Objective To examine associations of anti-cyclic citrullinated peptide (aCCP) antibody and rheumatoid factor (RF) concentrations with future disease activity in men with rheumatoid arthritis (RA).
Methods Outcome measures were examined in male US veterans with RA and included (1) proportion of observations in remission (disease activity score (DAS28) ≤2.6); (2) remission for ≥3 consecutive months; and (3) area under the curve (AUC) for DAS28. The associations of autoantibody concentration (per 100 unit increments) with outcomes were examined using multivariate regression.
Results 826 men with RA were included in the analysis; the mean (SD) age was 65 (10.5) years and follow-up was for 2.6 (1.3) years. Most were aCCP (75%) and RF (80%) positive. After multivariate adjustment, aCCP (OR 0.93; 95% CI 0.89 to 0.96) and RF concentrations (OR 0.92; 95% CI 0.90 to 0.94) were associated with a lower odds of remission, a lower proportion of observation in remission (p=0.017 and p=0.002, respectively) and greater AUC DAS28 (p=0.092 and p=0.007, respectively). Among patients with discordant autoantibody status, higher concentrations of both aCCP and RF trended towards an inverse association with remission (OR 0.93; 95% CI 0.83 to 1.05 and OR 0.80; 95% CI 0.59 to 1.10, respectively).
Conclusions Higher aCCP concentrations (particularly in RF-positive patients) are associated with increased disease activity in US veterans with RA, indicating that aCCP concentration is predictive of future disease outcomes in men.
Statistics from Altmetric.com
Rheumatoid factor (RF) has long been a component of the diagnostic criteria for rheumatoid arthritis (RA).1 Although the sensitivity of RF is approximately 75–80%, its specificity is modest. RF is found in 3–5% of the general population and in up to 30% of older people.2 Despite this limitation, RF is well studied and its associations with a more severe disease course have been extensively described. High RF concentrations, for example, are associated with unremitting disease, in addition to the presence of subcutaneous nodules and other extra-articular disease manifestations.3 4
Recently, anti-cyclic citrullinated peptide antibody (aCCP) has been shown to be reasonably sensitive (65–80%) and highly specific for RA (up to 98%).2 5 6 Not only is aCCP useful in diagnosis, its presence has been shown to predate clinical disease, by several years in some instances, and it is useful in identifying those with undifferentiated arthritis who will eventually develop RA.7 Moreover, aCCP status provides important prognostic information in RA. Numerous studies have shown that aCCP positivity is associated with joint damage and radiographic progression,8,–,10 and Rönnelid et al showed that aCCP seropositive patients had more active disease during follow-up than seronegative patients.11 However, only a limited number of studies have considered the magnitude of aCCP concentrations rather than qualitative aCCP status (seropositive or seronegative),4 8 11,–,14 and none has examined the prognostic value of aCCP in a male cohort of patients with RA.
While RA is more common in women than men at a ratio of approximately 3:1, there is still a significant proportion of patients with RA who are men. The literature is conflicted as to the impact of gender on disease course.15,–,17 There are data suggesting that men with RA have higher rates of autoantibody seropositivity and greater disease severity than women,15 16 but no studies examining associations of disease-related autoantibodies with disease course in large populations of men.
Given the potential toxicity of disease-modifying antirheumatic drugs (DMARDs) and the benefit of early/aggressive treatment, the prompt identification of patients with RA at greatest risk for unfavourable outcomes is important in guiding treatment decisions. The relative paucity of data on the prognostic value of aCCP concentration beyond seropositivity and the potential clinical utility of such information indicate a clear need for inquiry. The aim of this study was to examine the extent to which the magnitude of aCCP and RF concentrations are associated with disease activity in a well-characterised population of men with established RA.
Study subjects included US veterans enrolled in the Veterans Affairs RA (VARA) Registry,18 with active enrolment sites at VA Medical Centers in Brooklyn, New York; Dallas, Texas; Denver, Colorado; Iowa City, Iowa; Jackson, Mississippi; Omaha, Nebraska; Portland, Oregon; Salt Lake City, Utah; and Washington, DC. All patients satisfied American College of Rheumatology classification criteria for RA.1 The Registry includes VA beneficiaries with onset of RA after 18 years of age. There are no other inclusion/exclusion criteria for enrolment.
In addition to collecting serum and DNA at enrolment, the VARA Registry includes baseline and longitudinal clinical data, the latter collected as part of routine care. All treatment decisions during observation are made at the discretion of the treating rheumatologist. Enrolment variables include: diagnostic criteria (including subcutaneous nodules), comorbidity, smoking status (never, former or current), sociodemographics (education, race/ethnicity, age, sex), date of RA diagnosis and prior DMARD use. aCCP (IgG) was measured on banked serum using a second-generation ELISA (Diastat, Axis-Shield Diagnostics, Dundee, UK; positivity ≥5 U/ml). RF was determined by nephelometry (Siemens Healthcare Diagnostics, Munich, Germany; positivity ≥15 IU/ml).
