Article Text

Download PDFPDF

Extended report
Rheumatoid factor determines structural progression of rheumatoid arthritis dependent and independent of disease activity
  1. Daniel Aletaha1,
  2. Farideh Alasti1,
  3. Josef S Smolen1,2
  1. 1Division of Rheumatology, Department of Medicine 3, Medical University Vienna, Vienna, Austria
  2. 2Second Department of Medicine, Hietzing Hospital, Vienna, Austria
  1. Correspondence to Professor Daniel Aletaha, Division of Rheumatology, Department of Medicine 3, Medical University Vienna, Vienna 1090, Austria; daniel.aletaha{at}


Background Rheumatoid factor (RF) is prototypic for rheumatoid arthritis (RA) and serves diagnostic and prognostic purposes. RF is associated with joint destruction, but the role of disease activity as a potential mediator of these effects has not been clearly elucidated yet.

Objective To investigate if higher radiographic progression (Sharp score, ΔTSS) in RF+ patients is dependent or independent of disease activity.

Methods The authors performed a cross-sectional multivariate analysis at baseline and a matched cohort study in patients from five RA clinical trials. The authors pooled methotrexate treatment arms and compared ΔTSS in RF+ and RF− patients before and after matching for other associated variables.

Results Among 686 patients, 124 were RF− and 562 RF+, 343 having high (>160 U/ml) RF. ΔTSS was 1.03±5.83, 3.23±8.10 and 3.58±8.18 (p<0.0001), respectively, and similarly for erosions and joint space narrowing (JSN). After matching for other prognostically important variables, ΔTSS still was lower among 61 RF− versus 61 RF high+ patients (0.52±2.47 vs 3.09±8.28; p=0.028), mainly related to differences in erosion score (0.31±1.88 vs 2.07±5.62; p=0.035), but not JSN (0.21±1.26 vs 1.02±3.31; p=0.162).

Conclusions The data reveal that damage progression in seropositive RA patients is related to higher levels of disease activity and to independent effects of RF, particularly on bone damage. This calls for consideration of RF status irrespective of disease activity.

Statistics from

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.

One of the hallmarks of rheumatoid arthritis (RA) is the presence of autoantibodies. Among them, rheumatoid factor (RF) was the first autoantibody ever described in a chronic inflammatory condition,1 and is regarded as prototypic of RA. Moreover, it has diagnostic2–5 and prognostic implications. Although the risk of progression in RF+ patients is very evident from the literature, the mechanisms driving the relationship between RF and joint damage have not been fully clarified. Studies of the association between the presence and/or titre of RF and disease activity have provided equivocal results, with some studies reporting no relationship,6–9 and others suggesting independent influences of RF and disease activity on joint damage.10–12 Most of these analyses stemmed from observational data of patients with various degrees of disease activity and heterogeneous therapies.

The purpose of this study is to determine if the adverse prognostic implications of RF is mediated through its ability to increase disease activity, and only secondary through subsequent joint destruction, or whether the presence of RF per se leads to joint destruction that is independent of disease activity or both using a population of patients from pooled clinical trials with high disease activity upon entry and homogeneous methotrexate (MTX) therapy.

Patients and methods

Data sources

We evaluated patient level data of studies on RA patients who had mostly not experienced MTX therapy prior to their enrolment, as kindly provided by the sponsors of these trials.13–16 We were provided an 80%–90% random sample of subject data from the full trial datasets and all patients had active RA at entry into these studies. In the present investigation, we pooled patients who received MTX monotherapy for our analyses.


We obtained measures of RA disease activity including swollen joint counts and tender joint counts (SJC, TJC, on a 28-joint scale), the patient global assessment and evaluator global assessment of disease activity (PGA, EGA, based on a visual analogue scale from 0 to 10 cm), the C reactive protein (CRP in mg/dl) and erythrocyte sedimentation rate (ESR, in mm/h), pain scores (10 cm visual analogue scale) and functional scores (Health Assessment Questionnaire, HAQ) at baseline and follow-up at 3, 6 and 12 months. Further, the Simplified and Clinical Disease Activity Indices and the Disease Activity Score 28 were calculated.17–19

Radiographic data using Sharp scores or van der Heijde modified Sharp scores (TSS) were available in all trials. Radiographic changes were calculated by subtracting the baseline score from the score at 1 year, and radiographic progression was defined as changes ≥0.5.20 In addition, we defined rapid radiographic progression as an increase of ≥5 units/year.21

Main analysis

We compared radiographic progression in RF+ and RF− patients. To allow assessing RF effects that are independent of other important determinants of progression, we first performed a multivariate logistic regression model to identify independent predictors of radiographic progression. Demographic, clinical and functional variables were used as predictors; RF was not included in this analysis.

