Dr. Liang has received consulting fees, speaking fees, and/or honoraria from Roche, Amgen/Wyeth, Bristol-Myers Squibb, Pfizer, and Abbott (less than $10,000 each) and owns stock or stock options in Merck Frosst.
Henri A. Ménard,
McGill University Health Centre, Montreal, Quebec, Canada
Dr. Boire has received consulting fees, speaking fees, and/or honoraria from Bristol-Myers Squibb, Abbott Canada, Hoffmann-La Roche, Amgen Canada, Wyeth Canada, Pfizer Canada, and UCB Canada (less than $10,000 each) and owns stock or stock options in Pfizer.
To define the association of alleles encoding the HLA–DR rheumatoid arthritis (RA) protective epitope (DERAA) with the presence of RA-associated antibodies at study inclusion and with severe outcome in patients with early polyarthritis (EPA).
Consecutive EPA patients (n = 210) were evaluated early (mean of 4.8 months after diagnosis) and prospectively (for 30 months). HLA class II typing was performed by polymerase chain reaction using sequence-specific primers, and HLA–DR alleles DERAA, RA-associated shared epitope (SE), and non-SE/non-DERAA (neither SE nor DERAA) were identified. RA-associated antibodies identified were anti-Sa/citrullinated vimentin, anti–cyclic citrullinated peptide 2, and IgM rheumatoid factor. Severe disease was defined according to a preset threshold of joint destruction and/or functional limitation.
DERAA and SE alleles were present in 62 and 110 of the 210 EPA patients, respectively. At 30 months, severe disease was present in 78 patients (37%). In contrast to SE alleles, DERAA alleles were not associated with the production of RA-associated antibodies, but were strongly protective against severe disease at 30 months (odds ratio 0.30, P < 0.001). DERAA alleles emerged as a strong, independent protective marker on multivariate analysis. The protective effect of DERAA was seen only in patients who did not already have erosions at study inclusion, was independent of the presence of antibodies, but was not associated with spontaneous remission.
In our EPA cohort, the presence of a DERAA sequence was a strong independent predictor of a better prognosis, but only in the absence of erosive disease that was already present at inclusion. Identification of DERAA alleles may help in managing the large subgroup of EPA patients who do not have erosions at baseline.
The chronic inflammatory arthritides are a group of common complex diseases with distinct clinical and biologic characteristics (1). However, even within a single diagnostic entity such as rheumatoid arthritis (RA), patients differ markedly in their clinical course and outcomes. Establishing an individual's prognosis is even more complex in patients with early inflammatory joint disease (or, early polyarthritis [EPA]) because the more characteristic features, such as bone erosions in RA, appear relatively late in most patients (2–6). The current emphasis on early treatment of patients with EPA further complicates the issue by preventing or retarding the development of diagnostic features of a specific disease. It is with those considerations in mind that we elected to adopt the clinician's perspective and approached EPA patients from a prognostic, rather than a diagnostic, angle. This approach would allow us to better target the patients who need early aggressive interventions and spare many other patients the potential side effects of such interventions.
Many models have been proposed to establish prognosis in EPA patients. One of the most popular relies on a combination of clinical, serologic, and radiologic variables to predict the persistence of arthritis and the development of erosions (7). To build on these models, we previously focused on the evaluation of a powerful serologic marker for poor outcomes, anti-Sa antibodies (8). Another proposed prognostic biomarker is the RA-protective subset of HLA–DR alleles defined by the presence of an aspartic acid residue at position β70 of the HLA–DRB1 molecule (9). This biomarker was redefined as the “RA protective epitope” group of alleles and was defined according to the DERAA amino acid sequence at positions 70–74 in the DRB1 molecule (10, 11). However, comparing HLA–DR in patients with established RA versus controls makes it “impossible to determine empirically whether the affected population has an overrepresentation of a risk allele or an underrepresentation of a ‘protective’ allele” (12). An inception EPA cohort may be better suited for analysis of the impact of protective alleles, first on disease onset and then on disease outcomes.
