The now widely employed CCP2 assay, which detects circulating autoantibodies to a panel of synthetic, cyclic citrullinated peptides, carries a positive predictive value of 0.93 for rheumatoid arthritis (RA) among early undifferentiated arthritis (UA) patients,1 thus providing a cornerstone of recently evolved scoring systems that predict and classify early RA.2 However, approximately 80% of newly presenting UA patients are anti-CCP2-negative3 and a quarter of these evolve into RA within 3 years,1 thus experiencing diagnostic delay. The peptides used in the CCP2 assay do not necessarily correspond to in vivo generated proteins, yet citrullinated antigens of putative pathophysiological relevance have recently been described in association with specific circulating autoantibodies in RA.4 Therefore, we hypothesised that the detection of one or any such autoantibodies, as well as those against filaggrin components of the CCP1 assay,5 IgA or IgM rheumatoid factor (RF), might add to the diagnostic utility of CCP2 testing alone in predicting RA progression among early UA patients.

Consecutive adult UA patients with recent onset arthritis symptoms, naive to disease-modifying antirheumatic drugs and corticosteroids, were recruited from the Freeman Hospital early arthritis clinic, Newcastle upon Tyne, UK. Their diagnostic outcome was ascertained at the end of the study (1987 American College of Rheumatology classification criteria for RA6), at which point a median of 28-month follow-up had been completed (greater than 12 months in all cases). In addition to anti-CCP2, baseline sera were tested for antibodies with specificity for a range of citrullinated peptides, namely linear citrullinated filaggrin peptide (Fil-LC), a panel of synthetic, citrullinated, cyclic pro-filaggrin-derived peptides (cFc 1, 2, 5, 6A, 6B and 9),5 cyclic citrullinated α-enolase peptides, designated ‘−1’ and ‘−11’ (CEP1 and CEP11; Lundberg et al7), citrullinated fibrinogen (cit-Fbg; Snir et al8) and citrullinated vimentin (cit-Vim; Wegner et al4)—see references for peptide sequences; detailed methodologies available on request. Where available, auto-antibodies against non-citrullinated (native) versions of the same peptides were also assayed as negative controls.

Of the 81 eligible patients, 75 (93%) had definitive outcome diagnoses by the end of the study period and were included in the study. Twenty-nine (39%) of these evolved into RA over the study period (equivalent to the presumed pretest probability of RA), comparable with the proportion observed in other UA studies.1 9 10 A positive CCP2 assay was highly specific for an outcome of RA, carrying a positive predictive value of 0.93, but only 14 out of 29 sera from UA patients destined for a diagnosis of RA tested positive (sensitivity 0.48), and the negative predictive value was 0.75 (table 1). No single alternative assay had superior diagnostic performance to anti-CCP2 with respect to an outcome of RA among newly presenting UA patients (table 1 and figure 1). Careful scrutiny of figure 1 also illustrates that the consideration of assays in combination, employing permutations both with and without anti-CCP2, did not add value to CCP2 testing alone in predicting RA—an observation that was reinforced by extensive further calculations of diagnostic statistics (not shown). Hierarchical clustering of the data revealed that the assay whose results bore closest correlation to those of the CCP2 test among the sera studied was that for anti-cit-Fbg. Three RF assays clustered together as a distinct group, away from those for anti-citrullinated peptide autoantibodies (ACPAs), being generally characterised by higher proportions of false positives (figure 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Heat map representing results of 15 ELISA assays (columns labelled at base) among 75 sera from early undifferentiated arthritis patients (rows). Colour bar gives interpretation of colours; undet: undetermined, indicating that the autoantibodies against the non-citrullinated form of the specified peptide (negative control) were detected in a given sample and the overall result was discarded. Samples are grouped from top to bottom according to diagnostic outcome, with those evolving into rheumatoid arthritis (RA) towards the bottom. 1, osteoarthritis; 2, non-inflammatory arthritides; 3, non-RA inflammatory arthritides; 4, connective tissue disease. Overlying dendrogram depicts relative strength of correlations (Pearson; standard linkage) between assay outputs across the sample set.

In conclusion, the CCP2 test remains an invaluable tool in the assessment of UA. Nevertheless, the identification of novel biomarkers for the diagnosis of autoantibody-negative RA must now be a priority. Our study is the first to shed light on the role of potentially disease-relevant ACPAs as discriminatory tools alongside the CCP2 assay in the highly clinically relevant setting of early UA.