Abstract
Early rheumatoid arthritis (RA) and very early RA are major targets of research and clinical practice. Remission has become a realistic goal in the management of RA, particularly in early disease. The 2010 American College of Rheumatology/European League Against Rheumatism (ACR/EULAR) RA classification criteria, the EULAR treatment recommendations for RA, and the EULAR recommendations for the management of early arthritis focus on early disease and translate the knowledge related to early RA into classification and management. Nevertheless, there is a need for further improvement and progress. Results from 6 recent studies are summarized, evaluating the performance of the 2010 ACR/EULAR RA classification criteria. The data show a significant risk of misclassification, and highlight that overdiagnosis and underdiagnosis may become important issues if the criteria recommend synthetic and biological disease-modifying antirheumatic drugs. Therefore, some considerations are presented on how the current problems and limitations could be overcome in clinical practice and future research. A consensus is needed to better define the early phase of RA and differentiate from other early arthritis. The possible effect of misclassification on spontaneous and drug-induced remission of early and very early RA awaits further elucidation. Such research will eventually lead to more reliable diagnostic and classification criteria for new-onset RA.
- EARLY RHEUMATOID ARTHRITIS
- VERY EARLY RHEUMATOID ARTHRITIS
- UNDIFFERENTIATED ARTHRITIS
- MISCLASSIFICATION
- 2010 ACR/EULAR RHEUMATOID ARTHRITIS CLASSIFICATION CRITERIA
With the advent of more effective treatment strategies — early treatment, disease-modifying antirheumatic drug (DMARD) combinations, tumor necrosis factor-α inhibitors, and tight control — remission has become a realistic goal in the management of rheumatoid arthritis (RA), particularly in early disease. Thus, for some time diagnosis and treatment of early RA (ERA) and very early RA (VERA) have been major targets of research and clinical practice. The goal of the 2010 American College of Rheumatology/European League Against Rheumatism (ACR/EULAR) classification criteria is to identify RA among newly presenting patients with undifferentiated inflammatory synovitis. The criteria focus on features at earlier stages that are associated with persistent and/or erosive disease1. The EULAR treatment recommendations for RA similarly state that treatment with DMARD should be started as soon as the diagnosis of RA is made and should aim for remission or low disease activity2. Further, EULAR recommendations were formulated to improve the management of early arthritis (EA)3. The recommendations represent advances in turning the present knowledge related to early RA into scientific and practical reality. Nevertheless, there is a need for further improvement and progress. The questions: Is there a risk of misclassification and consequently overtreatment with aggressive therapy in ERA? Which future practices may overcome current problems and limitations?
ERA
Interest in the early symptoms of RA goes back to the 1940s, when several authors made the distinction between the “typical” form of disease characterized by a slow and insidious onset and progressing course, and several “atypical” forms, which are further distinguished according to the mode of onset: acute start in small joints, acute start in large joints, asymmetrical joint symptoms, low sedimentation rate, arthralgic symptoms, and uncharacteristic symptoms4. At that time, RA was considered a nonspecific syndrome that could be triggered by many diverse etiological factors such as psoriasis, urethritis, and ulcerative colitis5. Prompted by the discovery of the association of RA with the rheumatoid factor (RF), it was apparent that many seronegative patients are clinically and radiographically quite different from patients with seropositive RA. The idea that these seronegative arthritides were, in fact, entirely separate entities was mirrored by the Nomenclature and Classification of the Rheumatic Diseases proposed by the American Rheumatism Association in 19636. In that classification, RA, juvenile Still’s disease, ankylosing spondylitis, psoriatic arthritis (PsA), and Reiter’s syndrome were classified under separate headings with the common denominator “polyarthritis of unknown origin.”
The introduction of early arthritis clinics since the 1980s provided much information regarding EA, described factors predicting persistence and prognosis in ERA, and continued to show that the diagnosis of RA is often challenging, most importantly due to lack of a clinical or laboratory gold standard7,8,9. The construct of ERA and each component of the definition (i.e., “early” and “RA”) has an indeterminate aspect, especially as the criteria for diagnosing RA are based on established disease10. The duration for ERA and VERA in the literature varies widely: for ERA, 2–3 years, and for VERA, 6 weeks to 3 months. The working group of the 2010 RA classification scheme, although not developed as criteria for ERA, used data from early arthritis clinics and designed the study to earlier identify that subset of patients who are at sufficiently high risk of persistent and/or erosive disease as to be considered for intervention with DMARD and classified as having RA1. Conversely, following this current model underlying the disease construct “RA,” one may conclude that self-limiting and nonerosive arthritis is not RA. Undifferentiated arthritis (UA) and EA fall mostly into this category, but with variability in the extent of joint involvement and disease duration overlapping with RA. Most importantly, as known for some time and shown again recently, early and very early UA is usually RF-negative (49% to 75% and 91%, respectively) and anticitrullinated protein antibody (ACPA)-negative (71% to 76% and 90%)11. Most of the patients with seronegative EA will never develop RA, although no doubt there are patients with otherwise typical RA who are seronegative. On the other hand, numerous studies have shown that positive RF and ACPA have a very high prognostic value for the development from EA and very early arthritis to RA12.
