100 are plotted as 100.

Association of HLA–DR alleles with anti-CCP antibodies and RF.

Patients who possessed the HLA–DR4 allele were more likely to have anti-CCP antibodies than were those who did not possess HLA–DR4. This observation was true when all JRA patients were tested (OR 7.00, 95% CI 2.20–22.2) and when only subjects with polyarticular-onset JRA were evaluated (OR 5.20, 95% CI 1.30–20.9). In the control group, 4 of 688 children had anti-CCP antibodies, and all 4 possessed the HLA–DR4 allele. RF was also more likely to be positive in patients with HLA–DR4, but this association was significant only when it was tested in the combined JRA group (OR 2.74, 95% CI 1.09–6.92), and not when tested in the polyarticular-onset group only (OR 2.40, 95% CI 0.72–8.02) or in controls (OR 0.63, 95% CI 0.27–8.02). No subject was homozygous for HLA–DR4.

The relationship between other HLA types and anti-CCP antibodies or RF was also tested, and the results are shown in Table 3. Neither HLA–DR1, DR8, nor DR11 was significantly associated with positivity for either anti-CCP antibodies or RF. No patient with the HLA–DR8 allele was positive for anti-CCP antibodies, while 4 of 62 patients with HLA–DR1 and 2 of 65 patients with HLA–DR11 were positive for anti-CCP antibodies. The age at onset of JRA was higher in the HLA–DR4–positive group, when compared with the combined group of patients with all other HLA–DR alleles (P = 0.0056).

Table 3. HLA associations with CCP, RF, and age*
VariableHLA–DR4 (n = 38)HLA–DR1 (n = 62)HLA–DR8 (n = 52)HLA–DR11 (n = 65)
  • *

    For age at disease onset, data were available for 36 of the patients in the HLA–DR4 group, for 61 of the patients in the HLA–DR1 group, for all patients in the HLA–DR8 group, and for 63 of the patients in the HLA–DR11 group. CCP = cyclic citrullinated peptide; RF = rheumatoid factor.

Anti-CCP positive, no. (%)7 (18)4 (6)0 (0)2 (3)
RF positive, no. (%)8 (21)7 (11)2 (4)4 (6)
Age at disease onset, mean (median) years7.2 (6.7)6.1 (5.4)3.7 (2.7)4.2 (3.1)

Association of anti-CCP antibodies and RF with clinical features of disease.

The relationship between the presence of anti-CCP antibodies or RF and clinical features of JRA, including age at onset, radiographic erosions, iritis, and positive ANA, were evaluated. The age at onset of JRA was later in patients positive for anti-CCP antibodies than in those who were anti-CCP negative, with a mean age of 9.49 years in the anti-CCP antibody–positive group versus 4.70 years in the anti-CCP antibody–negative group (P = 0.0007). The mean age at onset of JRA in RF-positive patients was 7.09 years, versus 4.72 years in RF-negative patients (P = 0.029).

Radiographs were available for 76% of the JRA study population. Among 174 subjects for whom radiographic data were available, the presence of anti-CCP antibodies was significantly associated with erosions in all JRA patients (OR 14.3, 95% CI 3.01–67.9) and in the polyarticular-onset subgroup (OR 12.6, 95% CI 1.46–108). In addition, RF was associated with erosions in the combined JRA group (OR 12.0, 95% CI 3.74–38.3). In the polyarticular-onset subgroup, there was also a strong association between RF and erosions (P < 0.0001), although the OR could not be calculated due to the absence of any RF-positive subject without erosions.

Information on the presence or absence of iritis was available from the medical records of 192 JRA patients (83%), and results of ANA testing were available for 193 patients (84%). There was no significant association of anti-CCP antibodies or RF with the presence of iritis or ANA positivity, in either the entire JRA cohort or the polyarticular-onset group alone.

Concordance of autoantibodies in JRA ASPs.

The concordance of autoantibodies in multiplex families was evaluated in 100 ASPs. For the previous analysis, only 1 affected child per sibpair was included, but to evaluate concordance, we tested for anti-CCP antibodies in the second sibling. A total of 7 patients were anti-CCP antibody positive. This included 2 sibpairs that were concordant and 3 pairs that were discordant for the presence of anti-CCP antibodies. In ASPs, the rate of concordance for anti-CCP antibody positivity was 2%, concordance for anti-CCP antibody negativity was 95%, and discordant results for the presence of anti-CCP antibodies were present in 3% (κ = 0.5588, P = 0.002). Of the 2 sibpairs that were concordant for anti-CCP antibodies, 3 children had polyarticular-onset JRA, and 1 had pauciarticular-onset disease; in this child, a polyarticular disease course developed 6 months after the onset of JRA. The sibpair that was concordant for polyarticular-onset JRA and for anti-CCP antibodies was also concordant for HLA–DR4, while neither sibling in the other concordant pair had the HLA–DR4 allele.