In the present study, we tried to identify in consecutive EPA patients those who had the lowest and highest risks of reaching a preset severe outcome. To do so, we examined the interaction of DR alleles with erosive disease, controlling for RA-associated antibodies, including anti-Sa, that were present at the first evaluation as predictors of the development of severe disease at the 30-month benchmark. Our data indicate that the presence of erosions at the first evaluation entails the worst prognosis at 30 months. Conversely, among patients who do not have erosions at baseline, the presence of at least 1 DERAA allele identified a subgroup with the lowest propensity to develop severe disease while receiving conventional disease-modifying antirheumatic drug (DMARD) treatment aimed at disease remission.
PATIENTS AND METHODS
This cohort has previously been described in detail (8). Briefly, consecutive adult patients with synovitis affecting at least 3 joints for 1–12 months who had been evaluated at the Centre Hospitalier Universitaire de Sherbrooke (CHUS) were asked to participate in a longitudinal observational study. We excluded patients with bacterial or crystal-induced arthritis, a defined connective tissue disease (which included thorough autoantibody testing ), or systemic vasculitis according to the American College of Rheumatology (ACR; formerly, the American Rheumatism Association) criteria (14). We did not exclude patients who failed to fulfill the ACR criteria for RA (15) or had skin or nail lesions typical of psoriasis, inflammatory bowel disease, clinical features suggestive of a spondylarthropathy (16), signs or symptoms suggestive of a connective tissue disease not fulfilling specific ACR criteria, or positive HLA–B27.
Consenting patients were regularly followed up by rheumatologists and were treated according to the current approach of early and intensive therapy with DMARDs (17, 18). DMARD treatment was individualized, with the objective of sustained remission. Patients, rheumatologists, and coordinating nurses remained blinded to the patients' HLA–DR, anti–cyclic citrullinated peptide 2 (anti–CCP-2), and anti-Sa status. Serum and DNA samples were coded. The Ethics Review Board of the CHUS approved the study (ClinicalTrials.gov ID: NCT00512239).
A rheumatologist completed the physical examinations, including assessments of 68 joints for tenderness and 66 joints for swelling. A trained coordinator performed a structured interview at the inclusion visit and at each of the followup visits, which were scheduled at 18, 30, 42, and 60 months after the onset of inflammatory arthritis, to ensure more homogeneous disease duration upon followup. The time of onset of arthritis was assumed to be the week or month during which joint symptoms/signs had appeared or, in patients with previous musculoskeletal complaints (e.g., osteoarthritis), the time when signs or symptoms of inflammatory arthritis appeared.
Variables assessed included demographic features, tender and swollen joint counts, duration of morning stiffness, medication use at each visit and between visits, the modified Health Assessment Questionnaire (M-HAQ; range 0–3, representing normal function to maximal disability) (19), serum C-reactive protein (CRP) levels (upper limit of normal 8.0 mg/ liter), genomic HLA–DR typing (see below), serum IgM rheumatoid factor (IgM-RF) (using a RapiTex RF kit, with the threshold for a positive result set at 40 IU/ml; Dade-Behring, Newark, DE), measurement of anti-Sa antibodies as described previously (8), and measurement of anti–CCP-2 antibodies (using a QuantaLite kit, with a positive result at >20 units/ml; Inova Diagnostics, San Diego, CA). Radiographs of the hands and feet were obtained at study inclusion and at each scheduled assessment. Joint space narrowing and erosions were scored according to the modified Sharp/van der Heijde (SHS) method, with a maximum score of 448 units (20). All radiographs were read in known time sequence by 1 or 2 trained and blinded reviewers, 1 of whom is an investigator of this study (GB). The smallest detectable difference was shown to be ∼5 units, and it was assumed to be identical to the minimal clinically important difference (21, 22). The Disease Activity Score 28-joint assessment (DAS28) was calculated according to the appropriate formula (formulas available online at www.das-score.nl) using 4 variables (including the CRP) (23).
Severity required an M-HAQ score of at least 1.0 and/or within the upper one-third of the erosion component of the SHS. In the first consecutive 210 patients, the upper tertile started at 6 units at study inclusion, with 10 at 18 months and 14 at 30 months into the disease. Persistent arthritis was defined as the presence of at least 1 joint with synovitis and/or the current use of DMARDs or at least 10 mg of prednisone equivalent per day (7). Erosive disease was defined as a score of at least 5 on the erosion component of the SHS. Remission according to the DAS28 was defined as a DAS28 of <2.6.
