Can we accurately predict the development of rheumatoid arthritis in the preclinical phase?


Rheumatoid arthritis (RA) is an autoimmune disease characterized by joint destruction, which leads to functional decline and disability as well as increased mortality (1–4). A growing body of evidence indicates that joint destruction and functional decline are improved by early therapeutic intervention (5, 6). In addition, the risk of mortality is decreased in patients with RA who present early in the disease course, strongly suggesting that earlier diagnosis and treatment may have substantial long-term benefits (7). Because asymptomatic synovitis has been identified prior to major clinical manifestations of RA (8, 9), prevention of disability in RA might be even more successful if interruption of the process underlying joint destruction were carried out in the early phase of the disease or even before the disease is clinically evident, rather than when destructive synovitis is well established.

The article by Rantapää-Dahlqvist et al published in this issue of Arthritis & Rheumatism (10) demonstrates that antibodies to cyclic citrullinated peptide (CCP) as well as rheumatoid factor (RF) can be identified in stored serum from individuals who later develop clinically apparent RA. The authors conclude that such autoantibodies may be predictive of the future development of RA in asymptomatic individuals. This article raises the question of whether we can accurately identify individuals at risk for the development of RA during the preclinical phase of disease. If such an endeavor could be accomplished, then primary prevention or, at a minimum, very early intervention strategies could be implemented.

Over the last 60 years, several autoantibody systems have been proposed as diagnostic and/or prognostic tools for RA, but only the presence of RF is included as a serologic criterion in the American College of Rheumatology (formerly, the American Rheumatism Association) classification criteria for RA (11). However, the anti-CCP autoantibody system has received a great deal of study recently and may have an important role in providing supportive evidence for a diagnosis of RA. The anti-CCP autoantibody system includes a group of autoantibodies with shared reactivity to proteins containing arginine residues that have been modified to citrulline. These autoantibodies include antiperinuclear factor (APF), antikeratin antibody (AKA), antifilaggrin antibody (AFA), and anti-CCP antibodies. The APF autoantibody was first described in 1964 (12, 13). This antibody binds to perinuclear keratohyalin granules in the cytoplasm of buccal mucosa cells. The sensitivity of APF in RA ranges from 49% to 87%, but the specificity is only moderate, since the test result is positive in sera from individuals with other autoimmune diseases, including systemic sclerosis, autoimmune thyroiditis, and systemic lupus erythematosus (SLE) (12). AKA, an autoantibody reactive with a similar antigen, was first described in 1979 (14). AKA binds to the stratum corneum of rat esophagus and has a sensitivity of ∼59% and a specificity of ∼100% (12, 15). APF and AKA are measured by indirect immunofluorescence, which makes testing and interlaboratory comparisons of the results difficult. In part because of these problems, neither test has gained widespread clinical application (16).

More recently, the epithelial antigen recognized by AKA (17, 18) and APF (19) has been determined to be the protein filaggrin and, due to the apparent identity between APF and AKA, it has been proposed that these autoantibodies be renamed “antifilaggrin autoantibodies” (19). Filaggrin, or filament-aggregating protein, is formed during late differentiation of mammalian epithelial cells (20) and is involved in the regulation of cytokeratin intermediate filament aggregation. Filaggrin is formed as a precursor protein, profilaggrin, with 10–12 filaggrin repeats that are released by proteolytic cleavage during cell differentiation. Simultaneously, a number of the arginine residues in filaggrin are converted to citrulline as a posttranslational modification mediated by the enzyme peptidyl arginine deiminase. Importantly, the conversion of arginine to citrulline is required for AFA recognition of the filaggrin peptides (20, 21).

Filaggrin has only been detected in squamous epithelial cells (20). However, AFAs have been identified in the synovial membrane and are synthesized by plasma cells in the rheumatoid pannus (22). This has led to the hypothesis that autoantibodies against citrullinated filaggrin are cross-reactive with a citrullinated peptide within the synovium (20). Consistent with this possibility, target epitopes identified as citrullinated α- and β-chains of fibrin exist in the rheumatoid synovial membrane (23). Since citrullinated filaggrin does not exist in the joint, and is therefore unlikely to be the putative RA-specific autoantigen, van Venrooij and Pruijn recently proposed that this autoantibody system be called anti–citrullinated protein antibodies (24). Those authors hypothesized that citrullination of proteins within the inflamed rheumatoid joint may occur as a result of apoptosis (24).