Risk of misclassification and overuse of aggressive therapy
Six recent studies reported the performance of the 2010 RA classification criteria in prospective EA cohorts (symptom duration < 1 year to < 2 years), in a very early arthritis cohort (seen within 3 months of the onset of any symptom), and in undiagnosed subjects with joint symptoms (median duration of symptoms 18 weeks)13,14,15,16,17,18. The characteristics of these studies vary considerably as do the reported evaluation measures (Table 1). Only the sensitivities and specificities are available from all studies and may give a comparative overview of the performance of the 2010 RA classification criteria. Overall, patients with ERA and VERA are identified in 58%–91% and 62%–74%, respectively, a rather unsatisfactory result in view of the fact that the new criteria were developed to facilitate the early recognition of RA. Most importantly, because of the low specificities (47%–60% and 66%–78%, respectively), up to half and one-third of the patients may be misclassified as having ERA and VERA, respectively. These data highlight that overdiagnosis and underdiagnosis may become important issues if the criteria recommend use of synthetic and biological DMARD. Cader, et al in their study of patients with VERA suggested that the 2010 RA classification criteria will allow more rapid identification of patients requiring methotrexate (MTX) compared with the 1987 criteria if applied at baseline, but that misdiagnosis may become significant if these criteria are used to direct treatment within the phase when treatment makes the greatest difference — the first 3 months after symptom onset14. Recently Britsemmer, et al reported that 51% of ACR/EULAR “non-RA” patients compared to 86% of patients with RA were treated with MTX in the first year, suggesting that the rheumatologists in their clinic had a more aggressive approach to EA during the same period than the rheumatologists treating the cohorts that were used to derive the criteria17. Obviously, MTX is neither a “gold standard” for RA nor a static feature, as rheumatologists have a tendency to treat earlier and more aggressively.
In addition to misclassification and unjustified treatment, there are other potential problems with the new criteria. For example, how does defined erosiveness typical for RA according to the 2010 criteria lead to the diagnosis of RA? How reliable is it to compare RF tests done using different methods or different isotypes? Is it reliable to use tender joints equal to swollen joints? What tests have to be done before applying these criteria to patients with arthritis not otherwise explained?
Future directions
Classification criteria serve to define disease groups for clinical and epidemiologic research, facilitate selection of similar patients for clinical trials, and allow for comparison of results across studies19. If the criteria are not valid, participants without disease may be included in disease groups in studies, and participants with clear-cut disease may be excluded. Thus, the validity of classification criteria is critical to the ability to understand and treat rheumatic diseases. Validation of criteria sets for diagnostic purposes usually requires very high specificity with good sensitivity. By comparison, validation of criteria sets for use in clinical trials and epidemiologic studies requires a balance of sensitivity and specificity19. It has been argued that disease and classification criteria represent a continuum because every set of disease criteria is created as a classification and has the potential of becoming diagnostic if it has sufficient internal and especially external validity20. Accordingly, a diagnosis is, in fact, making a classification in an individual patient. Different possibilities may exist to address some of the problems and overcome the limitations of the new RA classification criteria to classify and diagnose ERA and VERA in clinical practice and research by using expert opinion and generating future evidence (Table 2)21,22.
In clinical practice, the rheumatologist as the expert can balance between possible or probable RA, depending on the level of confidence. Recently, when using a diagnostic certainty scale at baseline (0 to 100 visual analog scale), all the patients with a score > 75 at their inclusion in an EA cohort subsequently received a diagnosis of RA with, as a gold standard, satisfaction of the 1987 ACR criteria at the 2-year visit23. In another study, the replacement of rheumatoid nodules with ACPA as a criterion in the 1987 ACR classification criteria increased the sensitivity (87% vs 82%) without losing specificity (95.6% vs 95.6%) in ERA patients who had a disease duration ≤ 2 years24. An outcome study of EA reflecting the patient population seen by rheumatologists in everyday practice (symptom duration < 1 year) assessed levels of agreement between a diagnosis of RA recorded by an office-based rheumatologist at inclusion, then 2 and 10 years later25. Agreement between the baseline diagnosis and the final diagnosis was low. Therefore, Fautrel, in an editorial, favored other diagnostic terms such as “persistent arthritis” and “erosive arthritis” until the accurate diagnosis after followup22. The problem with the term “persistent arthritis” is that it can only be used in retrospect after 1–2 years have passed. Also, “erosive arthritis” is not a term to be used before RA can be diagnosed, since erosive arthritis leads to the diagnosis of RA directly, according to the 2010 criteria. Therefore, the term “UA” should be favored to stress the preliminary diagnosis and various possible outcomes: early stage of a defined arthritis that will meet criteria in time, a forme fruste or partial form of a classifiable disease, an overlap of more than 1 disease entity, or an arthritis of unknown origin that may (or may not) become differentiated in the future26. The term “UA” also encourages the physician to recognize the potential dynamic nature of inflammatory arthritis, which over time can persist as UA, progress to a specific diagnosis, or enter into remission, which implies the need for ongoing reevaluation.