Low-resolution and high-resolution genomic polymerase chain reaction (PCR) typing at the HLA–DR locus.
Genomic DNA was extracted from 2.5 ml of EDTA-treated whole blood (Wizard DNA kit; Promega, Madison, WI) and then stored at 4°C. Genomic typing was performed using sequence-specific primer PCR techniques specific for HLA class II molecules (24, 25). Primer sets for low-resolution typing of HLA–DR and HLA–DQ antigens and for high-resolution typing of DRB1*01, DRB1*04, DRB1*11, DRB1*13, and DRB1*14 were obtained from Pel-Freez Clinical Systems (Brown Deer, WI). After amplification, 10 μl of PCR products was resolved in a 2% agarose gel containing ethidium bromide and then visualized by ultraviolet transillumination.
Amplification patterns were interpreted according to manufacturer's instructions. HLA–DRB1 alleles encoding the amino acid sequences QRRAA (DRB1*0101, *0102, *0105, *0404, *0405, *0408, and *1402), QKRAA (DRB1*0401 and *0409), and RRRAA (DRB1*1001) were considered to encode the shared epitope (SE). The putative RA protective epitope DERAA was said to be present when the alleles DR*0103, *0402, *1102, *1103, *1301, *1302 were identified. Low-resolution typing results from 409 consecutive potential organ donors evaluated at the Laboratoire d'Histocompatibilité of the Institut National de Recherche Scientifique, Institut Armand Frappier, were used to approximate the frequency of the DR alleles in the general population of the Province of Quebec. Both the EPA and the organ donor groups were almost exclusively Caucasian.
Student's t-test or, if distribution was not normal, the Mann-Whitney U test was used to evaluate differences in continuous variables. Proportions were compared by chi-square test or Fisher's exact test when appropriate. To define the associations of the different groups of DR alleles (DERAA, SE, and DR3) with severe outcomes at 30 months and with the production of antibodies (anti-Sa, anti–CCP-2, and IgM-RF) at presentation, odds ratios (ORs) were calculated. To evaluate whether the presence of DERAA and of erosions (and in some cases of RA-associated antibodies) at study inclusion influenced the proportion of patients with severe disease, a graph of distribution was included. Logistic regression models were used to estimate the independent contribution of groups of DR alleles, antibodies, and baseline characteristics to predict the development of severe disease at 30 months. Each variable was first compared with severe disease in univariate analyses, and variables with a P value less than 0.1 were included in a complete logistic regression analysis without further selection. We also performed forward and backward stepwise logistic regression. The results of the different logistic regression models were compared. For all analyses, P values less than 0.05 were considered significant.
Characteristics of the patients at study inclusion and at the 30-month followup.
The main demographic, clinical, and biologic characteristics of the 210 patients are presented in Table 1. Since at least 3 joints with objective synovitis were required, >80% of the patients had already fulfilled the ACR criteria for RA at inclusion. Anti-Sa, anti–CCP-2, and IgM-RF antibodies were present at inclusion in 51 patients (24.3%), 78 patients (37.1%), and 91 patients (43.3%), respectively. Disease activity was high initially (mean DAS28 5.28), but had significantly decreased by 30 months (mean DAS28 2.6). This clinical improvement was further illustrated by a decrease in the number of patients with significant functional impairment (M-HAQ score of at least 1.0), from 99 at study entry to 24 at 30 months and by an increase in the number of patients in remission according to the DAS28, from 4 to 123 (58.6%). Despite this improvement in clinical disease activity, radiologic damage progressed significantly. Indeed, the number of patients with erosive disease (i.e., score of ≥5 on the erosion component of the SHS) increased from 41 to 113, and the number with severe disease (score of ≥14 on the erosion component of the SHS) increased from 23 to 64. Thus, despite a rapid diagnosis and the early introduction of treatments that effectively controlled disease activity in the majority of patients, erosive damage occurred in 53.8% of the patients, with severe disease developing in 37.1%.
Table 1. Characteristics of the 210 early polyarthritis patients at study inclusion and at 30 months*
All comparisons between study visits were statistically significant at P < 0.0001. ACR = American College of Rheumatology; RA = rheumatoid arthritis; anti–CCP-2 = anti–cyclic citrullinated peptide 2; IgM-RF = IgM rheumatoid factor; CRP = C-reactive protein; M-HAQ = modified Health Assessment Questionnaire; IQR = interquartile range; DAS28 = Disease Activity Score 28-joint assessment (4 variables); SHS = modified Sharp/van der Heijde score.