The results of our studies show that anti-CCP antibodies are significantly associated with the presence of HLA–DR4 alleles in patients with JRA. Anti-CCP antibodies are also associated with polyarticular onset, a polyarticular course, and erosive disease. Similar associations were observed with RF, and those patients who had polyarticular-onset JRA and were RF positive were much more likely to have anti-CCP antibodies.

JRA is distinct from RA, except for a particular subset of patients. It is well known that in patients with late onset of polyarticular disease (usually RF-positive female patients), a disease very similar to RA develops. Not surprisingly, this is the group of patients in whom anti-CCP antibodies have been shown to be more prevalent, although in a smaller proportion of patients than that reported in RA (18).

Anti-CCP antibodies have been shown to be present in the serum of patients with JRA. Two recent European studies showed prevalence rates of 2% and 15% among all JIA subtypes (17, 18). In one study, only 1 patient with polyarticular-onset, RF-positive disease was included, and that patient was negative for anti-CCP antibodies. The other study showed that among 11 patients with polyarticular-onset, RF-positive disease the prevalence of anti-CCP antibodies was 73%. This value is similar to the prevalence of this autoantibody in adults with RA. We observed a prevalence of anti-CCP antibodies of 5.6% when all JRA subtypes in our cohort were included. When only patients with polyarticular disease were included, anti-CCP antibodies were observed in 13%, and further limiting this cohort to patients with polyarticular-onset, RF-positive disease resulted in a prevalence of 57%. These observations suggest that the presence of anti-CCP antibodies is an immunologic feature in this particular subset of patients, who have the childhood equivalent of adult-onset RA.

Our results also established an association between the presence of anti-CCP antibodies and a more severe polyarticular disease course characterized by the presence of erosions. Our study was not prospective, and the time course for detection of anti-CCP antibodies and development of erosions cannot be determined. In addition, we had radiographic data for only 76% of our subjects, leading to a concern that our analyses might not have been representative of the entire cohort. However, there was little difference in mean disease duration (9.8 years versus 9.2 years) and onset type (for pauciarticular onset, 57% versus 70%; for polyarticular onset, 36% versus 25%; and for systemic onset, 6% versus 5%) between those subjects who had radiographs and those who did not, respectively, suggesting that the missing data most likely led only to a reduction in power, not to bias. Although prospective studies are needed to determine the prognostic significance of anti-CCP antibodies in erosive disease, it is possible that the presence of these antibodies at an early stage in the subset of patients with polyarticular-onset JRA might be useful in identifying patients who are at a higher risk for the development of more aggressive disease.

Associations between different HLA alleles and JRA subtypes have been described. The HLA–DR4 allele, which has been strongly associated with adult-onset RA, confers risk for the development of polyarticular-onset JRA but has a protective effect against the development of pauciarticular-onset JRA in younger children (3–5). Other JRA-associated HLA–DR alleles (i.e., HLA–DR1, DR8, and DR11) were not associated with anti-CCP antibodies. Although HLA–DR4 is observed in the subset of patients with an older age at disease onset, HLA–DR11 and DR8 are found in the group with a younger age at disease onset, which is consistent with the associated subtype distribution (23).

Even though HLA–DR1 can be present in both patients with younger age at the time of disease onset and in those with older age at onset and is part of the SE, we did not find any association between the presence of this allele and anti-CCP antibodies. HLA–DR subtyping was not performed, and it is possible that patients in the study who possessed HLA–DR1 alleles did not possess the SE, which is found only in association with the HLA–DRB1*0101 subtype. Another possibility is that the power of the study was inadequate to detect a weaker association between HLA–DR1 and anti-CCP antibodies.

The presence of HLA–DR4 was associated with the presence of anti-CCP antibodies in the JRA population; these findings are similar to those in RA. It has been proposed that the interaction between citrullinated self peptides and the SE carried by HLA–DR alleles such as DRB1*0404 and *0401 may result in the induction of anti-CCP antibodies in patients with adult-onset RA (25). In fact, citrullinated peptides fit better in the HLA–DR4 antigen binding groove than do similar arginine-containing peptides, linking anti-CCP antibodies to the SE hypothesis of RA (25). Our data demonstrate that a similar response may occur in the subset of patients with RF-positive, polyarticular-onset JRA.