The predictors significantly associated with progression (p<0.05) were then used to match RF− and RF+ patients. To achieve a good contrast in RF levels of the matched pairs, we compared RF− patients (ie, RF<20 U/ml) with RF ‘high+’ patients (RF >median of the positives, ie, in our cohort, RF>160 U/ml). We used the Mahalanobis distance for matching algorithm.22 The matching allowed us to compare the actually observed progression of RF− and RF high+ patients who had identical disease activity, whereas in a multivariate model progression rates would only be statistically adjusted and then estimated.

The difference in the radiographic progression in the matched RF−/RF high+ pairs of patients was statistically tested using the non-parametric Wilcoxon test and graphically depicted using probability plots.

Additional analyses

We next calculated the area under the curve from baseline, 3, and 6 months measurements of the matching variables to achieve an even more robust match. The area under the curve (not the baseline) values were then used for matching the groups.

We then investigated if the suspected RF effect on structural progression was also dependent on RF levels. Therefore, we matched and compared three groups (RF−, RF low+, RF high+) to see whether the radiographic progression would increase across these groups. For each RF− patient, the matching algorithm identified the best low positive match and the best high+ match. These multiple sets of three patients each were then analysed as matched triplets. Since quality of the matching algorithm was the priority, the number of well-matched triplets was expected to be low. We explored a trend for increasing progression with higher levels of RF.

To evaluate if the effects of RF on structural progression were related to differences in progression of joint space narrowing (JSN) or erosions, we repeated the main analysis with identification of JSN and erosion scores instead of TSS.

We then looked among the subset of patients, for whom 2-year radiographic data were available, whether these effects of baseline RF would also be apparent if a putative carry-over effect were considered.23

We finally performed similar analyses in the pooled population of patients treated with a combination of MTX plus tumour necrosis factor (TNF) inhibitor, in which we a priori did not expect major differences in radiographic progressions given the profound effects of this treatment on destructive processes.



Patient demographics and disease activity characteristics at baseline are shown in table 1, separate for the RF− and the low and high RF+ patients. Evidently, at baseline RF high+ compared with RF− patients have higher TSS and erosion (p=0.0058 and p<0.001, respectively), but not JSN scores (p=0.21). Moreover, ESR and CRP, inflammatory surrogates of RA, are significantly higher in the RF+ population, and so are Disease Activity Score 28 and Simplified Disease Activity Index, which comprise acute phase reactants, as well as the HAQ, a purely patient reported outcome (table 1). When progression of joint damage was assessed in the total population, TSS as well as erosion and JSN subscores were significantly lower in the RF− than RF+ populations (table 2).

Table 1

Patient characteristics at baseline for all patients, by rheumatoid factor (RF) status and for the matched subset of RF negative/positive patients (means±SD, unless otherwise indicated)

Table 2

Radiographic progression in rheumatoid factor (RF) positive and negative patients before and after matching

Main analysis

We had 675 cases with complete data for all the variables needed for multiple regression analysis, of which 350 (51.9%) showed the outcome (ie, radiographic progression ≥0.5 over 1 year). A significant association with the outcome was seen for duration of RA (p<0.0001), baseline radiographic damage (p=0.0371), baseline CRP (p=0.0166) and baseline ESR (p=0.0004). All other variables were not associated with radiographic progression by multiple regression (p≥0.05).

Figure 1A depicts the crude radiographic progression in RF− patients (RF<20 U/ml; n=124) as opposed to RF high+ patients (RF>160 U/ml; n=343). When patients were matched for duration, baseline damage, baseline CRP and baseline ESR, 61 pairs were obtained, who were not different for any of the baseline characteristics anymore (table 1). The radiographic progression was still significantly greater in the RF high+ compared with the RF− patients (mean of 0.52 units/year vs 3.09 units/year; p=0.028; figure 1B, table 2). In the matched pairs, there was also a different proportion of progressors (ie, with a TSS increase ≥0.5 units/year: 27.9% vs 44.3%; p=0.059 by χ2 test; table 2) and of rapid progressors (ie, with a TSS increase of ≥5 units/year: 6.6% vs 19.7%; p=0.03). Similar results were obtained when we matched by the area under the CRP or ESR curves instead of their baseline values (progression of 1.00±5.78 units/year in the RF− and of 3.57±8.14 units/year in the high− RF+ group; p<0.0001).

