Article Text

Download PDFPDF

Concise report
Rheumatoid arthritis phenotype at presentation differs depending on the number of autoantibodies present
  1. V F A M Derksen1,
  2. S Ajeganova1,2,
  3. L A Trouw1,
  4. A H M van der Helm-van Mil1,
  5. I Hafström2,
  6. T W J Huizinga1,
  7. R E M Toes1,
  8. B Svensson3,
  9. D van der Woude1
  1. 1Department of Rheumatology, Leiden University Medical Center, Leiden, The Netherlands
  2. 2Rheumatology Unit, Department of Medicine, Karolinska Institutet, Karolinska University Hospital Huddinge, Stockholm, Sweden
  3. 3Section of Rheumatology, Department of Clinical Sciences, Lund University, Lund, Sweden
  1. Correspondence to Dr Diane van der Woude, Department of Rheumatology, Leiden University Medical Center, C1-R-041, Albinusdreef 2, PO Box 9600, Leiden 2300 RC, The Netherlands; dvanderwoude{at}


Objectives In rheumatoid arthritis (RA), seropositive and seronegative disease may be two entities with different underlying pathophysiological mechanisms, long-term outcomes and disease presentations. However, the effect of the conjoint presence of multiple autoantibodies, as proxy for a more pronounced humoral autoimmune response, on clinical phenotype remains unclear. Therefore, this study investigates the association between the number of autoantibodies and initial clinical presentation in two independent cohorts of patients with early RA.

Methods Autoantibody status (rheumatoid factor, anticitrullinated protein antibodies and anticarbamylated protein antibodies) was determined at baseline in the Leiden Early Arthritis Cohort (n=828) and the Swedish BARFOT (Better Anti-Rheumatic Farmaco-Therapy, n=802) study. The association between the number of autoantibodies and baseline clinical characteristics was investigated using univariable and multivariable ordinal regression.

Results In both cohorts, the following independent associations were found in multivariable analysis: patients with a higher number of RA-associated antibodies were younger, more often smokers, had a longer symptom duration and a higher erythrocyte sedimentation rate at presentation compared with patients with few autoantibodies.

Conclusions The number of autoantibodies, reflecting the breadth of the humoral autoimmune response, is associated with the clinical presentation of RA. Predisease pathophysiology is thus reflected by the initial clinical phenotype.

  • Rheumatoid Arthritis
  • Autoantibodies
  • Autoimmunity
View Full Text

Statistics from


Approximately 60% of patients with early rheumatoid arthritis (RA) are positive for RA-associated autoantibodies such as rheumatoid factor (RF), anticitrullinated protein antibodies (ACPA) and anticarbamylated protein antibodies (anti-CarP). Anti-CarP is a novel autoantibody system in RA and is associated with the development of RA in patients with arthralgia and radiographic damage.1 ,2 Autoantibodies may develop years before disease onset and are commonly seen as markers for underlying autoimmune pathophysiology. Thus, clinical phenotype at presentation may be a reflection of pathophysiological mechanisms taking place before disease onset.

Seropositive and seronegative RA differs in terms of risk factors and disease course. Seropositivity, especially ACPA positivity, is associated with risk factors such as smoking and human leucocyte antigen (HLA) shared epitope (SE) alleles.3 Seropositive patients also have a worse disease outcome with more radiological damage over time.4 In contrast, seronegative patients have recently been described to have more joint inflammation at initial presentation.5 Even though novel autoantibodies may be discovered in the future, seropositive and seronegative RA as defined at the moment is likely to differ in underlying pathophysiology and phenotype.5 ,6

However, within the seropositive subset, a varying number of autoantibodies can be found, with the presence of several autoantibodies indicating a break of tolerance to more autoantigens. Yet, the specific clinical implications of a broader humoral autoimmune response remain unclear, although this could be very important since broadening of the autoantibody response could be amenable to therapeutic intervention. Therefore, this study investigates the association between the number of autoantibodies, as proxy for a broader humoral autoimmune response, and initial clinical presentation in patients with RA.

