Introduction

Rheumatoid arthritis (RA) is a multi-faceted inflammatory disease affecting joints and surrounding structures as well as other organ systems. RA has traditionally been associated with a substantial burden of disease, loss of work productivity and increased mortality . The pathophysiology of RA is not completely understood, and no single test or gold standard exists to confirm the diagnosis. Hence, the diagnosis is made based on a set of findings and symptoms typical for the RA phenotype rather than measurement of the specific pathogenic processes that lead to this phenotype. It is pivotal to be able to identify the RA phenotype at an early stage of disease development, because timely intervention with disease-modifying anti-rheumatic drugs (DMARDs) has been shown to improve the outcome of this disease . The window of opportunity hypothesis states that there is a time frame in the early stages of disease development where treatment could delay or prevent the disease from becoming chronic and effectively reduce the chance of erosive changes to the joints . Significant advances in the understanding of the underlying disease processes that cause inflammation has led to the development of targeted therapies against molecules that are involved in this inflammation (e.g., cytokines like tumour necrosis factor (TNF)-alpha and interleukin (IL)-6). These new biologic agents not only act to alleviate symptoms but also intervene directly with the disease process and therefore have the potential to improve long-term outcome in RA patients. Remission is now a realistic option in RA, partly due to these novel medications and partly as a result of modern targeted treatment strategies using early optimised doses of methotrexate (MTX) with additional treatment being escalated more frequently .

Classification criteria vs. diagnostic criteria

To create a common background for disease definition, both diagnostic and classification criteria for RA have been developed over the years. Although diagnostic and classification criteria may consist of the same type of clinical, laboratory or other markers, they have different aims. The purpose of classification criteria is to ensure homogeneity among patients in clinical studies. They should be applied to patients in whom a diagnosis has already been made and should ideally have high specificity (close to 100%), so that patients without the disease are not misclassified. Classification criteria should provide a dichotomous answer (yes/no). By contrast, diagnostic criteria facilitate the clinician in making a diagnosis in an individual patient. The prevalence or pretest probability of the disease is of interest, as this will influence the predictive performance of the criteria . For example, the risk of a patient having RA will be relatively high in an early arthritis clinic (EAC) and relatively low in a primary-care facility. Diagnostic criteria should ideally have high positive predictive value and likelihood ratios and should provide a graded estimate of probability of disease (e.g., expressed by percent chance, a score or in categories like definite/probable/possible) .

Classification criteria vs. diagnostic criteria

To create a common background for disease definition, both diagnostic and classification criteria for RA have been developed over the years. Although diagnostic and classification criteria may consist of the same type of clinical, laboratory or other markers, they have different aims. The purpose of classification criteria is to ensure homogeneity among patients in clinical studies. They should be applied to patients in whom a diagnosis has already been made and should ideally have high specificity (close to 100%), so that patients without the disease are not misclassified. Classification criteria should provide a dichotomous answer (yes/no). By contrast, diagnostic criteria facilitate the clinician in making a diagnosis in an individual patient. The prevalence or pretest probability of the disease is of interest, as this will influence the predictive performance of the criteria . For example, the risk of a patient having RA will be relatively high in an early arthritis clinic (EAC) and relatively low in a primary-care facility. Diagnostic criteria should ideally have high positive predictive value and likelihood ratios and should provide a graded estimate of probability of disease (e.g., expressed by percent chance, a score or in categories like definite/probable/possible) .

Previous criteria sets

The 1958 diagnostic criteria for RA were the first set of criteria to gain widespread use ( Table 1 a) . The criteria were originally published in 1956, but were revised 2 years later. These criteria consisted of a set of 11 elements, where 5/11 had to be fulfilled for a patient to be considered to have definite RA. Two of the criteria were histologic features and one required synovial fluid examination. These three criteria were omitted in a later revision in 1961 in Rome, and a set of exclusion criteria was added . The criteria showed good sensitivity and specificity when compared to the later 1987 criteria . Interestingly, a patient would be considered to have probable RA if three or four of the 1958 criteria were met and the exclusion criteria were fulfilled.