Paralleling investigations to identify the target autoantigens of this system have been efforts to develop robust and clinically useful serologic tests. Paimela et al (25) used filaggrin purified from human skin as the antigen in an enzyme-linked immunosorbent assay (ELISA) to evaluate serum from patients with established RA; the reported sensitivity of this assay was only 54%. Schellekens et al (20, 26) designed a synthetic cyclic citrullinated peptide with deiminated arginines for use as the antigen in an ELISA for anti-CCP autoantibodies. This assay demonstrated an impressive specificity of 98% in sera from patients with established RA and 96% in sera from patients with early RA. The sensitivity of this anti-CCP ELISA was 68% and 48% in sera from patients with established RA and early RA, respectively (26). RF is the gold standard test for serologic support of a diagnosis of RA but has a specificity of only 90–95%, depending upon the age of the individual tested (12). Therefore, detection of anti-CCP antibodies has the potential to better distinguish early RA from other non-RA inflammatory polyarthritides. Although the sensitivity of anti-CCP antibodies is less than that of RF (68% versus 80% [27]), anti-CCP antibodies have been found in the sera of patients with RF-negative RA (20, 26). Consequently, one clinical use of anti-CCP antibody testing could be in combination with RF testing in order to improve the sensitivity of the serologic diagnosis of RA. Additionally, the anti-CCP autoantibody system has been associated with progression to erosive RA and a more severe disease course (26, 28, 29). Therefore, this autoantibody system may contribute to clinical decision-making regarding therapy for RA.

Importantly, the results reported by Rantapää-Dahlqvist et al (10) suggest that detection of anti-CCP antibodies may also be of value in identifying individuals with RA prior to the onset of symptoms and destructive synovitis. Consistent with this hypothesis, various autoantibodies from the anti-CCP autoantibody system have been observed preclinically. Aho and colleagues (12, 15, 30) identified preclinical AKA and APF autoantibodies in 20% of stored serum samples from 69 patients who later developed RA. The same group of investigators also showed that the presence of AKA or AFA increased the risk of seropositive RA development 5-fold in individuals with preclinically positive RF (31).

In the studies reported by Rantapää-Dahlqvist and colleagues, stored serum samples were obtained from individuals with RA that had been previously collected as part of 2 cohorts, the Northern Sweden Health and Disease Study cohort and the Maternity cohort of Northern Sweden. A total of 83 individuals with RA (85% female) as well as matching controls were selected for analysis of samples that had been collected a median of 2.5 years prior to presentation with RA symptoms (i.e., the preclinical phase). Anti-CCP antibodies and RF of all isotypes were detected during the preclinical phase of RA; the prevalence of anti-CCP antibodies, IgG-RF, IgM-RF, and IgA-RF was 33.7%, 16.9%, 19.3%, and 33.7%, respectively. Additionally, 21 samples collected a median of 10.9 years prior to presentation with RA symptoms were tested, but without controls for comparison. The longest intervals predating the onset of RA symptoms were 9 years, 22 years, and 19 years for anti-CCP antibodies, IgA-RF, and IgG/IgM-RF, respectively. Anti-CCP antibodies and IgA-RF had statistically significant odds ratios for the future development of RA in a multivariate conditional logistic regression model, 28.9 (4.3–192.6) and 11.4 (1.3–98.0), respectively. Anti-CCP antibodies had a positive predictive value (PPV) of 82% for the development of RA in this cohort, but the PPV was only 16% when calculated for the general population using an estimated prevalence of RA equal to 1%.

While these results are striking, there are several limitations to this study. First, it should be noted that some serum samples were collected during pregnancy, possibly leading to an underestimation of the prevalence of autoantibodies in preclinical RA, given that pregnancy may be protective against the development of new-onset RA. Second, the authors state that seropositivity for anti-CCP persisted in subsequent preclinical samples and that the titer increased until symptoms of RA were manifested. This conclusion cannot be fully evaluated, since there were only 15 subjects from whom 2 preclinical samples were available for testing, and the results of these tests were not reported separately. Furthermore, the comparison appears to be between the overall preclinical mean anti-CCP antibody level and the mean anti-CCP level measured during early RA. Thus, a limitation of this study is that the cohorts lacked the serial serologic specimens necessary for a detailed analysis of the stability and level of preclinical autoantibody seropositivity over time. Third, the limits of the anti-CCP autoantibody system as a screening test in the general population should also be noted. The authors estimate the PPV of anti-CCP antibodies for the future development of RA in the general population to be only 16%. Therefore, in order to use the anti-CCP antibody test to predict the future development of RA, it should be combined with an evaluation for other risk factors for RA, such as genetic predisposition, family history, and the presence of seropositivity for other autoantibodies.

This study raises the important clinical issue concerning the identification of a stage of preclinical autoimmunity in RA. An increased understanding of the evolution of preclinical autoimmunity in autoimmune diseases such as type 1 diabetes mellitus (DM) and SLE has recently emerged (32). The current concept is that autoimmune diseases such as type 1 DM and SLE may exhibit 3 phases during their development. The first phase is defined as a state of susceptibility resulting from genetic risk factors, either alone or in combination, that predispose to the loss of self tolerance. The second phase is the transition in an individual with genetic risk to a state of autoreactivity with autoantibody production that is not yet associated with clinically apparent disease that would meet diagnostic or classification criteria. This transition from risk to initial autoimmunity occurs because of the introduction of an environmental factor or perhaps because of the stochastic nature of the immune response. This preclinical autoimmune phase is associated with either the development of one or a small number of autoantibodies and precedes the commonly observed substantial epitope spreading that is associated with major tissue injury. The third phase is the development of clinically apparent disease, which is characterized by extensive immune-mediated tissue destruction.

