Article Text

Download PDFPDF

Quantifying anti-cyclic citrullinated peptide titres: clinical utility and association with tobacco exposure in patients with rheumatoid arthritis
  1. D M Lee,
  2. R Phillips,
  3. E M Hagan,
  4. L B Chibnik,
  5. K H Costenbader,
  6. P H Schur
  1. Department of Medicine, Division of Rheumatology, Immunology and Allergy, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts, USA
  1. D M Lee, Brigham and Women’s Hospital, Rheumatology, Immunology and Allergy, Smith 552B, 1 Jimmy Fund Way, Boston, Massachusetts 02115, USA; dlee1{at}


Objective: To determine the significance of quantitative levels of antibodies to cyclic citrullinated peptides (anti-CCP) in a population of patients with rheumatoid arthritis (RA).

Methods: A total of 241 consecutive sera from patients with RA sent from a large rheumatology clinic for laboratory testing were selected for precisely quantifying anti-CCP antibody titres with the anti-CCP2 assay. Patient charts were reviewed for demographic information, smoking history, clinical diagnosis, rheumatoid factor (RF) titre, radiographic information and other laboratory information (erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) level). Correlations with anti-CCP titre and RF titre, disease parameters and smoking history were assessed.

Results: We confirm previous findings that anti-CCP seropositivity is associated with a higher incidence of erosions in patients with RA (56% vs 20% CCP+ vs CCP−, κ = 0.297, p<0.001). We also found a moderate correlation between anti-CCP titre and RF titre. However, we failed to find an association between anti-CCP titre and presence of erosions, between anti-CCP titre and CRP or ESR level, or between anti-CCP titre and age or disease duration. Interestingly, we did find significantly higher anti-CCP titres in patients with a history of smoking (452 units/ml vs 229 units/ml, smokers vs non-smokers, respectively; p = 0.02).

Conclusions: Although anti-CCP titres were not associated with clinical parameters of disease, they are increased in patients with RA with exposure to tobacco. By contrast, no elevation in RF was noted in patients with a history of smoking. These observations are consistent with a pathogenic contribution of smoking to RA and suggest the immune stimulus for anti-CCP is distinct from that for RF.

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.

Since their original descriptions,1 2 multiple studies have now established the clinical utility of measuring anti-cyclic citrullinated peptide (CCP) autoantibodies via commercial assays as a biomarker in rheumatoid arthritis (RA).326 In addition to displaying a pooled sensitivity of 67% and specificity of 95% for the diagnosis of RA,27 it is now established that anti-CCP autoreactivity is prognostic for disease severity, radiographic erosions, as well as development of RA in synovitis of recent onset.1926 28 Rather than supplanting measurement of rheumatoid factor (RF), the utility of anti-CCP has been demonstrated by itself and in combination with measurement of serum RF. The consensus that has emerged from these studies is that while anti-CCP and RF are frequently overlapping in their presence and prognostic significance, they are also independently informative in a significant subset of patients. The differences between subsets of patients with RA with distinct autoantibody profiles remain poorly understood.

In addition to its performance as a biomarker, recent evidence has also suggested a pathogenic association between anti-citrulline autoantibodies (measured by the anti-CCP test) and RA (reviewed in van Gaalen et al).29 Retrospective studies in multiple cohorts have demonstrated that measurable anti-CCP autoantibodies frequently precede development of clinical arthritis with evidence for an increase in frequency and titres proximal to disease onset.19 30 At a population level, recent studies have shown that anti-CCP presence correlates with presence of specific human leukocyte antigen (HLA)-DR alleles with a particularly strong association with the “shared epitope” genes.26 3133 At a basic mechanisms level, citrullinated peptides from demonstrated autoantigens bind more avidly to major histocompatability complex (MHC) containing the shared epitope than native peptides and is thereby capable of activating CD4+ T lymphocytes, thus providing a plausible molecular basis for the genetic association seen in patients with RA.34

An additional connection between smoking (a strongly associated environmental risk factor for RA) and anti-citrulline autoantibodies provides further potential insight into mechanisms of disease evolution. It has long been appreciated that smoking confers an increased risk of RA.3539 Although not replicated in all cohorts, recent studies demonstrate that the increased risk conferred by tobacco exposure is restricted to the anti-CCP+ subset of subjects and is associated with HLA-DR alleles.32 33 Furthermore, pulmonary lavage samples from smokers shows citrullination of cellular proteins that is not present in non-smoking subjects thereby providing a source of citrullinated antigens to induce autoantibody production. Moreover, citrulline post translational modification of tissue proteins occurs in diseased joints and joint tissue citrullination levels elevate markedly in the context of inflammation in humans and experimental rodents.4044