Measures collected at enrolment and during follow-up for this study included tender and swollen joint counts (0–28), erythrocyte sedimentation rate (ESR, mm/h), pain (0–10), a 10-item multidimensional Health Assessment Questionnaire score (range 0–3),19 patient global well-being (100 mm visual analogue scale) and treatments. RA treatments were classified into three categories: (1) biological agents (infliximab, etanercept, adalimumab, rituximab and abatacept); (2) non-biological DMARDs (methotrexate, sulfasalazine, leflunomide, minocycline, doxy-cycline and hydroxychloroquine); and (3) prednisone.
Patients were excluded if autoantibody data (both RF and aCCP) were not available. Patients were also excluded if they had only a single clinical observation and/or follow-up duration ≤6 months. Finally, patients for whom the four-variable disease activity score (DAS) based on 28 joint counts and ESR (DAS28, 4V-ESR)20 could not be calculated after accounting for missing variables were also excluded.
Three outcomes were examined: (1) achievement of sustained remission, defined as at least two consecutive observations separated by at least 3 months with DAS28 ≤2.6, a level which has been widely used as indicative of remission21; (2) proportion of follow-up time spent in remission (a continuous measure, range 0–1) to account for differences in follow-up duration (time in remission credited only when remission was maintained for at least two sequential visits); and (3) the area under the DAS28 curve (AUC) per year of follow-up (see example in online supplement).
For missing components of DAS28 (tender and swollen joint counts (3.7% of observations missing), ESR (7.4% of observations missing) and patient global well-being (11.1% of observations missing)), a last observation carried forward approach was used. Imputation for missing DAS28 values was not used due to poor correlation of ESR and global well-being with other clinical measures. If there were no preceding observations, the first observation available was carried backwards. For educational level (missing in 203 patients), dichotomised as high school graduate or not, and smoking status (missing for 11 patients), missing values were imputed (Stata10 command ‘uvis’, StataCorp, College Station, Texas, USA). Education level was imputed based on age, race/ethnicity, gender and smoking status, while smoking status was imputed based on age, race, gender, education level and comorbidities of chronic obstructive pulmonary disease (COPD) and ischaemic heart disease (IHD).
Baseline aCCP and RF were considered dichotomously (seropositive vs seronegative), categorically as low, moderate and high concentration (using tertiles of seropositive patients and referent to seronegative) and continuously. Among aCCP positive patients, the thresholds for low, moderate and high concentration groups were ≥5, ≥126.3 and ≥333.2 U/ml. For RF seropositive patients, the thresholds for the categorical groups were ≥15, ≥89.5 and ≥340.5 IU/ml.
Associations between aCCP, RF and the three outcomes were examined using linear regression for continuous outcomes and logistic regression for dichotomous outcomes. Given the high concordance of aCCP and RF status (84%) and correlation of continuous values (r=0.44, p<0.001), these autoantibodies were modelled separately in primary analyses. All models were adjusted for age. Additional covariates included in multivariate models were race, education, disease duration, follow-up time, smoking status, comorbidity burden (a count of presence of diabetes mellitus, IHD, hypertension, cerebrovascular disease, chronic kidney disease, hyperlipidaemia and COPD; range 0–7), DMARD and/or prednisone use, and disease remission status (DAS28 ≤2.6 vs DAS28 >2.6) at enrolment. In linear regression models examining the proportion of time in remission, the β coefficients correspond to fractional changes in time spent in remission (eg, βi=0.12 corresponds to 12% more time in remission for each increase in the ith covariate). Pharmacological interventions based on the three aforementioned categories were accounted for using two approaches: use of medication at enrolment and the addition of new medications (within classes) during follow-up. Differences based on the enrolment site were accounted for by cluster analysis. In subanalyses, we examined the associations of aCCP and RF with outcomes among patients with recent onset disease, defined as disease duration of <2 years at enrolment. In exploratory analyses we also examined the associations of autoantibody concentrations with our outcomes in those patients with discordant autoantibody status to more closely examine the importance of the individual autoantibodies.
Of the 1285 patients enrolled in the VARA Registry at 1 July 2009, aCCP and RF data were available for 1098. After excluding women, those with fewer than two observations and those with <6 months follow-up, 827 patients were eligible for analysis. Finally, after removing a single patient with insufficient data for calculation of DAS28, 826 patients (7937 observations) remained.