Patients and methods


Two independent early RA cohorts from the Netherlands and Sweden were analysed: the Leiden Early Arthritis Clinic (EAC) and the Better Anti-Rheumatic Pharmacotherapy (BARFOT) project (details described elsewhere).7 ,8 All patients had a short symptom duration (<24 months EAC and <12 months BARFOT), were disease-modifying anti-rheumatic drug (DMARD)-naive and had known status for all three autoantibodies (RF, ACPA and anti-CarP). All patients fulfilled the 1987 ACR criteria for RA (in EAC within 1 year of follow-up and in BARFOT at inclusion). Informed consent was obtained and the studies were approved by the local medical ethics committees. At baseline, information about demographics, smoking, family history and disease characteristics was recorded. The location of initial symptoms was documented only in EAC.

Autoantibody measurements

All antibody measurements were performed in baseline sera. IgM-RF was measured by commercial ELISA in EAC and agglutination test (SERODIA-RA) in BARFOT. In both cohorts, ACPA were determined by anti-second-generation cyclic citrullinated peptide (CCP2) ELISA (Euro-Diagnostica) with manufacturer's cut-offs. Antibodies against carbamylated fetal calf serum were measured using in-house ELISAs in Leiden for both cohorts as described previously.2 The cut-off was set at the mean plus two times the SD of the specific anti-CarP antibody reactivity of healthy controls matched for country of origin (the Netherlands and Sweden). The presence of ACPA isotypes, ACPA fine specificities and SE alleles was determined in EAC for a subset of patients as described previously.9 ,10

Statistical analysis

Multiple imputations were used in both cohorts to deal with random missing data. Markov chain Monte Carlo imputations (m=20) using linear and logistic regression for continuous and categorical variables, respectively, were done in SPSS V.23. Ordinal regression was performed to compare baseline characteristics between patients harbouring 0–3 autoantibodies, and results of each imputed data set were pooled to yield ORs and 95% CI, which signify the increase in association with every additional autoantibody. The Holm-Bonferroni method was applied to correct for multiple testing.11 Variables with a univariable p value <0.10 were included in a multivariable model, excluding highly correlated variables. Associations between SE positivity and baseline characteristics were analysed using Mann-Whitney U tests or χ2 tests.


In both cohorts, the distribution of the number of autoantibodies was similar. The majority of patients was either seronegative (31% EAC and 33% BARFOT) or triple positive for RF, anti-CCP2 and anti-CarP (35% EAC and 29% BARFOT). Baseline characteristics differed between the two cohorts at several points (see online supplementary table S1), reflecting differences in inclusion criteria and referral systems.

Several phenotypic characteristics were significantly associated with the number of autoantibodies in both EAC and BARFOT. In univariable and multivariable analyses, the following independent associations were found in both cohorts (tables 1 and 2): patients with additional RA-associated antibodies were younger, more often smokers, had longer symptom duration and higher erythrocyte sedimentation rate (ESR) at presentation (figure 1). Furthermore, some associations were found in one cohort but could not be replicated, like body mass index (BMI), swollen joint count (SJC)28 and tender joint count (TJC)28.

Table 1

EAC: number of autoantibodies and clinical characteristics at presentation

Table 2

BARFOT: number of autoantibodies and clinical characteristics at presentation

Figure 1

Means and SDs or percentages of clinical characteristics, which were significant in multivariable analysis in both cohorts, for increasing number of autoantibodies. EAC, Early Arthritis Clinic; BARFOT, Better Anti-Rheumatic Farmaco-Therapy.

To investigate whether underlying genetic risk factors could partly explain our findings, we analysed the association between SE alleles (presence vs absence) and initial clinical presentation in EAC. SE-positive patients indeed had a significant longer symptom duration and a trend towards being younger, more often smokers and having a higher ESR, but associations were less strong than with the number of autoantibodies.