In 1987, the American Rheumatism Association developed classification criteria to ensure that individuals enrolled in clinical trials really had RA ( Table 1 b) . The criteria were developed using patients with longstanding disease (mean duration 8 years) and had a sensitivity of 91% and a specificity of 89% in the original population. The criteria are useful for distinguishing patients with established RA from those with other inflammatory arthritides, but have proven less valuable for identification of RA in the early phases of the disease or to determine who would benefit from early intervention . Several features included in the 1987 criteria are infrequently found in early disease, such as erosive changes and rheumatoid nodules. With modern treatment strategies and the acceptance of the concept of RA as an evolving disease, the need for new criteria facilitating early identification has become apparent. In addition, the convincing body of evidence regarding the importance of antibodies against citrullinated proteins (ACPAs) in RA has been highlighted in the debate regarding the need for new criteria .

The 2010 ACR/EULAR classification criteria for RA

The new classification criteria for RA were developed by a task force consisting of members from the American College of Rheumatology (ACR) and the European League Against Rheumatism (EULAR) and were published in 2010 ( Table 1 c) . The aim of the task force was to identify patients with undifferentiated arthritis who were likely to develop a chronic, erosive disease. The criteria were designed to form a basis for initiation of DMARDs in such patients. These were developed through three elaborate phases . In the initial, data-driven phase, data from nine European and Canadian EACs on 3115 patients with early undifferentiated arthritis (UA) were used to identify clinical and laboratory markers associated with the rheumatologist’s decision to start MTX therapy. The independent contribution of each variable was determined through univariate regression modelling, principal component analysis and subsequent multivariate regression analysis. Swollen and tender joints in the fingers and wrists, acute-phase reactants and levels of serological markers were found to contribute to the predefined outcome. Phase II was consensus based and included a panel of 24 rheumatologists (12 from North America and 12 from Europe) who engaged in a Delphi process. The purpose of this phase was to identify clinical and laboratory factors deemed to influence the probability of developing “persistent inflammatory and/or erosive arthritis that is currently considered to be RA.” This was obtained with the help of a set of real-life case scenarios that each clinician contributed to the process. Through review of these case histories by the expert panel, the domains that were important in determining the probability of developing RA were identified. The weights of these domains and categories within the domains were determined by use of decision-support computer software ( www.1000minds.com ) and decision-science theory. In the final and third phase of the criteria development, the results of phases I and II were combined. The scoring system was refined and the optimal cut-point to define ‘definite’ RA was determined. The cut-point was optimised to ensure high sensitivity to capture patients who could benefit from MTX therapy. The final criteria set consists of a scoring system with four domains: distribution of swollen and tender joints (range 0–5 points), serology (range 0–3 points), acute-phase reactants (range 0–1 point) and duration of joint swelling (range 0–1 points). In order to be evaluated by the criteria, a patient must have at least one joint with definite clinical synovitis (swelling) and the synovitis should not be better explained by another disease. A patient with a score of 6 or more out of 10 possible points is considered to have definite RA. Patients with evidence of erosive disease typical for RA with a history compatible with prior fulfilment of the criteria are automatically classified as RA. ‘Erosive disease typical for RA’ was recently defined as a cortical break (erosion) seen in at least three separate joints in the following sites: the proximal interphalangeal (PIP), the metacarpophalangeal (MCP), the wrist (counted as one joint) and the metatarsophalangeal (MTP) joints on radiographs of both hands and feet . The fulfilment of the criteria can also be determined by use of a tree algorithm, yielding the same results as the list format.

Performance of the 2010 criteria according to different reference standards

Many studies have tried to validate the new classification criteria. However, the studies have used different gold standards for validation, different follow-up times and different populations. Here we summarise the main findings that have been published so far according to the reference standards used.