An association of HLA–DR4 and anti-CCP antibodies was also observed in healthy children enrolled in a prospective study of diabetes (DAISY) (24). Anti-CCP antibodies were observed in <1% of these children, but all children with anti-CCP antibodies carried the HLA–DR4 allele. This association with HLA–DR4 was not demonstrated for RF. This finding provides further support for the theory that citrullination of peptides allows for a better fit in the HLA–DR4 antigen binding groove, leading to production of anti-CCP antibodies.

In multiplex families with JRA, concordance of clinical phenotype for this cohort has been reported previously (22). We evaluated concordance of anti-CCP antibodies in multiplex families with JRA and found that the majority were concordant negative for anti-CCP antibodies, 2% were concordant positive, and 3% were discordant. The concordance of autoantibodies was statistically significant, but the high rate of concordant-negative values, along with the low frequency of anti-CCP antibody positivity, may limit the clinical significance and temper interpretation of the kappa statistic. The most common type of JRA onset in the multiplex population is pauciarticular, and this onset type is not associated with anti-CCP antibodies. This likely explains the low frequency of anti-CCP antibodies and the difficulty with interpreting concordance rates in this population. In previous studies of the same population, a high concordance rate for the presence or absence of ANA and RF was observed, suggesting that multiplex families have similar serologic features, and the current findings may add to this conclusion (22).

The association between HLA–DR4 and anti-CCP antibodies was also observed in the multiplex population. It is noteworthy that both sibpairs concordant for anti-CCP antibodies were also concordant for the subtype of JRA with a polyarticular course. This finding reinforces the notion that anti-CCP antibodies are associated not only with onset type but also, and possibly more clinically relevant, with disease course. As discussed above, this was not a prospective study, and a predictive value of anti-CCP antibodies for the development of a polyarticular course of JRA cannot be assigned.

In this study, an existing registry was used to test for autoantibodies that were not included as part of the original study design. There were some limitations of the study design, including the fact that some of the data in the registry were based on chart review, and some data were missing. The chart review method could specifically affect the ability to detect iritis, because some cases of iritis might be missed when assessment was performed in this way. Although some data were missing, we believe that for the majority of patients clinical information was complete enough to allow valid statistical analysis. Specifically, in 76–91% of subjects data on radiographic erosions, iritis, ANA, and disease course were available.

The controls used in this study were children enrolled in a prospective study of the development of type 1 diabetes. These children were healthy but were at higher risk for the development of autoimmune diabetes, and thus the prevalence of anti-CCP and RF autoantibodies may have been slightly higher than that in the general population. Therefore, if anything, this may have resulted in an underestimation of the difference in antibody prevalence between certain types of JRA and the general population of children. The healthy controls in this study were younger than the patients with JRA, and it is possible that the rate of anti-CCP antibody positivity would be higher in older healthy children. However, the rate of anti-CCP antibody positivity in healthy adults is also low, and we would not expect the age difference to significantly affect the results.

Another limitation of the study is that in the patients with JRA, serum samples were obtained several years after the onset of disease. Therefore, we cannot comment on the likelihood of anti-CCP antibodies being observed early in JRA. In addition, tests for anti-CCP antibodies were conducted at only 1 time point, making it impossible to determine the time course for development of these antibodies. We have made the assumption that anti-CCP antibodies remain present over time, without significant seroconversion from either negative to positive or positive to negative. This assumption is based in part on observations by Mikuls et al, showing that anti-CCP antibodies do not convert from positive to negative following the initiation of disease-modifying antirheumatic drug therapy (26). If there were a high rate of conversion from negative to positive, anti-CCP antibodies would not be as useful in predicting the prognosis in patients with early-onset JRA. Conversely, if a high rate of conversion from positive to negative was associated with control of disease, the results of this study might underestimate the prevalence of anti-CCP antibodies in JRA. Analogous to RA, in which anti-CCP antibodies have been observed both prior to the diagnosis of RA and in early undifferentiated arthritis, we would expect anti-CCP antibodies to be present early in JRA, but prospective studies are needed to determine whether this is true. In addition, we would expect the presence or absence of anti-CCP antibodies to remain relatively constant over time, but this could be proven only by prospective studies.

Because JRA comprises many disease types, we thought it was necessary to analyze our data within disease-type subgroups, even though this resulted in small numbers and large confidence intervals for some of the analyses. We would not have been able to detect many of these clinically relevant associations without this type of subgroup analysis.

The association of the presence of anti-CCP antibodies and a polyarticular disease course should be further investigated. Future studies will prospectively determine the prevalence of this antibody prior to or at the onset of disease, as well as further delineate its value in the prediction of outcome, especially in the subset of older patients with polyarticular-onset JRA.