Evidence from animal models further supports the potential contribution of anti-citrulline antibodies to the pathogenesis of inflammatory arthritis. Recent studies in the collagen induced arthritis model have demonstrated that anti-citrulline autoantibodies are induced by immunisation with collagen and that establishing tolerance to citrullinated peptides blocks development of disease.42 Perhaps more compelling is the ability to passively transfer anti-citrulline antibodies to mice with subclinical arthritis and initiate robust symmetric inflammatory synovitis. Interestingly, passive transfer of anti-citrulline antibodies into mice without any inflammatory stimulus does not elicit arthritis and citrullination of joint tissue proteins is evident in mice with subclinical collagen-induced arthritis (CIA). Taken together, these observations have suggested a new disease model wherein the combination of anti-citrulline antibodies and inflammation-induced citrullination of tissue proteins can conspire to induce an autoimmune inflammatory arthritis (reviewed in van Gaalen et al).29

These substantial insights notwithstanding, there remain many questions that require further investigation. A significant topic worthy of further scrutiny is the contribution of anti-CCP titre to disease activity. Assays for clinical measurement of anti-CCP typically feature a limited dynamic range. It is appreciated that many patients with RA have anti-CCP titres greater than the upper limit of the anti-CCP assay, but precise quantification of anti-CCP titres requires serial dilution of serum; a laborious and substantially more costly requirement. Thus, the significant majority of studies document presence of anti-CCP autoantibodies, but do not proceed to quantify anti-CCP titres and the association of titre with disease parameters. Noting the high frequency of anti-CCP titres greater than assay in patients with RA in our clinic, we elected to examine the association between anti-CCP titre and disease parameters as well as assess the utility of full quantification of anti-CCP titres in clinical practice. While we have found an interesting association between tobacco exposure and anti-CCP titre, we thus far have not found clinical utility in full quantification of anti-CCP titres.


Serum samples

A total of 241 unique consecutive sera from patients with RA sent from the Brigham and Women’s Hospital (BWH) Arthritis Center to the BWH Clinical Immunology Laboratory for rheumatology testing were selected for further analysis. The BWH Arthritis Center is a large rheumatology practice comprised of 41 practicing rheumatologists with an outpatient count of >35 000 visits/year. Patient charts were reviewed for demographic information, clinical diagnosis, smoking history, age of RA onset, years of disease, radiographic information, RF measurement, erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) laboratory values. For smoking history, chart examination included assessment of cigarette smoking history (current, past or never smoking). Diagnosis of RA was established by attending rheumatologist diagnosis and/or by review of laboratory, radiological and clinic notes and by applying American College of Rheumatology (ACR) classification criteria.45 All studies received Institutional Review Board approval. The majority (80.1%) of these patients were female with a wide variation in age (21−95 years) (table 1). Mean (SD) age of RA onset was 48 (15) years.

Table 1 Subject demographics

Radiographic analysis

Radiographic identification of joint erosions in hands and feet was investigated using electronic medical records. Joint radiographs were available for 214 of the 241 patients. All radiographic diagnoses were abstracted from formal interpretations by an American Board of Radiology certified attending radiologist.

Data measurement and analysis

Anti-CCP measurement

Anti-CCP activity was determined on sera from all subjects by ELISA using the Diastat anti-CCP2 assay (Axis-Shield, Dundee, UK). When CCP2 results were initially greater than 100 units, sera were retested at 1:16, 1:64 and 1:128 dilutions until a measurement within the linear range of the assay was obtained. The CCP2 was then derived by calculating the measured result by the dilution factor of the sera. RF titre (available for 214/241 subjects) performed by nephelometry (N Latex RF, Dade-Behring, Newark, Delaware, USA)46, ESR (available for 211/241 subjects)47 and CRP (available for 225/241 subjects; CardioPhase hsCRP, Dade-Behring) were obtained from subject medical records.

Statistical analyses and graphs were performed using the Prism V. 4.0 package (Graphpad, San Diego, California, USA). The Student two-tailed t test, Spearman correlation tests and κ coefficient tests of agreement were performed to assess statistical significance.


