Learning about the natural history of rheumatoid arthritis development through prospective study of subjects at high risk of rheumatoid arthritis–related autoimmunity


  • Kevin D. Deane

    Corresponding author
    1. University of Colorado School of Medicine, Aurora
    • Division of Rheumatology, University of Colorado School of Medicine, 1775 Aurora Court, Mail Stop B-115, Aurora, CO 80045
    Search for more papers by this author
    • Dr. Deane is included on a patent application that has been filed for the use of biomarkers to predict clinically actionable events in rheumatoid arthritis.

Multiple studies have demonstrated abnormalities of rheumatoid arthritis (RA)–related autoantibodies and biomarkers of inflammation prior to the presence of the signs and symptoms of inflammatory arthritis (1–18). Many of these studies were performed using retrospectively assembled samples from subjects who fortuitously had biologic specimens available from before their diagnosis of RA. Therefore, our understanding of the precise temporal relationship between biomarker abnormalities and the onset of the signs and symptoms of inflammatory arthritis is limited. However, despite these limitations, overall these studies have had a profound impact on the understanding of the early development of RA. Furthermore, these studies, in conjunction with studies that have established genetic risk factors for RA (such as the shared epitope [19]), support the concept that RA development occurs in several phases, as depicted in Figure 1.

Figure 1.

Phases of development of rheumatoid arthritis (RA). In this model of RA development, disease begins with genetic risk (phase 1), followed by asymptomatic inflammation and autoimmunity (phase 2), with eventual progression to symptomatic inflammatory arthritis (IA) (phase 3) that may progress to classifiable RA (phase 4). Currently, the asymptomatic phases of disease development (phases 1 and 2) can be termed the “preclinical” period of development, although this nomenclature is in evolution. Not all subjects who are at risk of developing RA progress through all of these phases, and some subjects may have resolution of inflammation, autoimmunity, and even inflammatory arthritis (demonstrated by the bidirectional arrows). The mechanisms of transition between these phases are not well understood but likely involve complex relationships between genetic and environmental factors (which may differ between phases), age-related and stochastic immunologic changes, as well as psychosocial factors, access to health care, and response to therapy.

In this model of RA development, genetic risk of disease (phase 1) is followed by a period of asymptomatic autoimmunity and inflammation (phase 2), marked by detectable abnormalities of autoantibodies and inflammatory markers and, presumably, the absence of signs and symptoms of inflammatory arthritis. Over time, the presence of signs and symptoms of unclassified inflammatory arthritis develop (phase 3), with eventual progression of disease to classifiable RA (phase 4) based on established classification schemes, such as the American College of Rheumatology 1987 revised criteria (20). This model can also include a phase 5, which encompasses the behavior of autoimmunity and inflammation after the onset of symptomatic inflammatory arthritis, including remissions, exacerbations, response to specific therapies, evolving biomarkers, and extraarticular disease.

Importantly, this model of RA development best applies to autoantibody-positive disease because of readily identifiable biomarkers. Additionally, not all subjects who are at risk of future RA, and not even all of those who have developed detectable autoimmunity, will progress through all phases of disease. Furthermore, the findings of some studies demonstrating that levels of RA-related autoantibodies are elevated only after inflammatory arthritis has become apparent indicate that, at least in some cases, detectable circulating autoimmunity may not precede inflammatory arthritis, at least by biomarker measures available at the time of the initial presentation of inflammatory arthritis (21).

The term “preclinical RA” is currently in common use to describe the earliest “asymptomatic” phases of RA development (phases 1 and 2); however, the nomenclature for this period of evolution of RA is itself in evolution. In particular, there is debate about the appropriateness of the term “preclinical RA” to describe the phases of RA that include genetic risk and asymptomatic autoimmunity, when in fact subjects in these phases may never develop symptomatic inflammatory arthritis defined as RA, and as such, it may be inappropriate to label them as having “preclinical RA.” These issues are important, and efforts are currently under way to standardize the nomenclature regarding the development of RA (22). However, nomenclature issues aside, this model of RA development strongly suggests that the immunologic events important to the initial development of autoimmunity occur in the preclinical period of RA, and that it is crucial to study this early period in detail to understand how RA initially develops (23, 24).

But what is the best way to study the earliest phases of RA development? One approach would be to continue biomarker analyses of biologic specimens fortuitously collected prior to a diagnosis of RA. Despite limitations, such studies have yielded valuable information about preclinical RA. However, because of the highly complex nature of the genetic, environmental, and immune system interactions that drive the initial generation of RA-related autoimmunity, such studies are unlikely to provide the data necessary for a comprehensive understanding of early RA development. Therefore, optimal investigations of the preclinical period of RA development should be performed through prospectively conducted natural history studies, where researchers can perform real-time detailed investigations of the development and evolution of RA-related autoimmunity. Furthermore, while not imperative, from a practical standpoint, such natural history studies may have a greater likelihood of success if they target subjects at high risk of RA so that outcomes of autoimmunity are enriched.

