The new ACR/EULAR classification criteria for RA: how are the new criteria performing in the clinic?

Transcription

1 RHEUMATOLOGY Rheumatology 2012;51:vi10 vi15 doi: /rheumatology/kes280 The new ACR/EULAR classification criteria for RA: how are the new criteria performing in the clinic? Vivian P. Bykerk 1,2 and Elena M. Massarotti 3,4 Abstract The objective of the 2010 ACR/European League Against Rheumatism classification criteria for RA was to distinguish patients at high risk for developing persistent erosive and/or inflammatory disease from those with undifferentiated inflammatory arthritis. These criteria were developed for use in clinical trials; in order to implement these criteria most effectively, they need to be validated in real-world settings. The 1987 criteria may have led to underdiagnosis in the case of patients with positive anti-citrullinated peptide antibody values but no evidence of radiographic progression of joint erosion, or overdiagnosis in the case of some patients with FM; similarly, the possibility that the 2010 criteria may result in overdiagnosis cannot be excluded. Prospective validation of the 2010 criteria has been carried out in several cohorts, with reported sensitivities ranging from 0.50 to 0.60 and specificities from 0.88 to The sensitivity and specificity of the 2010 criteria were 0.74 and 0.66 when compared against the gold standard of needing MTX therapy in the opinion of experienced clinicians, and 0.69 and 0.72 against the standard of having persistent synovitis despite DMARDs after 1 year. Other comparisons have yielded similar sensitivities and specificities, ranging up to 0.85 for the gold standard of needing MTX therapy. Questions remain concerning the utility of the 2010 criteria for non-arthritis health care practitioners, who may be less than expert in identifying swollen joints and may underestimate the number of joints affected by synovitis. US may be of value in the future, but its role remains to be validated. Key words: rheumatoid arthritis, ACR/EULAR classification criteria, radiographic progression, persistent synovitis, fibromyalgia, under/overdiagnosis of RA. Introduction Recently new classification criteria were developed through a joint process involving the ACR and the European League Against Rheumatism (EULAR) in order to facilitate identification of patients with undifferentiated arthritis (UA) who are at high risk of developing persistent/ erosive disease, with an emphasis on the application of these criteria to patients with early rather than established disease [1, 2]. The previous (1987) classification criteria emphasized the identification of patients with established RA [3]. The new criteria could potentially 1 Inflammatory Arthritis Center, Hospital for Special Surgery, New York, 2 Department of Medicine, Weill Cornell Medical School, New York, NY, 3 Department of Medicine, Harvard Medical School and 4 Center for Clinical Therapeutics, Division of Rheumatology, Brigham and Women s Hospital, Boston, MA, USA. Submitted 10 January 2012; revised version accepted 14 September Correspondence to: Vivian Bykerk, Hospital for Special Surgery 535 East 70th Street, 6th Floor, New York, NY 10021, USA. Bykerkv@hss.edu permit the study of patients with earlier disease to better evaluate new treatment paradigms and new therapies that may be effective in arresting the progression of RA at an early stage. Classification criteria are needed as a basis for a common approach to disease definition to allow for comparison across studies. The objective of this review is to summarize the presented/published data available on the performance of the new criteria in the clinical setting and to discuss the potential advantages and pitfalls of the new criteria. Clinical presentations with the new criteria The 1987 classification criteria for RA were never intended for use in clinical practice. They were developed primarily for clinical research purposes. Nonetheless, they came to define what we called RA. One of the most significant limitations of these criteria was their low sensitivity in diagnosing patients with early disease. The advent of increasingly effective treatment paradigms, incorporating the use of conventional and biologic DMARDs, has made it! The Author Published by Oxford University Press on behalf of the British Society for Rheumatology. All rights reserved. For Permissions, please journals.permissions@oup.com