The initial goal of this study was to confirm our impression that a significant number of patients have anti-CCP titres that are higher than those measured in the linear range of the commercial anti-CCP2 assay. We therefore acquired 241 sequential discarded sera sent for laboratory testing from a large academic primary rheumatology clinic. Indeed, we found that of the 73% of subjects with a positive anti-CCP test, 42% had initial anti-CCP measurements higher than assay (⩾100 Units).

Further characterisation of our cohort reveals a patient population similar to other published cohorts (table 1). The majority (80%) were female and were RF+ (83%). No difference in RF (329 units/ml vs 291 units/ml) or anti-CCP titre (329 units/ml vs 301 units/ml) was seen between male and female subjects, respectively. Overall mean (standard error of the mean (SEM)) antibody titres were 298 (50) units/ml for RF and 318 (50) units/ml for anti-CCP. From a severity standpoint, 46% of subjects who had radiographic data available demonstrated evidence of hand and/or foot erosions. Consistent with previous studies, there was substantial overlap in RF and anti-CCP reactivity with 93% of anti-CCP+ sera containing RF activity and 81% of RF+ sera containing anti-CCP reactivity (κ = 0.417, p<0.001) (table 2). Given the extensive overlap of humoral RF and anti-CCP autoantibodies, we assessed a correlation between levels of RF and anti-CCP reactivity. Indeed, we found a moderate correlation between RF and CCP titres in these patients (Spearman r = 0.52, p<0.001) although several subjects displayed substantial disparity in the level of autoantibodies present (fig 1).

Figure 1 Correlation of rheumatoid factor (RF) and cyclic citrullinated peptide (CCP) titres. Shown is a scattergraph of anti-CCP and RF values in our cohort of patients with rheumatoid arthritis (RA). n = 214 subjects. Spearman r = 0.52, p<0.001.
Table 2 Comparison of anti-cyclic citrullinated peptide (CCP) and rheumatoid factor (RF) reactivity

To examine an association between anti-CCP and disease severity, we assessed anti-CCP seropositivity and titres in patients with or without radiographic evidence of erosions. As noted in numerous other cohorts,4 4859 we found the cohort of anti-CCP seropositive subjects demonstrated increased frequency of erosions when compared to the anti-CCP seronegative cohort (56% vs 20% CCP+ vs CCP-, κ = 0.279, p<0.001) (table 3). However, examination of anti-CCP titres in subjects with or without erosions reveals overlapping levels in these patient populations (331.1 (±59.26 vs 310.0 ±75.31 units/ml erosion(+) vs erosion(−), p = 0.83) (fig 2). Since an association between erosions and RF seropositivity has been appreciated in some cohorts,48 50 51 53 54 58 6070 and not in others,49 5557 71–76 we further investigated this association in our cohort and found a significantly increased frequency of erosion in patients who were RF seropositive (47% vs 28%, κ = 0.094, p<0.001) (table 4). As with anti-CCP titre, there was no significant difference in the titre of RF in those with or without erosions (fig 2B).

Figure 2 Autoantibody levels and erosive disease in rheumatoid arthritis (RA). Shown are average levels of anti-cyclic citrullinated peptide (CCP) (A) or rheumatoid factor (RF) (B) in patients with RA with or without radiographic erosions. Error bars show standard error of the mean (SEM). A. n = 98 erosion (+) and 116 erosion (−) p = 0.83. B. n = 85 erosion (+) and 108 erosion (−) p = 0.37.
Table 3 Comparison of autoantibodies and radiographic erosions
Table 4 Comparison of autoantibodies and radiographic erosions

We next examined an association between markers of inflammation and elevated anti-CCP. Here we utilised the erythrocyte sedimentation rate and C-reactive protein values as surrogates for degree of systemic inflammation in our cohort of patients. Somewhat surprisingly, we found no correlation with either marker of inflammation and anti-CCP titre (fig 3). We further extended these studies by examining an association between inflammatory indices and RF titre. Similarly, we found no meaningful association between ESR or CRP and RF titre.

Figure 3 Association between autoantibody levels and indices of inflammation. Shown are scattergraphs of anti-cyclic citrullinated peptide (CCP) (A,B) and rheumatoid factor (RF) (C,D) plotted against either C-reactive protein (CRP) or erythrocyte sedimentation rate (ESR). Note the lack of meaningful correlation in these comparisons. A. n = 225 subjects. Spearman r = 0.16, p = 0.003. B. n = 211 subjects. Spearman r = 0.16, p = 0.018. C. n = 198 subjects. Spearman r = 0.21, p = 0.0027. D. n = 189 subjects. Spearman r = 0.13, p = 0.07.

