RF, autoantibodies that react with the Fc portion of IgG, have been on the rheumatologic stage for longer than 60 years and are still the only established serologic marker of RA 8. RF can be detected in 60–80% of RA patients and are thus fairly sensitive in diagnosing RA. However, the specificity of RF as a diagnostic indicator for RA is substantially inferior to that of autoantibodies used for the diagnosis of other rheumatic autoimmune diseases. Moreover, RF occur less frequently or are of low titer in early disease stages when a clear diagnosis is often not yet possible. Apart from RA, high-titer RF are also present in the majority of patients with primary Sjögren's syndrome and can be found in lower proportions (and usually also in lower titers) in all other rheumatic autoimmune diseases, as well as in a number of non-autoimmune conditions such as osteoarthritis or chronic infections 68.

The disease specificity of RF for rheumatic diseases is dependent on the concentration (titer) used as a cut-off and is considerably higher at elevated titers, albeit at the expense of sensitivity (Table 3). Furthermore, RF can be a prognostic indicator because they are related to increased severity of RA, such as the presence of erosive disease, more rapid disease progression, and worse outcome. Although RF may be involved in RA pathogenesis, their role is still not entirely clear. Interestingly, RF do not occur in experimental models of RA. Only MRL/lpr mice that suffer from lupus-like disease with anti-dsDNA and anti-Sm antibodies express RF 11. In addition, these mice present with arthritis, which may become erosive, as well as other overlapping disease features 12.

Antikeratin antibodies (AKA) were first described in 1979 13 and have been repeatedly shown to be highly specific for RA 914. The antigen targeted by AKA was identified in 1993 as the intermediate filament-aggregating protein filaggrin, which is expressed exclusively in keratinizing epithelial cells 15. Subsequent studies showed that the AKA/antifilaggrin antibodies recognized epitopes that contained the amino acid citrulline, which is generated post-translationally from arginine by the enzyme peptidylarginine deiminase 1617. These antibodies are therefore now generally called anticitrullinated peptide or anticitrulline antibodies. However, these autoantibodies recognize citrulline-containing peptides, which are not necessarily derived from filaggrin. Nevertheless, the disease specificity of anticitrulline antibodies for RA is high (96%), and the reported sensitivity of RA patients for anticitrulline autoantibodies is 65%, which is comparable with that of RF. The diagnostic value of anticitrulline antibodies is further substantiated by their presence in sera from patients with early RA, although they seem to occur mainly in RF-positive sera 1819.

As filaggrin is an epidermal protein, it presumably does not represent the actual antigen of the anticitrulline autoimmune response, which remains to be identified. A promising candidate antigen for this autoimmune response is fibrin, which is present in the synovium of RA patients. It has recently been shown that antifilaggrin and anticitrulline antibodies recognize citrullinated fibrin 22. It is thus conceivable that locally produced antibodies against citrulli-nated target structures may contribute to the inflammatory and destructive processes in the rheumatoid joint 20. Interestingly, among experimental forms of arthritis, anti-citrulline antibodies have not been observed (Table 2).

Anti-A2/anti-RA33 antibodies are directed to the heterogeneous nuclear ribonucleoprotein A2 (hnRNP-A2), a nuclear protein that is involved in mRNA splicing and transport 2123. The antibodies occur in approximately one-third of RA patients but can also be detected in 20–30% of patients with SLE and in up to 40% of patients with the rare overlap syndrome mixed connective tissue disease (MCTD). The sensitivity of RA patients for anti-A2/anti-RA33 autoantibodies is therefore low (~40%). Nevertheless, in a representative cohort of patients with various rheumatic diseases including autoimmune and non-autoimmune arthritides, the specificity of anti-A2/anti-RA33 antibodies for RA was approximately 90% 2425. However, if a diagnosis of SLE and MCTD (or MCTD alone) is excluded or if there is an absence of autoantibodies associated with SLE (such as anti-DNA, anti-Sm, and anti-U1 RNP antibodies), the specificity of anti-RA33 antibodies for RA can be as high as 96% 25.

Importantly, other arthritides such as osteoarthritis, reactive arthritis, and psoriatic arthropathy are usually anti-A2/anti-RA33 negative. Moreover, these antibodies may be present in early disease stages, particularly in RF-negative sera; although, as is true with other autoantibodies, at slightly lower frequency than in established disease 1426. The antigen targeted is more or less ubiquitously expressed, although expression levels may greatly differ between tissues. Recent data from our laboratory indicate that hnRNP-A2/RA33 is overexpressed in synovial membranes of RA patients, where it might form a target for autoreactive B and T cells 10. Of particular interest, anti-A2/anti-RA33 antibodies are present in early stages of disease in MRL/lpr mice, where they precede anti-dsDNA antibody formation, and in tumor necrosis factor-α transgenic mice (which develop severe erosive arthritis similar to RA) 710. These data suggest that autoimmunity against hnRNP-A2/RA33 may be involved in the pathophysiology of these models, and possibly in human disease.

Antinuclear antibodies are found in approximately 50% of RA patients. With the exception of anti-RA33 and antibodies to Epstein-Barr virus nuclear antigen, specific antinuclear antibody subsets among patients with connective tissue diseases are very rare. Anti-Ro may occasionally be present in RA patients, especially if they suffer from secondary Sjögren's syndrome. Although antibodies to double-stranded DNA are also very rare among RA patients, it is of interest that in the course of infliximab therapy both an increase (or de novo occurrence) of antinuclear antibodies and the development of antibodies to double-stranded DNA, particularly of the IgM type, have been observed 2728. Nevertheless, the presence of drug-induced lupus-like syndrome is rare (approximately 0.5% of infliximab-treated patients) and is no more common than that observed with several other disease-modifying antirheumatic drugs. Furthermore, it usually does not involve major organs 27, and is reversible upon mild therapeutic measures or cessation of therapy.

Anti-Sa antibodies are directed to a 50 kDa protein of unknown structure and function that has been isolated from human tissues (spleen, placenta, rheumatoid syn-ovium). Anti-Sa autoantibodies are detected in approximately 40% of patients with established RA but less often in patients with early disease 2930. The reported specificity of anti-Sa antibodies for RA range between 92 and 98%, which compares favorably with marker antibodies for other autoimmune diseases. As suggested by Hayem et al. 31, who found the incidence of these antibodies to be significantly increased in RA patients with severe destructive disease, determination of anti-Sa antibodies may be of prognostic benefit. In a recent prospective study in patients with recent onset synovitis, anti-Sa had the highest specificity and prognostic value of all autoantibodies investigated 18.

Autoantibodies to a ubiquitously expressed 68 kDa glyco-protein were described in 1995 by Blass et al. 32. The target of anti-p68 antibodies was recently identified as the chaperone, or stress protein immunoglobulin heavy-chain binding protein (BiP). It is also known as glucose-regulated protein of 78 kDa (grp78), a member of the 70 kDa heat-shock protein family, and is localized in the endoplasmic reticulum 33. Anti-BiP autoantibodies are found in the sera of more than 60% of RA patients 32. The specificity of anti-BiP antibodies for RA has been reported as 96%, making these antibodies promising candidates for the diagnosis of RA. Furthermore, it will be important to learn whether anti-BiP antibodies develop early in the disease process. Similar to hnRNP-A2/RA33 and fibrin, BiP has been shown to be highly expressed in synovial tissue. Because it seems to form a target for autoreactive T cells of RA patients, BiP may be one of the antigens driving the pathologic autoimmune process in RA 33.