The patients were predominantly Caucasians with well-established RA (table 1). They had a mean (SD) of 3.96 (1.51) visits per year of follow-up. Additional characteristics, including measures of disease activity at enrolment, follow-up duration, medication use, comorbidity score and autoantibody status are summarised in table 1. Most of the patients were positive for aCCP (75%) and RF (80%). Discordant autoantibody status was relatively uncommon (16%), while dual autoantibody positivity was far more common (69%). In addition, patients had moderate disease activity as reflected by a mean DAS28 of 3.96 (1.65) at enrolment.
An episode of sustained remission (including ≥2 consecutive visits >3 months apart) was achieved by 245 (30%) patients. After multivariate adjustment, both baseline aCCP (OR 0.67; 95% CI 0.50 to 0.92) and RF (OR 0.60; 95% CI 0.49 to 0.75) positivity were associated with a lower odds of sustained remission (figure 1). Likewise, higher concentrations of baseline aCCP and RF (modelled as a continuous variable and in ordered categories) were also associated with a lower likelihood of sustained remission. The associations of ordered autoantibody categories with remission were most pronounced at the highest concentrations. Covariates associated with a higher likelihood of sustained remission in models examining continuous aCCP concentration included years of follow-up (OR 1.52; 95% CI 1.33 to 1.74) and clinical remission at enrolment (OR 10.8; 95% CI 8.40 to 13.8). Covariates independently associated with a lower likelihood of remission included older age (OR 0.97; 95% CI 0.95 to 0.98), the addition of three new pharmacological agents during follow-up (OR 0.27; 95% CI 0.09 to 0.88) and four comorbid conditions (OR 0.33; 95% CI 0.13 to 0.85).
Proportion of time in remission
The mean (SD) proportion of follow-up time spent in remission was 0.15 (0.28). After multivariate adjustments, aCCP (β=−0.057; p=0.057) and RF (β=−0.062; p=0.023) positivity were associated with less time in remission, although aCCP positivity did not reach significance (table 2). As with sustained remission, higher levels of aCCP and RF (modelled as concentrations and in ordered categories) were associated with less follow-up time in remission (table 2). Covariates significantly associated with a lower proportion of follow-up in remission in multivariate models examining aCCP concentration included the addition of two (β=−0.083, p=0.022) or three (β=−0.109, p=0.004) new pharmacological agents during follow-up and the addition of prednisone one, two or three times during follow-up (β1=−0.028, p=0.013; β2=−0.073, p<0.001; β3=0.158, p<0.001). The only covariate independently associated with an increased time in remission was clinical remission at enrolment (β=0.327; p=0.001).
Area under the curve
Analyses of DAS28 AUC revealed similar associations as those observed with sustained remission and proportion of time in remission. Following multivariate adjustment, both aCCP (β=0.17; p=0.023) and RF (β=0.28; p=0.001) positivity were associated with larger AUC (table 2). Referent to seronegative patients, those with moderate-titre aCCP and RF were associated with greater AUC, as was high-titre RF and RF concentration (table 2). For both high-titre aCCP and aCCP concentration, there was a strong trend towards a significant inverse relationship (table 2). Covariates significantly associated with a greater DAS28 AUC in multivariate models examining aCCP concentrations included the addition of one (β=0.22, p=0.037), two (β=0.65, p=0.002), three (β=0.95, p<0.001) or four (β=0.92, p=0.003) new pharmacological agents during follow-up. Covariates independently associated with a decrease in DAS28 AUC included clinical remission at enrolment (β=−1.19; p=0.002) and the presence of two (β=−0.16, p=0.024) comorbid conditions. The results for DAS28 AUC and all other outcomes examined above were not changed with removal of clinical remission at enrolment from the models (data not shown).
In an age-adjusted analysis of patients with recent-onset RA (disease duration ≤2 years, n=162), aCCP (OR 0.74; 95% CI 0.62 to 0.88) positivity was inversely associated with sustained remission, while RF positivity (OR 0.51; 95% CI 0.23 to 1.11) only trended towards a significant inverse relationship. The results were not changed after multivariate adjustment (data not shown). The inverse associations of aCCP and RF concentrations with sustained remission in patients with recent-onset disease were greatest in those with higher serum concentrations (data not shown). Patterns of associations of aCCP and RF seropositivity with DAS28 AUC and proportion of follow-up in remission were similar to those described for sustained remission (data not shown).
In multivariate models examining the relationship between dual autoantibody status and sustained remission, aCCP(−)/RF(+) status (OR 0.27; 95% CI 0.10 to 0.71) and aCCP(+)/RF(+) status (OR 0.42; 95% CI 0.23 to 0.74) were inversely associated with sustained remission, referent to aCCP(−)/RF(−) status. aCCP(+)/RF(−) status only trended towards significance (OR 0.47; 95% CI 0.11 to 2.09).