Previous studies have demonstrated that anti-CCP reactivity can precede disease onset and that anti-CCP values elevate in proximity to disease onset.19 30 These observations prompted us to examine an association between disease duration and anti-CCP titre. These analyses demonstrate no association between duration of disease and anti-CCP titre (fig 4A). We also assessed an association between age and anti-CCP titre. As seen in fig 4B, there is no observable association between age and anti-CCP titre in this cohort. Examination of RF titre also shows lack of association with age or disease duration (Fig. 4C,D).

Figure 4 Association between autoantibodies and age or years of disease. Shown are scattergraphs of anti-cyclic citrullinated peptide (CCP) (A,B) and rheumatoid factor (RF) (C,D) plotted against either subject age or years of disease. Note the lack of meaningful correlation in these comparisons. A. n = 215 subjects. Spearman r = −0.05, p = 0.43. B. n = 241 subjects. Spearman r = −0.04, p = 0.58. C. n = 199 subjects. Spearman r = −0.08, p = 0.25. D. Spearman r = 0.36, p = 0.36, n = 214 subjects.

Finally, given the recent demonstration that smoking associates with anti-CCP reactivity and rheumatoid arthritis, we investigated an association between smoking history and the level of anti-CCP reactivity in patients with RA. Indeed, we found that patients with RA with a history of tobacco use demonstrate a mean anti-CCP value twice that seen in non-smoking RA subjects (452 units/ml vs 229 units/ml respectively, p = 0.02) (fig 5A). Although not statistically significant, further separation of subjects with tobacco exposure into those who currently smoke and those who have discontinued tobacco use demonstrates the highest average anti-CCP values in the cohort of patients actively smoking (anti-CCP = 540, 391 and 229 units/ml respectively, active smoker vs past smoker vs no tobacco use history, p = 0.59 current vs past smoker, p = 0.02 current smoker vs no tobacco use and p = 0.11 past smoker vs no tobacco use respectively). Furthermore the highest mean anti-CCP titre (669 units/ml) was observed in those patients who had erosions and were current smokers (vs a mean anti-CCP of 217 units/ml in those patients without erosions and not current smokers (p = 0.021). Having noted an association between smoking history and anti-CCP titre, we assessed the RF levels in subjects with or without tobacco exposure. Here, we found no discernable difference in RF titres associated with exposure to tobacco (fig 5B).

Figure 5 Autoantibody levels and tobacco exposure. Shown are average levels of anti-cyclic citrullinated peptide (CCP) (A) or rheumatoid factor (RF) (B) in patients with rheumatoid arthritis (RA) with or without smoking exposure. Error bars show standard error of the mean (SEM). A. n = 62 with and 141 without smoking exposure. p = 0.025. B. n = 55 with and 127 without smoking exposure. p = 0.84.


Multiple studies have demonstrated the clinical utility of measuring anti-citrulline autoantibodies using anti-CCP assays in patients with inflammatory arthritis. From a pathogenesis standpoint, numerous associations between anti-CCP and rheumatoid arthritis suggest a mechanistic contribution to disease including presence prior to disease onset, presence of autoantigen in target articular tissues and presence of immune complexes in arthritic joints.29 Furthermore, recent evidence has also suggested a pathogenic contribution in rodent models of arthritis.42 While these observations have provided substantial insights, there remain many facets of the clinical usage of the anti-CCP assay as well as the contribution of these autoantibodies to disease that remain poorly understood.

Our findings demonstrate a remarkable spectrum in titres for autoantibodies to anti-CCP in patients with RA. Since most studies assessing the utility of the anti-CCP assay in arthritis have examined the presence of anti-CCP and its association with clinical parameters, the majority of clinical laboratories perform this assay on serum samples at a fixed dilution and report readings that fall above the linear range of the assay standard curve as a single value (either “>assay” or “>100” for the anti-CCP2 assay). Thus, the utility of serial dilution measurements to quantify the true titre of autoantibodies remains largely undefined either from a prognostic standpoint or from a disease mechanism standpoint. The frequency of undiluted titres greater than the assay range provides impetus for further scrutiny of the clinical import of fully quantifying levels of anti-CCP reactivity.