No. (%) female
Age, mean ± SD years
57 ± 16
Duration of symptoms, mean ± SD months
4.85 ± 3.2
No. (%) fulfilling ≥4 ACR criteria for RA
Antibodies, no. (%) of patients
Anti–CCP-2 >20 units/ml
IgM-RF ≥40 IU/ml
No. (%) with CRP ≥8.0 mg/liter
Median (IQR) score
No. (%) with M-HAQ score ≥1.0
Mean ± SD score
5.28 ± 1.4
2.6 ± 1.1
No. (%) in remission (DAS28 <2.6)
No. (%) with active disease (DAS28 ≥3.2)
Median (IQR) score
No. (%) with an SHS for erosions ≥5
No. (%) with severe disease
No. (%) with an SHS ≥14
Distribution of HLA–DR alleles among EPA patients.
The distribution of DR alleles defined at low resolution was compared in our cohort of 210 EPA patients and in a group of 409 potential organ donors from the Province of Quebec (see Patients and Methods). The group of organ donors was assumed to be representative of the Caucasian genetic makeup of the local regional population, which is mostly of North European origin. EPA patients more frequently carried the DR4 and DR1 alleles (58.1% in EPA versus 38.4% in organ donors; P < 0.0001). In contrast, there was a decreased prevalence of DR7 (14.3% versus 24.0%; P = 0.0002) and, to a lesser degree, DR8 (2.9% versus 7.6%; P = 0.01895), as well as the DR11 and DR13 alleles (38.6% versus 47.9%; P = 0.02675). This suggested that the clinical development of EPA was facilitated by the presence of DR alleles containing most of the SE alleles, but may be slightly decreased in the presence of DR alleles containing most of the DERAA alleles. Calculations using DR alleles defined at low resolution represent only a crude approximation, however, since some DR1 alleles are not SE (e.g., the DR*0103 allele is in fact a DERAA allele) and some DR11 and DR13 alleles are not DERAA (e.g., DR*1101 and *1303). The role of DR7 and DR8 in the development of EPA merits further study, particularly since a similar negative association with these alleles was previously reported in Japanese EPA patients (26).
Table 2 shows the global distribution of DR alleles genotyped in our EPA cohort. The DERAA alleles were found in 62 patients, including 20 who were heterozygous for the DERAA and the SE alleles and 6 who carried 2 DERAA alleles. The most frequent DERAA alleles in our cohort were DRB1*1301 (18 patients), DRB1*1302 (16 patients), and DRB1*0103 (16 patients). SE alleles were found in 110 patients, including 24 carrying 2 SE alleles. The most frequent SE alleles in our cohort were DRB1*0101 (45 patients), *0401 (40 patients), and *0404 (27 patients). In our cohort, 58 patients carried no SE or DERAA allele, including 29 patients carrying at least 1 DR3 allele.
Table 2. Distribution of HLA–DRB1 alleles in the 210 early polyarthritis patients*
No. (%) of patients
DERAA is the rheumatoid arthritis “protective” epitope allele. SE = shared epitope.
At least 1 SE allele
2 SE alleles
1 DR3 allele
At least 1 DERAA allele
2 DERAA alleles
1 DR3 allele
Total with SE and/or DERAA alleles
No SE and no DERAA alleles
At least 1 DR3 allele
DERAA alleles and protection against the development of severe disease.
Over the first 30 months of disease, 113 patients (53.8%) developed erosive disease according to our preset criteria (i.e., score of ≥5 on the erosion component of the SHS) (Table 1). Of these, 41 already had erosions at the study inclusion visit. The presence of at least 1 DERAA allele had a significant protective effect on the development of erosive disease (OR 0.51, P < 0.05), while no similar protection was found with the SE allele (Table 3). The presence of a DERAA allele was also associated with a decrease in the risk of developing severe disease at 30 months (OR 0.30, P < 0.001). This contrasted with the absence of association between SE alleles and severe disease. Similarly, the presence of a DR3 allele was not associated with erosive disease or with severe disease at 30 months.