Figure 1

Radiographic progression in rheumatoid factor negative (<20 U/ml) and positive rheumatoid before (A) and after (B) matching. This figure is only reproduced in colour in the online version.

RF level-effect response

Matching three groups by RF (RF−, n=124; RF low+, n=219; RF high+, n=343), we found 29 matched triplets. Matching was successful and no single variable at baseline differed across these groups statistically (p>0.05, Kruskal–Wallis test), without a notable trend across the three groups for any variable. Despite this similarity in baseline characteristics and the low number of triplets, there was a clear, although statistically non-significant, trend towards a larger effect with higher RF titres, progression in RF− being 0.19±2.47, in RF low+ 1.15±3.72 and in RF high+ 3.77±10.8 (p=0.19, Kruskal–Wallis test).

Effects on JSN and erosions

Figure 2A,B shows the differences for RF− and RF high+ patients in progression of JSN and erosions. Only progression of erosions (0.31±1.88 to 2.07±5.62, p=0.035) but not JSN (0.21±1.26 to 1.02±3.31, p=0.16) was significant.

Figure 2

Progression of joint space narrowing (A) and erosions (B) in rheumatoid factor positive and negative patients after matching. This figure is only reproduced in colour in the online version.

Also in the level-response analysis described above, only erosions showed a respective trend across the three groups (−0.09±2.47, 1.06±3.65 and 2.66±7.49, respectively), and JSN did not (0.29±1.71, 0.09±0.32 and 1.11±3.53, respectively).

Consideration of a carry-over effect

When 25 matched pairs of RF− and RF high+ patients who had second year radiographic data available (subset of patients enrolled in the PREMIER, TEMPO and ERA trials) were analysed, we again found numerically less progression in the RF− individuals between months 12 and 24 for TSS (0.00±1.72 vs 0.78±2.74, respectively), as well as for erosions (0.02±1.41 vs 0.50±2.11), and lesser differences (−0.02±0.46 vs 0.28±1.04) for JSN scores; however, the overall progression in the RF+ population was much higher during year 1 (table 2).

Results in patients treated with TNF inhibitor–MTX combination

In the pooled combination groups of TNF inhibitor plus MTX, we observed a crude progression among the RF− and RF+ patients of 0.24±3.54 and 0.42±4.71 units/year, respectively (p=0.79). After matching as in the main analysis (n=76 pairs), progression in the RF− was 0.51±4.02 units/year versus 0.71±3.19 units/year in the RF high+ group (ie, RF>177 U/ml) (p=0.90).


For many decades, RF has been regarded as a risk factor for more aggressive RA and thus rapidly progressive destructive disease.10 ,24 Whether this effect was related to an increase of disease activity or independent of the inflammatory response has not been unequivocally answered in the literature.6–12 In contrast to most of these studies, our data did not come from observational databases, but from clinical trials and comprised large numbers of patients with homogeneous therapy.

The data reveal that this association is primarily mediated by a higher disease activity in RF+ than RF− patients; the former exhibited higher disease activity by composite measures, higher HAQ scores and particularly higher acute phase reactant levels; also, progression of radiographic scores was significantly higher in the RF+ population, and this pertained to both progression of erosions and JSN. However, on top of increasing disease activity, the presence of RF conveys an activity independent effect on joint damage; this, however, relates particularly to bone erosions but not JSN.

In accordance with these data, the TSS progressed significantly faster over 1 year among RF+ compared with RF− patients. When seropositive and seronegative patients were matched for disease characteristics associated with radiographic progression (including baseline joint damage), thus eliminating the effect of these traditional predictors of joint damage, radiographic progression was still significantly greater in RF+ compared with RF− patients.

In line with previous data, combination therapy of TNF blockers with MTX generally inhibited progression of joint damage irrespective of RF status.