In this issue of Arthritis & Rheumatism, El-Gabalawy and colleagues present results from just such a natural history study, using a unique Canadian cohort of North American Native subjects who are first-degree relatives of probands with RA (25). These first-degree relatives did not have inflammatory arthritis, based on joint examination at the time of study, and were considered to be at high risk of RA based on their first-degree relative status as well as the fact that they were recruited from a population of Cree and Ojibway North American Natives with an ∼3% prevalence of RA (26). Therefore, these subjects provide a unique opportunity to identify individuals in real-time who may be in the preclinical period of RA development.

El-Gabalawy et al tested serum samples for levels of RA-related autoantibodies, 42 cytokines/chemokines using a multiplex assay, and high-sensitivity C-reactive protein (hsCRP) in a cross-sectional manner in these first-degree relatives, the probands with RA, and in 2 healthy control groups of North American Natives and Caucasians. The controls were from a similar geographic region as the first-degree relatives and were screened to ensure that they did not have a known personal or family history of rheumatic disease. The authors found that a high proportion of probands and first-degree relatives were positive for RF and anti–cyclic citrullinated peptide (anti-CCP) antibodies (anti–CCP-2 enzyme-linked immunosorbent assay [ELISA]; Inova Diagnostics), even when using a cutoff for anti-CCP positivity of ≥40 units. (The manufacturer's suggested cutoff is ≥20 units.) Using this higher cutoff, ∼88% of the RA patients were RF positive, and 81% were anti-CCP positive. Of the North American Native first-degree relatives, ∼34% were positive for RF, and ∼9% were positive for anti-CCP.

First-degree relatives also had differences in levels of multiple cytokines/chemokines compared to controls; in subsequent discrimination analyses, these differences allowed for significant separation of North American Native first-degree relatives from both North American Native controls and Caucasian controls. Furthermore, an elevated monocyte chemotactic protein 1 (MCP-1) level was the strongest predictor of group membership (North American Native first-degree relatives versus controls). (Elevations of MCP-1 levels were similar even in followup ELISA testing.) Of note, the levels of most of the 42 cytokines/chemokines studied were elevated in first-degree relatives compared to controls; however, several were lower, including interferon-γ–inducible 10-kd protein and CD40 ligand. Additionally, levels of multiple cytokines/chemokines and hsCRP were abnormal (typically elevated) even in autoantibody-negative first-degree relatives when compared to controls. The authors conclude that abnormal levels of multiple cytokines/chemokines and hsCRP, even in the absence of RA-related autoantibody positivity, may be part of the risk profile for developing RA, and longitudinal studies of these first-degree relatives are underway.

The findings of autoimmunity and inflammation in this cohort of North American Native first-degree relatives should be of great interest to researchers in the field of preclinical RA, although there are several questions that immediately arise from this study, the chief of which relate to the relationship between these cytokine/chemokine biomarker abnormalities and the risk of RA-related autoimmunity and future clinically apparent inflammatory arthritis/RA. Do these findings, especially in the autoantibody-negative first-degree relatives, suggest that autoimmunity arises out of a general background of inflammation? In particular, since the authors found that MCP-1 most strongly distinguished first-degree relatives from controls, do elevations of this biomarker tell us something specific about the pathogenesis of RA, or is the strength of the discriminate power of MCP-1 levels due to the vagaries of this study's biomarker assays and statistical analyses?

Elevations of MCP-1 levels prior to the onset of symptomatic RA have been found in other studies, strengthening the possibility that this chemokine may play an important, specific role in RA pathogenesis (10), but what is that role? MCP-1 is generated by many cells, including monocytes, macrophages, and osteoclasts, and it has several functions, including acting as a chemoattractant for monocytes, basophils, T cells, and dendritic cells. We do not know, however, which of these cells or actions of MCP-1 is most relevant in the first-degree relatives included in the study by El-Gabalawy et al. And what about the abnormalities of the multiple other biomarkers in the first-degree relatives compared to controls? Do these indicate specific mechanisms important to the pathogenesis of RA, or do they merely reflect a general tendency toward inflammation in these at-risk individuals?

It is tempting to defer any conclusions about the role of cytokine/chemokine, hsCRP, and autoantibody abnormalities in the pathogenesis of RA until adequate followup is performed to determine the relationship between these biomarkers and outcomes of incident clinically apparent RA. This is because, as some argue, it is difficult to label a state of inflammation or autoimmunity as “preclinical RA” if an at-risk subject does not develop inflammatory arthritis or classifiable RA. However, the questions “What leads to the initial development of RA-related autoimmunity?” and “What predicts the onset of clinically apparent inflammatory arthritis once autoimmunity has developed?” should be viewed as quite different, because it is highly likely that the mechanisms that lead to the initial generation of autoimmunity are separate from those that lead from autoimmunity to clinically apparent disease. It will be of great interest to see the results of longitudinal followup of the subjects included in the study by El-Gabalawy and colleagues and to learn what factors, biomarker or otherwise, are associated with the development and progression, or even resolution, of autoimmunity as well as clinically apparent inflammatory arthritis/RA, keeping in mind that studying the evolution of autoimmunity itself even in the absence of clinically apparent inflammatory arthritis may be extremely valuable for our understanding of the pathogenesis of RA.