Table 3. HLA–DR alleles and development of erosive disease or severe disease at the 30-month evaluation*
No. with allele(s)
Erosive disease at 30 months
Severe disease at 30 months
No. of patients
OR (95% CI)
No. of patients
OR (95% CI)
Erosive disease was defined as the presence of ≥5 erosions. DERAA is the rheumatoid arthritis “protective” epitope allele. OR = odds ratio (relative to −/− genotype); 95% CI = 95% confidence interval; SE = shared epitope.
Surprisingly, the apparent protection conferred by the DERAA alleles against severe disease was not associated with clinically milder disease at presentation. For example, the mean number of joints with synovitis, the mean DAS28 scores, the mean CRP levels, as well as other markers of inflammation were not significantly different in DERAA and non-DERAA patients (P > 0.3) (data not shown). Similarly, DERAA-associated protection was not accompanied by spontaneous remission maintained without DMARDs. Indeed, ∼76% of the DERAA patients had persistent disease at 30 months, as compared with 83% of the non-DERAA patients. In addition, the intensity of treatment, as reflected by the number of DMARDs used, was similar at 30 months in DERAA and non-DERAA patients (data not shown).
DERAA alleles and RA-associated antibodies.
The presence of IgM-RF and anti–CCP-2 antibodies in RA patients has been associated with more erosive and more severe disease. Conversely, DERAA alleles might identify a subset of patients with milder seronegative EPA. To verify this possibility, we examined the associations between DR alleles and RA-associated antibodies (Table 4). No significant association of DERAA with the production of any of the 3 RA-associated antibodies was observed, since the prevalence of RA-associated antibodies was similar in patients with and without DERAA. However, the known DR associations with RA-associated antibodies were present in our cohort (27). First, carriage of 2 SE alleles was strongly and positively associated with the production of antibodies, with an OR of 5.97 for anti-Sa, 9.50 for anti–CCP-2, and 7.38 for IgM-RF antibodies (Table 4). The association of a single dose of SE with IgM-RF and anti–CCP-2 antibodies was weaker, but remained significant. Second, the production of anti–CCP-2 was negatively associated with carriage of DR3 alleles (OR 0.38, P < 0.001).
Table 4. HLA–DR alleles and autoantibodies at study inclusion in the 210 patients*
No. with allele(s)
No. of patients
OR (95% CI)
No. of patients
OR (95% CI)
No. of patients
OR (95% CI)
DERAA is the rheumatoid arthritis “protective” epitope allele. Anti–CCP-2 = anti–cyclic citrullinated peptide 2; IgM-RF = IgM rheumatoid factor; OR = odds ratio (relative to −/− genotype); 95% CI = 95% confidence interval; SE = shared epitope.
Quite interestingly, in contrast to the DR3 associations with anti–CCP-2, there was no significant association of single-dose SE alleles with the production of anti-Sa antibodies (Table 4). Similarly, a negative association between anti-Sa and DR3 was not observed. This is the first indication that antibodies to citrullinated proteins such as anti-Sa/citrullinated vimentin might have DR associations distinct from those of anti–CCP-2 antibodies.
Independent prognostic information in EPA through genotyping of DERAA alleles.
A multivariate logistic regression analysis was performed to assess whether the presence of DERAA alleles might contribute independent prognostic information at the first evaluation of the study patients (Table 5). In this model, erosive disease, the presence of RA-associated antibodies (especially anti-Sa), and age ≥50 years (8) contributed significantly to the prediction of severe disease, with an OR of 6.56, 3.57 and 2.85, respectively. Conversely, the presence of a DERAA allele contributed significantly to the prevention of severe disease (OR 0.21, P < 0.01). The magnitude of the protective effect of DERAA alleles against severe disease was thus similar to the detrimental effect of already erosive disease on initial radiographs. Similar odds ratios were obtained with and without inclusion of SE in the model using stepwise and complete multivariate analyses. This confirmed the lack of additional information carried by the SE status as observed in univariate analyses.
Table 5. Multivariate logistic regression model for severe disease at 30 months*
No. of patients
No. with severe disease
OR (95% CI)
DERAA is the rheumatoid arthritis “protective” epitope allele. OR = odds ratio (relative to −/− genotype); 95% CI = 95% confidence interval; SE = shared epitope; IgM-RF = IgM rheumatoid factor; anti–CCP-2 = anti–cyclic citrullinated peptide 2; SHS = modified Sharp/van der Heijde score.