Prediction of clinical RA diagnosis/expert opinion

This gold standard has high face validity, but is potentially biased by the individual physician’s experience and preferences and influenced by what information was available to the doctor at the time of diagnosis (e.g., ACPA status). Importantly, this outcome is also biased by the old criteria themselves, as the rheumatologist trained to think of RA using the paradigm of the 1987 criteria was likely to have these in ‘the back of his/her head’ when diagnosing RA. Circularity is a challenge when using expert opinion of RA as the reference standard.

Several studies have evaluated this outcome ( Table 3 ) . The sensitivities ranged from 0.59 to 0.95, while specificities ranged as widely as 0.34–0.96. Radner et al. found the pooled estimates to be 0.88 for sensitivity and 0.48 for specificity . Substantial heterogeneity between the study populations was seen, with regard to duration of symptoms, inclusion- and exclusion criteria (e.g., untreated polyarthritis lasting <6 weeks and early arthritis (EA) lasting for <2 years ²³ ) and follow-up times ranging from 6 months to 20 years. These differences probably explain some of the large variations in specificity. Britsemmer compared the performance of the 2010 and 1987 criteria to predict a clinical RA diagnosis by 1 year and found that the 1987 criteria had lower sensitivity (0.65 vs. 0.85) but higher specificity (0.80 vs. 0.64) . In a French study, the 2010 and 1987 criteria had similar sensitivities (0.58 vs. 0.60) for predicting a diagnosis of RA after 10 years, but the 2010 criteria actually had higher specificity (0.89 vs. 0.75) . The 2010 criteria also showed better specificity for predicting RA after long-term follow-up (0.96 vs. 0.86) in the study by Kaarela et al. .

Table 3

Validation studies for the 2010 ACR/EULAR classification criteria for rheumatoid arthritis: RA by expert opinion.

Study	Patients	Inclusion period	n	Outcome assessment	Sensitivity	Specificity
Biliavska	SAVE study cohort Early arthritis ≤ 16 weeks symptom duration	2004–2007	303	Investigator’s diagnosis RA 1 year	0.85	0.64
Britsemmer	Jan van Breemen Institute EAC SJC ≥ 2, symptom duration < 2 years, no prior DMARD	2000	455	Expert opinion RA 1 year	0.90	0.48
Cornec	Early arthritis, SJC ≥ 1, duration <1 year	1995–1997	164	Expert opinion RA 10 years	0.60	0.89
Fautrel	ESPOIR cohort SJC ≥ 2, duration 6 weeks-6 months	2002–2005	811	Expert opinion RA 2 years	0.95	0.34
Kaarela	RA patients, arthritis duration < 6 months vs. controls with definite SpA (ReA, PsA, AS) and SJC ≥ 1	1973–1975	216	Clinical diagnosis determined by long-term follow-up (20 years)	0.79	0.96
Kennish	Consecutive patients with joint symptoms in a rheumatology private practice	2010–2011	126	Rheumatologist’s diagnosis	0.97	0.55
Mourao	Untreated polyarthritis < 6 weeks	2005–2009	37	RA defined by clinical diagnosis plus fulfilment of 1987 criteria	0.71	0.56
Reneses	Early arthritis 4–12 months, SJC ≥ 2	2002–2006	201	Expert opinion RA 1 year	0.87	0.73

Only gold members can continue reading. Log In or Register to continue

Share this:

• Click to share on Twitter (Opens in new window)
• Click to share on Facebook (Opens in new window)

Related

Related posts:

Can rheumatoid arthritis be prevented? To screen or not to screen: How to find and identify very early arthritis Imaging in early rheumatoid arthritis What is the best treatment strategy for early RA? Outcomes of early rheumatoid arthritis – The WHO ICF framework Is there a place for initial treatment with biological DMARDs in the early phase of RA?

Stay updated, free articles. Join our Telegram channel

Join