Thereafter, it was investigated whether the increasing prevalence of one autoantibody in particular among patients with an increasing number of autoantibodies might be responsible for the observed observations. Stratifications for all three autoantibodies were performed, comparing single-positive patients with patients harbouring one or two additional autoantibodies (data for anti-CCP2 shown in online supplementary tables S2 and S3). This resulted in small groups with limited power, but for most variables there was still a trend visible for increasing numbers of autoantibodies. Overall, the effect of the higher number of autoantibodies seemed stronger than the effect of individual autoantibodies.

Besides the number of autoantibodies, other features of a broad autoantibody response are increased levels, ACPA fine specificity and isotype usage. In EAC, there was an association between clinical phenotype and these features (see online supplementary tables S4–S11), although not as significant and consistent as with the number of autoantibodies. This could be due to inclusion of only ACPA-positive individuals in these subanalyses.


This study describes for the first time the association between initial clinical presentation and the number of autoantibodies. We found that younger age, smoking, longer symptom duration and higher ESR were independently associated with having additional autoantibodies. We replicated these results in a second independent cohort.

Patients with a higher number of autoantibodies also have higher autoantibody levels, more fine specificities and broadened isotype usage. These features also appeared to be associated with clinical phenotype. Since all reflect the breadth of underlying autoimmune pathophysiology, this suggests an association between initial clinical presentation and the extent of humoral autoimmunity.

The underlying genetic predisposition associated with autoantibody positivity could possibly explain part of the observations, for example, why patients with multiple autoantibodies develop RA at a younger age.12 This idea is supported by the association between clinical presentation and HLA SE alleles, representing part of the genetic predisposition. The association between the presence of additional autoantibodies and longer symptom duration is in line with a previous publication.13 It was recently reported that seronegative individuals have higher joint counts at presentation.5 We found in addition that increasing numbers of autoantibodies are associated with lower tender joint counts (and swollen joint counts in EAC). This may possibly reflect the higher number of joints required for seronegative individuals to fulfil classification criteria as well as intrinsic pathophysiologic differences. Overall, the findings suggest that the genetic predisposition underlying autoantibody-positive RA may result in an early, smouldering disease onset, which is difficult to recognise promptly by patients and their general practitioners.

Our study has several limitations. Some variables had a large amount of missing data. However, since these data were generally missing at random and multiple imputations yielded very similar results, it seems unlikely that missing data biased our conclusions. Although this study focuses on the simultaneous presence of autoantibodies, stratifications for the different autoantibodies were performed to determine whether the observed effects are due to the presence of a single autoantibody. This analysis lacks power to draw solid conclusions, but there were still trends visible. Although we cannot exclude certain effects of specific autoantibodies, overall the effect of a higher number of autoantibodies seemed stronger than the effect of individual autoantibodies. Another limitation is the dissimilarity in baseline characteristics between both cohorts, which might be due to difference in inclusion criteria or referral systems between the Netherlands and Sweden.7 ,8

Nevertheless, the use of two large, independent cohorts enabled us to distinguish between patients with 1 versus 2 versus 3 autoantibodies and thereby discover new dose-dependent associations consistent across different populations. To further enhance our understanding of the link between humoral autoimmunity and clinical phenotype, more studies are warranted investigating disease evolution, possibly by studying immunological processes in patients with arthralgia progressing towards RA. Our study also has implications for the early recognition and treatment of RA. The fact that patients with additional autoantibodies had a longer symptom duration, while they are particularly predisposed to a more severe disease course, might indicate the need for better targeted early recognition strategies to reach these patients.

In conclusion, the number of autoantibodies present is associated with clinical phenotype at presentation, indicating that the breadth of the humoral autoimmune response affects the initial clinical presentation of RA.


View Abstract


  • Handling editor Tore K Kvien

  • Contributors All authors were involved in study design, data collection and interpretation and approved the final version of the manuscript. VFAMD and DvdW performed the statistical analyses and wrote the manuscript.

  • Funding This study was supported by the European Union Seventh Framework Programme integrated project Euro-TEAM.

  • Competing interests None declared.

  • Patient consent Obtained.

  • Ethics approval Ethics Committee of the Leiden University Medical Center; regional ethics committee in Stockholm, Sweden.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.