The accuracy of a prediction rule developed to estimate the chance of progression to RA in individual patients presenting with UA was investigated in 3 cohorts of patients with disease duration between 4 weeks and 2 years27. The prediction rule has an excellent discriminative ability for assessing the likelihood of progression to RA (classified according to the 1987 criteria) after ≥ 1 year of followup with positive predictive values between 93% and 100% and negative predictive values between 83% and 86%. Moreover, the prediction rule is validated using data from early arthritis clinics in Germany, the United Kingdom, Canada, Russia, and Japan28. The application of this rule will allow individualized treatment decision-making for patients with UA as long as the 2010 criteria are not improved to classify ERA and VERA.
In addition, modern imaging techniques (sonography, magnetic resonance imaging) not included in the 2010 criteria are very effective in identifying erosions and are increasingly performed in clinical practice to diagnose ERA and VERA29,30. A study evaluating the use of ultrasound joint counts in the prediction of RA in patients with very early synovitis (duration ≤ 3 months) showed that by adding grey-scale and power Doppler scanning of metacarpophalangeal joints, wrists, and metatarsophalangeal joints to the 2010 criteria, more patients were classified as having RA, including several later classified as RA by the 1987 criteria, one with ultrasound erosions31.
There are also worthwhile targets for future research (Table 2). Remarkably, no appropriate clinical and laboratory measurements for spondyloarthritis (SpA) and reactive arthritis (ReA) were systematically recorded in the datasets selected for phase 1 of the 2010 RA criteria despite availability of tests for HLA-B27 and bacterial infections at the initiation of the EA cohorts32. Applying a comprehensive diagnostic program including HLA-B27 typing and microbiological testing for infective agents (Chlamydia trachomatis, Yersinia enterocolitica, pseudotuberculosis, Borrelia burgdorferi, Campylobacter jejuni), we have shown in a prospective 2-year survey of patients with EA (< 1 year duration) that the 1987 ACR criteria have a good diagnostic performance (sensitivity 90%, specificity 90%) for ERA33. A population-based study in southern Sweden using such a comprehensive program diagnosed ReA and UA more frequently than RA34. Therefore, it seems essential to include these diagnostic measurements into datasets for development of future diagnostic and classification criteria, although the 3E Initiative in Rheumatology recommended not testing for ReA, and deemed that HLA-B27 typing was helpful only in specific settings35. In addition, one may test if the application of evaluated classification criteria, such as not fulfilling classification criteria for PsA and for peripheral SpA, perform as exclusion criteria instead of leaving it to expert opinion whether the synovitis is not better explained by another disease. This may enhance the specificity of the 2010 ACR/EULAR criteria. Further, it would be worthwhile to reinvestigate the 2010 ACR/EULAR criteria with the “classification tree” method used for the 1987 ACR classification criteria, which yields a better accuracy than the traditional format (4 of 7 criteria) of the 1987 ACR classification criteria, especially in early disease36. A tree format is available for the 2010 criteria: presence of synovitis is required (condition 1), followed by absence of a better alternative diagnosis (condition 2), and then by absence of erosions typical for RA (condition 3). Only patients meeting all 3 conditions are eligible for scoring. This tree format is therefore a prerequisite and does not change the results. The situation is different from the tree versus list versions of the 1987 ACR classification criteria, where the tree format is an alternative to the list format giving the lowest number of misclassifications36. Finally, Corrao, et al suggested use of a statistically driven process to weight each criterion by likelihood ratios, instead of the formulation process of the 2010 criteria, based on a form of structured consent by expert opinion37. Positive and negative likelihood ratios (and their 95% CI) could be computed for groups with similar criteria to help clinicians manage patients with EA effectively, using Bayesian reasoning. A totally new approach may be a very recently described highly reproducible, automated, multiplex biomarker assay testing for autoantibodies, cytokines, and bone-turnover products that can reliably distinguish between patients with early RA (< 6 months’ duration), those with other inflammatory arthritides (PsA, AS), and healthy individuals38. The exploratory study provided high sensitivity (84.2%) and specificity (93.8%) in the diagnostic discrimination of RA, paving the way for future optimized classification and diagnosis.
Improvement of the accuracy of diagnosis and classification of ERA and VERA is a continuous challenge. The recent EULAR guidelines for the management of EA3, the multinational evidence-based recommendations on how to investigate and followup undifferentiated peripheral inflammatory arthritis33, and the algorithm for identification of undifferentiated peripheral inflammatory arthritis39 are steps forward to promote awareness among treating physicians of the importance of treating early and to improve patient outcomes. But the overriding risk of misclassification is that of overtreatment with potentially toxic agents (e.g., DMARD), even considering that poor recognition and inadequate intervention in the earliest phases of inflammatory arthritis may occur more often. Therefore, a consensus is needed to better define the early phase of RA and differentiate it from other EA. A possible effect of misclassification on the “window of opportunity” hypothesis and on spontaneous and drug-free remission of ERA and VERA also awaits further elucidation40. Such activities should lead to more reliable diagnostic and classification criteria for new-onset RA.
- Accepted for publication September 24, 2011.