Given the recent insightful studies suggesting a gene-environment interaction between HLA-DRB1 alleles and smoking exposure,32 33 the significant elevation of anti-CCP titres in tobacco-exposed subjects in this cohort is intriguing. In combination with numerous studies documenting the strong association between cigarette smoking and RA,3539 77 and cigarette smoking and the presence36 38 7881 and titre of rheumatoid factors 82 83 this finding supports the notion that the association between smoking and RA is causal and mediated at least in part by contributing to expansion of citrulline directed autoantibodies. However, our observation of an elevated mean anti-CCP titre (230 units) in the non-tobacco exposed subset in our cohort also suggests that stimuli other than tobacco exposure can drive development of anti-citrulline autoantibodies. The relationship between the presence of HLA-DRB1 alleles and other genetic risk factors for RA and anti-CCP antibody titres is an enticing avenue for further research and these studies are ongoing.

Our results show a surprising lack of correlation between anti-CCP titre and other disease parameters. More specifically, we found no association with markers of systemic inflammation (either ESR or CRP), with disease severity as manifested by joint erosions, or with age or disease duration. Given the evidence arguing for a pathogenic contribution from anti-citrulline autoantibodies in RA, simple modelling would predict that higher autoantibody titres should drive more aggressive disease whose manifestation (CRP, erosions and others) should correlate with titre. Numerous considerations could explain this apparent discrepancy including autoantibody isotype profile, autoantibody glycosylation state84 or the subset of anti-citrulline autoantibodies captured in the commercial anti-CCP ELISA assays. Given that the sequence of the citrullinated peptides used in commercial assays is proprietary, it is impossible to know the true “autoreactivity” of the antibodies measured in this context. It is thus plausible that truly pathogenic anti-citrulline autoantibodies are distinct from those measured in the commercial anti-CCP assay used in this study. At the very least, this observation underscores the complexity of pathogenic processes in RA and the continued need for research into the role of autoantibodies in this disease.

The moderate correlation between elevations in RF and anti-CCP are also intriguing from a mechanistic standpoint. Since both assays measure a parameter associated with humoral immunity and since both assays are highly associated with a diagnosis of RA, it is intuitive to hypothesise that their elevations should correlate in this autoimmune disease. The modest degree of correlation in elevation of both autoantibodies and striking instances of disparity in autoantibody titres highlights the complexity of the autoimmune response within the B cell compartment of the adaptive immune system in RA and is consistent with the notion that distinct primary means of immunological activation can conspire to drive the autoimmune inflammatory response in RA.

Although accurately quantifying anti-CCP titre provides potential mechanistic associations of interest in RA, our study does not provide compelling rationale for performing these measurements in the clinical management of patients with disease. In contrast to the strong data from previous studies and replicated in part here that shows the utility of simply identifying the presence of anti-CCP autoantibodies aids in confirming the diagnosis of RA and in prognosticating increased severity, we found no connection between anti-CCP titre and disease severity as assessed by radiographic erosions or by measurement of inflammatory indices. These findings are consistent with recently published findings in other cohorts.56

Our observations need to be interpreted in light of the limitations of the cohort and methods employed in this study. Radiographic analyses were limited to documentation of erosions presence or absence; no formal measurements of radiographic severity were performed. We can thus not exclude the possibility that an association between anti-CCP titre and radiographic severity exists. In addition, measurements of CCP, RF and inflammatory markers values were not performed concurrently. We cannot therefore account for time-dependent fluctuations in these values as a confounder in these studies. Furthermore, the cross-sectional design of this study precludes detection of association of changes in anti-CCP titre and disease parameters in RA. Documentation of an association between disease severity or response to therapeutic interventions and a temporal change in anti-CCP titre would provide strong rationale for accurate quantification of anti-CCP reactivity in clinical management of RA. This cross-sectional study also did not account for the multiple different RA medications received by the subjects. Finally, our study size precludes identification of small associations between anti-CCP titre and the RA clinical parameters where we failed to find correlation.

In summary, accurate quantification of anti-CCP titres in a cohort of subjects with RA from a large outpatient practice reveals substantial variation of anti-CCP autoantibody levels that is not appreciated in standard laboratory testing of undiluted blood samples. While these measurements reveal associations with potential for further mechanistic insights, the current findings do not support full anti-CCP quantification in routine clinical practice. Further investigation of potential clinical utility is warranted.


The authors gratefully acknowledge the expert technical assistance of Siobhan Gunn.



  • Competing interests: None declared.

  • Funding: Funding support was received from the Cogan Family Foundation.

  • Ethics approval: Ethics approval was obtained.