Other questions arising from this study relate to the source of the cytokine/chemokine abnormalities in the first-degree relatives. Are these circulating cytokine/chemokine abnormalities due to tissue injury at some specific anatomic site? The first-degree relatives apparently did not have inflammatory arthritis on clinical evaluation at the time of blood sampling, but could they have had subtle synovial inflammation that was responsible for these biomarker abnormalities? A recent study by van de Sande and colleagues involved a small group of subjects from Northern Europe who were RF and/or anti–citrullinated protein antibody positive in the absence of clinically apparent inflammatory arthritis. Synovial biopsy specimens obtained from the knees of those subjects showed no significant inflammation (27). If those negative synovial findings were not due to sampling error, then they support the notion that circulating markers of autoimmunity and inflammation in the North American Native first-degree relatives in the study by El-Gabalawy et al may be present without synovitis.

However, if the joints of the North American Native first-degree relatives are not producing biomarker abnormalities, is some other site of inflammation driving these biomarker elevations? Smoking is known to drive systemic inflammation, perhaps due to tissue injury in the lung (28). A high proportion of first-degree relatives (65%) reported being a current smoker at the time of the study, but this was not significantly different from the proportion of North American Native controls who were current smokers (62%), so it seems unlikely that smoking-related inflammation alone could explain these differences. It could be, as the authors suggest, that the first-degree relatives have genetically related “overactive” inflammatory pathways or perhaps “underactive” immunoregulatory pathways that contribute to the risk of RA. All of these issues need to be explored in future studies.

While offering intriguing data regarding the potential role of inflammation in the risk of developing RA-related autoimmunity, the study by El-Gabalawy et al also faced several challenges. First, there are several issues regarding multiplex testing of biomarkers, including controversies as to what biologic specimen yields optimal assessment of cytokines/chemokines (serum versus plasma), statistical issues pertaining to multiple testing, variability of results depending on the processing/handling of biologic specimens prior to testing, variability of results in repeated testing of the same samples or in testing of serial samples collected in a longitudinal manner, and the question of how to translate biomarker abnormalities identified in groups to an individual's risk of RA-related autoimmunity (29). Also, recent findings by Todd and colleagues have shown that elevations of RF within samples may falsely amplify results from multiplex assays unless special efforts are taken to limit the effect of RF, including the use of RF-blocking or RF-depleting agents (30). El-Gabalawy and colleagues have minimized some of these concerns regarding multiplex testing by collecting and processing samples in a uniform manner and by overlapping sample testing to ensure consistency. However, they did not use RF-depleting strategies, and this may be viewed by some as a major limitation, although the observation that levels of multiple cytokines/chemokines were elevated even in seronegative first-degree relatives when compared to controls suggests that there are significant elevations of the levels of these biomarkers even in the absence of RF. Of note, these issues with multiplex testing are complex, and researchers in the field of preclinical RA will need to come to a consensus regarding the methodologies for such testing going forward.

A second challenge is that the first-degree relatives were derived from a population that has a high prevalence of RA as well as a high prevalence of risk factors for RA (such as the presence of the shared epitope and smoking); as such, some may question the applicability of these findings to other populations. However, rather than being a liability, it may be that studying these North American Native first-degree relatives who have an increased risk of RA will lead to key insights into the evolution of RA that would be difficult to identify in lower-risk populations.

Finally, factors not evaluated by El-Gabalawy et al could drive the elevations of biomarker levels in the first-degree relatives. The authors did adjust their analyses to some extent to account for the possibility that a few families with large numbers of first-degree relatives could drive their findings, as well as for factors, including smoking, that could influence biomarker level differences between controls. The authors' careful construction of two control groups (North American Native and Caucasian) that were from the same geographic area as the first-degree relatives and probands also may help adjust for many unknown environmental factors pertinent to the region of study that may influence autoimmunity and inflammation. However, as the authors point out, factors such as obesity and mucosal inflammation or genetic factors related to specific pathways of inflammation may be driving autoimmunity and inflammation in these first-degree relatives, and these will need to be evaluated in future studies.

In summary, prospective natural history studies of the preclinical period of RA development are crucial to our understanding of the complex factors related to the development of RA. The cross-sectional data from such a natural history study presented by El-Gabalawy and colleagues suggest that there may be biomarker evidence of risk for RA-related autoimmunity even beyond autoantibodies. However, substantial further prospective work needs to be done to understand the biologic relevance of these findings, especially as they relate to specific pathways of disease development, incident autoimmunity, and clinically apparent inflammatory arthritis/RA. The rheumatology community should look forward to the results of longitudinal analyses from this group, as well as forthcoming data from other studies of the natural history of RA.


Dr. Deane drafted the article, revised it critically for important intellectual content, and approved the final version to be published.