The presence of these 3 disease phases is well established for type 1 DM (33). During the first phase of disease, susceptibility to type 1 DM is conferred by genes in the HLA region and to a smaller extent by genes on other chromosomes (33). Then, genetically predisposed individuals progress to the second phase of disease, with immune-mediated destruction of the pancreatic islet beta cells that normally secrete insulin (33, 34). In type 1 DM, autoantibodies against islet cells typically appear by age 5, and the development of multiple autoantibodies is strongly predictive of eventual advancement to clinical disease (33–36). At the point at which an individual progresses to the third phase of disease, displaying clinically apparent diabetes, there has already been a progressive decline in beta cell function, and the majority of beta cells have been damaged or destroyed (33, 34). Although this last phase is the one in which the vast majority of patients seek medical care, prevention strategies must focus on identifying patients during earlier phases of disease.

There are currently large prospective studies attempting to do just that. For example, the Diabetes and Autoimmunity Study in the Young (DAISY) (37) is a prospective study designed to identify individuals at high risk for the development of type 1 DM so that prophylactic intervention can be applied during the preclinical phase. DAISY investigators have been serially evaluating over time the relatives of probands with type 1 DM as well as a newborn general population cohort since 1994 in order to quantify the risk of developing type 1 DM. Staging consists of genetic screening for HLA haplotypes conveying increased disease risk and immunologic screening for autoantibodies reactive with islet cells and insulin. This staging can predict risk for the development of type 1 DM with remarkably accurate precision (37, 38). Moreover, some at-risk individuals are currently being studied in a large multicenter primary prevention trial called the Diabetes Prevention Trial–Type 1 Diabetes (35) whose goal is to prevent or delay the development of overt diabetes. Interventions used by the Diabetes Prevention Trial have included the administration of parenteral or oral insulin. The results of parenteral insulin therapy were disappointing, despite several pilot studies indicating successful delay of disease development with insulin therapy (38–40). However, oral insulin trials are still in progress. A similar study, the European Nicotinamide Diabetes Intervention Trial (ENDIT), is also addressing whether nicotinamide will reduce the rate of progression to DM in at-risk relatives (41). In addition, treatment with “tolerizing,” nondepleting anti-CD3 antibodies has been successful in delaying disease onset in animal models, and this intervention has been proposed as a potential preventive measure for the clinical onset of hyperglycemia (34, 42).

Important to studies of the rheumatic diseases, Arbuckle et al (43) have published results of a retrospective case–control study in which 55% of 133 SLE patients were found to have anti–double-stranded DNA (anti-dsDNA) autoantibodies in stored serum samples up to 9.3 years before the diagnosis of clinical SLE (43). The mean onset of autoantibody positivity was 2.7 years prior to disease diagnosis. Applying the hypothesis of 3 distinct phases of autoimmune disease to SLE, it is likely that an individual with an underlying genetic predisposition for the development of SLE is exposed to some triggering factor, such as estrogen, ultraviolet light, or an infectious agent, that stimulates the production of anti-dsDNA autoantibodies. Eventually, with epitope spreading and time, this individual develops clinical symptoms of SLE.

The data presented by Rantapää-Dahlqvist et al provide an increased understanding of the preclinical phase of RA. Individuals carrying the shared epitope of HLA–DR alleles have a well-defined increase in relative risk for the development of RA (44, 45). Such predisposed individuals may at some point encounter an environmental factor, such as an infection, hormonal influence, or tobacco exposure, which leads to the production of preclinical autoantibodies, such as anti-CCP or RF (Figure 1). Eventually, in a subset of at-risk individuals, clinical symptoms develop, and the individual is diagnosed as having RA. It remains to be determined whether the early appearance of RF and anti-CCP autoantibodies in RA represents a primary loss of tolerance to self antigens or a secondary response to subclinical tissue injury with exposure to novel or abundant autoantigens.

Figure 1.

Three phases of rheumatoid arthritis (RA) and means of detecting preclinical disease that parallel ongoing studies in other autoimmune diseases, such as type 1 diabetes mellitus and systemic lupus erythematosus. RF = rheumatoid factor; anti-CCP = anti–cyclic citrullinated peptide.

While the data presented by Rantapää-Dahlqvist et al provide insight into the preclinical phase of RA, long-term prospective studies similar to those in type 1 DM are needed in order to more definitively characterize the evolution of this autoimmune disease. It is possible that with such a study, the additive effect of individual contributors such as genetic predisposition, environmental exposures, and possible loss of self tolerance can be more fully understood. Such a study would also allow for the more accurate identification of the onset of clinical RA as well as the stability of autoantibodies over time so that the predictive value of such autoantibodies could be more precisely defined. Nonetheless, this retrospective study and others that identify the preclinical presence of autoantibodies in RA may provide a major breakthrough in our understanding of this disease and eventually in its clinical management. Further strategies designed to identify individuals at high risk for the development of RA may allow for the introduction of prevention or very early treatment strategies with the potential to greatly affect the morbidity and mortality of this very debilitating disease.