We have shown that higher absolute concentrations of aCCP and RF in a cohort of male US veterans with RA are associated with poorer clinical outcomes including greater disease activity and less sustained remission over time. These associations appear to be explained primarily by those patients with the highest autoantibody concentrations. Our analyses did not reveal predictive superiority of one autoantibody over the other, contrary to previous reports suggesting that aCCP is a stronger predictor than RF in outcomes such as radiographic changes.22 23
Our results regarding aCCP magnitude are consistent with recent reports examining predominantly female patients with RA. Berglin et al found that aCCP concentration, beyond aCCP positivity, was indicative of more aggressive radiographic progression and worse disease severity,8 and del Val del Amo et al reported similar findings.12 Higher aCCP concentrations at baseline were associated with greater disease activity during follow-up11 and with more aggressive radiographic progression.13 Turesson et al showed in a small case–control study that patients with extra-articular disease tended to have higher aCCP concentrations, although the association did not reach significance.4 Only the study by Lee et al14 failed to show associations between aCCP titres and disease severity markers. While these studies showed relationships between aCCP concentration and disease activity, our study is unique in its simultaneous consideration of both aCCP and RF concentrations, its larger sample size and its inclusion of only men with RA.
The results of this investigation may have limited generalisability, particularly since this study population was composed entirely of men. However, men with RA are historically under-represented in epidemiological studies, even excluded from major studies conducted in the USA.24 25 Furthermore, men with RA have been shown to have a more aggressive disease course than women, one characterised by increased radiographic progression, a higher prevalence of extra-articular manifestations and greater disease-related mortality.15 16 The VA represents the largest integrated healthcare system in the USA and, as such, provides a unique opportunity to study a diverse and arguably vulnerable RA population where socioeconomic barriers to healthcare access are limited.
Another possible limitation is the length of disease duration at enrolment with a mean of approximately 12 years and the corresponding receipt of previous RA-related therapies. This is potentially important since other investigations have shown that aCCP and RF concentrations can be influenced by DMARD treatment, although the magnitude of this effect appears to be modest (approximately 10% change in serum concentrations associated with biological therapy).26 Furthermore, changes in aCCP related to RA treatments only rarely result in a change in aCCP status (eg, seroconversion from positive to negative) and are generally insufficient to cause changes in the aCCP categories examined in this study.27
Our exploratory subanalyses limited to a select number of patients with recent-onset disease showed risk estimates that were similar to those from analyses of the entire study population, suggesting that the associations of aCCP and RF concentrations with disease activity are independent of disease duration. Although we examined DMARD (biological and non-biological) and glucocorticoid use as potential confounders (both at baseline and initiations during follow-up), we did not examine the associations of autoantibody concentrations with treatment responses to specific therapeutic agents given the heterogeneity in treatments received. This may be relevant since the associations of autoantibody concentrations with disease activity could be operative through specific effects on treatment response. We also recognise that the definition of sustained clinical remission in RA is not universally established. However, by considering disease burden in three different outcome variables, which showed consistent relationships with aCCP magnitude, the limitations of these definitions appear to be less relevant.
In summary, our results show that higher baseline aCCP concentrations are associated with greater disease activity over time in men with established RA, a prognostic attribute that extends beyond that of the qualitative (seropositive vs seronegative) status. Although limited in power, our exploratory analyses of patients with discordant aCCP/RF status failed to show that either autoantibody was dominant in driving the observed relationships. These results emphasise the importance of further investigations in larger, more diverse RA populations including individuals with discordant antibody status.
The authors thank Ms Debra Bergman and Mr Bart Hamilton for their assistance in this work and the many US veterans who have generously participated in this research.
Funding BJM was supported by the Ephraim P Engleman Endowed Resident Reseach Preceptorship from the American College of Rheumatology (ACR) Research and Education Foundation. This work was funded by a grant from NIH/NIAMS (R03 AR054539, PI TRM). The VARA Registry has received research support from the Health Services Research & Development (HSR&D) Program of the Veterans Health Administration (VHA) in addition to unrestricted research funds from Abbott Laboratories and Bristol-Myers Squibb. TRM receives research support from NIAMS (K23 AR050004) and the VHA (VA Merit). LC is supported by a VA HSR&D Career Development Award.
Competing interests None.
Ethics approval This study was conducted with the approval of the IRB from all the enrolling sites (VA Medical Centers in Dallas, Texas; Denver, Colorado; Jackson, Mississippi; Omaha, Nebraska; Portland, Oregon; Salt Lake City, Utah; and Washington, DC) and patients provided informed written consent prior to enrolment.
Provenance and peer review Not commissioned; externally peer reviewed.
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.