How can the effect of RF be explained? Synovial fluids of RF+ RA patients have been observed to contain immune complexes and exhibit low complement levels,25–29 suggesting activation of complement by immune complexes. It is also well established that immune complexes of RA joints contain RF.28 ,30 Immune complexes, by binding to Fc- and/or Toll-like receptors, activate production of TNF,31–34 a cytokine inducing a cascade of other cytokines, molecules and cells,35 the inhibition of which is beneficial in RA.36 Further, some experimental models of arthritis heavily depend on the presence of autoantibodies37 ,38 and their presence may also enhance cytokine dependent forms of experimental arthritis.39

Explaining the effects of RF that are unrelated to an increase of disease activity may be more difficult. However, immune complexes when binding to Fc-receptors on macrophages activate the spleen tyrosine kinase (Syk) signal transduction cascade.40 ,41 Syk signals via at least two pathways: one of these leads to induction of mitogen activated protein kinases, especially C-jun N-terminal kinases which induce proinflammatory cytokines, in line with the above notion on the increase of disease activity; these cytokines can activate receptor activator of NFκB ligand (RANKL), a pivotal osteoclastogenic molecule. Importantly, C-jun N-terminal kinases can also directly lead to overexpression of RANKL which could partly explain the activity independent effects of RF. Moreover, Syk also induces nuclear factor of activated T cells c1 (NFATc1) via phospholipase Cγ, and NFATc1 is yet another important transcription factor for osteoclast generation.42 ,43 Osteoclasts derive from bone marrow progenitor cells and thus from the monocyte/macrophage lineage.

In contrast to the activity dependent events, the activity independent effects were significantly associated only with progression of the erosion score, whereas the increase of JSN did not reach significance, in line with the above notion on the effects of syk signalling in osteoclasts.

Our study is based on a secondary data analysis. This means that the data had not been obtained for the purpose of testing the present hypothesis. In this regard, our results must be taken as hypothesis generating. However, to test the current hypothesis, the (retrospective) matching approach is one of the best applicable, as even a prospective randomised trial will show considerable differences in the baseline characteristics of seropositive and seronegative patients, despite similar inclusion criteria. In this respect, we consider the large number of patients with provenance from clinical trials, who were required to have high disease activity at baseline and were then newly started on a well-defined intervention, to be the strength of our study. The issue of generalisability is relevant as patients with substantial comorbidities were excluded, but this may be less important in this study that uses a structural outcome as end point. Also, the pooling of different trials studying patients from various centres, countries and regions may yet again increase the generalisability of our findings.

We did not have endpoint measurements of RF, and thus were not able to test if the association is still present after RF titres have potentially changed during MTX therapy, as previously described.44

Yet another limitation of this study is the lack of data on antibodies to citrullinated proteins (ACPA), since these were not assessed in the respective trials. Since ACPA and RF are broadly overlapping45 and joint damage appears related to ACPA and RF levels,46 ,47 it is not clear at present whether the structural changes described here are mainly related to the ACPA+ subpopulation of RF+ patients or not. This question will have to be addressed in additional studies. While some analyses show a higher correlation of ACPA with joint damage than of RF48 it should be borne in mind that the recent NICE assessment as well as the 2010 American College of Rheumatology–European League Against Rheumatism classification criteria for RA did not lend stronger impact on ACPA than RF.4 ,49

The novelty of our findings, however, relates to the cross-sectional and longitudinal assessment of a large database of patients with active RA and the matched analysis of observed (and not modelled) data from a homogenously treated patient population and pertains to the demonstration of two clear relationships: one between seropositivity and disease activity with consequential activity dependent accrual of joint damage, and the other as an activity independent effect of RF+ status on erosions, but not JSN; importantly, the theoretical background, with disease activity leading to induction of bony and cartilage damage and immune complexes amplifying osteoclastogenesis beyond their effects on macrophages and disease activity, is fully in line with our conclusions.

Taken together, our study indicates that there is a role of RF in structural progression of RA that relates to, but also reaches beyond, a direct effect on disease activity. This is in line with the current recommendations by the European League Against Rheumatism which considers RF an independent bad prognostic factor in patients upon experiencing an insufficient response to MTX therapy.50 Along these lines, our data imply that the presence of RF should also be an indicator of the need for more rapid changes toward more intensive therapy in patients newly started on MTX; treatment targets and strategies in RA may thus differ in seropositive and seronegative patients, and may not solely be based on a specific disease activity level.


We thank Abbott, MSD and Pfizer for providing the random sample of the original data from their trials. This study was supported through Coordination Theme 1 (Health) of the European Community's FP7; Grant Agreement number HEALTH-F2-2008-223404 (Masterswitch). This is a publication of the Joint and Bone Center for Diagnosis, Research and Therapy of Musculoskeletal Disorders of the Medical University of Vienna.



  • Contributors DA and JS: Study design, data analysis and drafting of manuscript. FA: Data analysis.

  • Funding Theme 1 (Health) of the European Community's FP7.

  • Disclosure D Aletaha and J Smolen: Received consulting and/or speaking honoraria from Abbott, MSD and Pfizer. F Alasti: None.

  • Patient consent Not obtained.

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