Protective effect of DERAA alleles only in patients without erosions at study inclusion.
In the above multivariate logistic regression model, the presence of erosive disease at study inclusion was a strong marker of future severe disease. Indeed, almost three-fourths of patients who already had erosive disease at study inclusion (30 of 41 patients [73.2%]) developed severe disease over the following 30 months (Figure 1). Patients with erosive disease at baseline represented a subset of ∼20% of our cohort in which the presence of a DERAA allele did not have a protective effect. In contrast, in the remaining 80% of patients who did not have erosions at study inclusion, the presence of a DERAA allele was associated with a much lower risk of developing severe disease (8.0% versus 36.1%; P = 0.0001).
Despite their significantly positive association (OR 3.57) with severe disease in the multivariate analyses (Table 5), anti-Sa antibodies failed to refine the predictive model in patients without erosions at study inclusion. This may represent a type II error, since our cohort contained only 8 patients without erosions at baseline who were anti-Sa positive and carried a DERAA allele, of which 1 patient had a severe outcome at 30 months. Our growing cohort of patients may enable us to answer that specific question in the near future.
We have observed that, in patients with early polyarthritis, carriage of at least 1 allele of the DRB1 molecules containing the DERAA sequence at positions 70–74 was associated with a decreased risk of developing severe disease by 30 months. Our results are thus concordant with previous reports of the protective associations of the HLA–DRB1 DERAA alleles in established cohorts of RA patients (10, 28) and in early RA (29). A protective role of DERAA alleles in EPA was also previously suggested in an underpowered study (30).
The design and nature of our cohort allowed us to analyze in more detail the context of the protection afforded by the presence of a DERAA allele. First, the protection was statistically valid only in patients who did not have erosions at the first evaluation. The failure to observe such a protective effect of DERAA alleles in the BeSt (Behandelstrategieën voor Reumatoide Artritis) trial (31) can thus be partly explained by the almost uniformly erosive status of the patients selected for the trial. Second, the protective influence of a DERAA allele was not associated with an absence of production of RA-associated antibodies. This suggests that the protection afforded by DERAA does not simply counteract the mechanisms used by SE alleles, which appear to associate with severity through the production of antibodies to citrullinated antigens (32, 33). A similar lack of protection of DERAA alleles against RA-associated antibodies in established RA was recently reported (34). Third, the protective influence of DERAA alleles did not imply clinically milder disease at study inclusion or spontaneous remission over time. Indeed, the clinical disease activity at inclusion, the proportion of persistent disease, and the types and numbers of DMARDs still being prescribed at 30 months were similar in the DERAA and non-DERAA groups (data not shown). As a consequence, genotyping at the DR locus for the presence of a DERAA may represent a useful addition to the initial evaluation and treatment of EPA patients.
Since the presence of erosive disease at the first evaluation overrides the contribution of the protective DERAA alleles, this suggests that other genetic and/or environmental influences determine an as-yet-undefined “erosive factor.” This putative “erosive factor” was present both in seronegative and in seropositive EPA patients (Table 2). It may thus result from the action of nonimmune mediators (e.g., osteoclastogenesis) that are unlikely to be regulated by DR alleles. Alternatively, once erosions have occurred, the pathologic process may no longer be accessible to regulation by the DERAA alleles. The classic SE at the DR locus had no influence on the development of erosive disease or on the progression to severe disease. Previous studies of SE alleles in EPA patients have reached similar conclusions about the SE, possibly because current treatments alter its influence on severity (31–33, 35, 36). Analysis of the characteristics of patients with early aggressive erosive disease deserves more focused research since close to 75% of such patients reached the severe disease threshold by 30 months despite aggressive treatment with nonbiologic DMARDs. We postulate that this group of initial “fast eroders” represents a large proportion of the EPA patients who could benefit from the early addition of biologic agents.
In our cohort, we could confirm known associations between RA-associated antibodies and SE DR alleles (27, 33, 37–39). Anti–CCP-2 antibodies were strongly and positively associated with a double dose of SE, and strongly and negatively associated with the presence of DR3. Despite these strong associations, it is noteworthy that 24 of the 78 patients (30.8%) patients producing anti–CCP-2 antibodies at study inclusion carried no SE allele. We could not explore in this study the proposed mother–fetus transmission of SE-positive cells as a contributor to the development of seropositive RA (40, 41). However, we could, for the first time, examine the DR associations of anti-Sa antibodies. Unlike anti–CCP-2, anti-Sa antibodies were not associated with a single dose of the SE and were not negatively associated with DR3. These data suggest diverging immunogenetics underlying the production of anti-Sa and anti–CCP-2 antibodies, two antibodies that target citrullinated antigens.
Among patients without erosions at study inclusion, the presence of anti-Sa antibodies was associated with an increased risk of severe disease. Anti-Sa, but not anti–CCP-2, was an independent marker of severity in our multivariate analyses. Similarly, a commercial assay using mutated citrullinated vimentin was better than an anti–CCP-2 assay in predicting a poor radiologic outcome in EPA (42). This suggests that antibodies to citrullinated peptides (e.g., CCP-2) and antibodies to citrullinated proteins may not be prognostically equivalent. Although a trend was observed, insufficient power prevented us from confirming whether the presence of a DERAA allele can antagonize the increased risk for severe disease associated with anti-Sa antibodies.
Our cohort has some unique characteristics worth stressing. First, patients were seen in a single referral center early after the onset of signs or symptoms of disease (mean of 4.8 months) and were thus sick enough to be fast tracked to rheumatologists. Second, followup visits were scheduled at fixed intervals from clinical disease onset. Third, patients were usually treated within weeks with conventional DMARD combinations, which usually included methotrexate. Although there was no predetermined drug regimen, our implicit objective was to obtain and maintain clinical and biologic remission. This goal was attained in 123 patients (59%) at 30 months into disease. Fourth, only patients in whom at least 3 joints had objective signs of synovitis were included. Fifth, patients were almost exclusively Caucasian.
A brief discussion of “severe disease” as the primary outcome used in our study is also warranted. We adopted a priori (8) a composite definition for severe disease that could be used in EPA to identify early the patients most affected by arthritis. This definition included structural and functional components of joint damage, as scored by the SHS and the M-HAQ, respectively. Initially, we defined structural severity as being the upper tertile of the erosion component of the SHS at any given point. In the first consecutive 210 patients, the upper tertile started at 6 units at study inclusion, with 10 units at 18 months and 14 at 30 months into the disease. The latter value of 14 units (5% of the maximal erosion component of the SHS) was thus used to indicate the presence of severe disease at 30 months. The functional component used a threshold of 1.0 for the M-HAQ as an indicator of severe disease. It follows that at study inclusion, most patients designated as having severe disease were so labeled because of severe functional limitations. In contrast, at 30 months into disease, only 14 of the 78 patients with severe disease were so labeled because of their M-HAQ score. Of the remaining 64 patients with severe disease at 30 months, 10 also had an M-HAQ score of at least 1.0 (Table 1).
In conclusion, caring for patients with EPA ideally requires the early identification of those with the worse and those with the best expected outcomes. Our data confirm that defining the erosion status is the most important initial step in the evaluation of an EPA patient because it identifies those with the worse prognosis. Patients who already have erosions should benefit early from the maximally effective available treatments, since rapid progression of disease occurs in more than 70% of EPA patients despite aggressive treatment with nonbiologic DMARDs. In patients without erosions, genotyping at the HLA–DR locus for the presence of a DERAA allele may identify those who are most likely to respond best to conventional DMARDs and, thus, most likely to have the best outcomes when adequately treated. Finally, measuring antibodies to citrullinated proteins (e.g., anti-Sa), rather than to CCP-2 peptides, may further improve the prognostic classification of EPA patients who do not have erosions at the first evaluation. This approach still leaves a large group of patients without erosions at study inclusion in whom reliable biomarkers remain to be identified.
Dr. Boire had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study design. Cossette, Ménard, Boire.
Acquisition of data. Daniel, de Brum-Fernandes, Liang, Ménard, Boire.
Analysis and interpretation of data. Carrier, Cossette, Daniel, de Brum-Fernandes, Ménard, Boire.
Manuscript preparation. Carrier, Cossette, Daniel, de Brum-Fernandes, Liang, Ménard, Boire.