2 The new ACR/EULAR RA classification criteria possible to prevent destructive disease in patients who are identified early in the course of disease rather than after progression to irreversible radiographic changes. Patients are now presenting earlier to the clinic with joint pain involving only one or two joints and, not infrequently, with positive anti-citrullinated peptide antibodies (ACPAs) and elevated concentrations of acute-phase reactants (ESR and CRP) [4]. These patients do not typically present with nodules or joint erosions and until now may have been labelled as having UA. It is thought that the new 2010 criteria will more appropriately classify these patients as likely to progress to persistent and potentially destructive disease. However, the new 2010 criteria were not developed for use in clinical practice. In order to best identify patients with early disease in the clinic, the criteria will need to be validated in cohorts representing real-world settings [1, 2, 5 7]. MTX is often selected as the first-line non-biologic DMARD. The decision to initiate MTX therapy can be triggered by several factors, including the presence of polyarticular or erosive disease. There are scenarios, however, in which the decision to begin MTX therapy is less clear. One example would be where clinicians would be highly suspicious that the patient has RA but does not meet the 1987 ACR classification criteria for RA. Such an example would be a 22-year-old woman presenting with 4 weeks of joint pain; 45 min of morning stiffness; swelling of the left forefoot, right wrist and second MCP joint (nine swollen small joints); a positive anti-acpa (twice the upper limit of normal); an elevated CRP concentration; and normal radiographs. Using the new 2010 criteria, a score of 3 points would be given for having 4 10 small joints involved. She would score 2 points for having a low titre of ACPAs and 1 point for an elevated CRP value. Her total score of 6 points would meet the 2010 classification criteria for RA. However, this patient would have been undiagnosed using the 1987 criteria, meeting only three of seven criteria. She would not have symmetric arthritis, rheumatoid nodules or radiographic changes [3]. Based on the new criteria, one could justify beginning MTX therapy in such a patient [8, 9]. This example illustrates why these criteria were developed. The importance of continuing to consider the differential diagnosis when using any set of diagnostic criteria is illustrated in the following scenario. Consider a 42-year-old woman with a 16-week history of overall pain; an hour of vague, generalized morning stiffness; tenderness in 30 joints with swelling noted in the right second PIP joint; 18/18 FM tender points; a negative RF of <8 IU/l (normal <20 IU/l); normal radiographs; and a marginally elevated ESR of 26 mm/h (normal <20 mm/h). With the new criteria, this patient could be considered as having more than 10 joints involved with >6 weeks of symptoms and score 1 for an abnormal ESR. This patient would inappropriately score 7 on the new criteria, classifying her disorder as RA, even though she has FM. Thus the 2010 classification criteria may also, on occasion, result in overdiagnosis of RA. This case highlights the issue that, without carefully excluding other diagnoses, there is a risk for overdiagnosis. Although some would argue that a patient such as this warrants a trial of anti-inflammatory therapy, perhaps even prednisone or MTX, one needs to count joints with synovitis very carefully. If there is doubt as to which joints have active synovitis and one chooses not to have the patient undergo a trial of therapy, high-sensitivity imaging may play an important role in determining if there are joints with confirmed synovitis. Furthermore, repeated evaluation of the patient is important even in suspected inflammatory arthritis, as it may later become apparent that the patient s symptoms are due to early RA or to another condition. Many studies still need to be considered prior to widely applying these criteria to clinical practice. These need to include the issues of risk benefit and cost-effectiveness of a trial of therapy in questionable cases as opposed to further delineation of synovitis, bone oedema and erosions using musculoskeletal US, MRI or even synovial biopsy. Role of erosions in the new diagnostic criteria The 2010 classification committee determined that if a patient presenting with synovitis in the absence of another diagnosis demonstrated a typical RA erosion, then the diagnosis of RA was met. This issue of what constitutes a typical RA erosion may lead to some confusion, particularly in the setting of erosive OA or PsA. Some authors have also suggested that the presence of one erosion may not predict persistent disease [10]. Several authors have now shown that erosions seen on US and bone oedema detected on MRI predict the development of radiographic erosions [11]. The role of these technologies in diagnosing RA going forward remains to be determined. How do the criteria perform in the validation cohorts? The question arises whether the criteria will perform as well in observational cohorts of patients with new-onset inflammatory arthritis as in the cohort sets studied in the development of the 2010 classification. The criteria were satisfied in 87 97% of patients who began treatment with MTX [1, 2, 5]. Several other investigators have gone on to apply the criteria to patients in their cohorts with early inflammatory arthritis. Overall, the criteria have now been tested in several clinical cohorts and settings. Prospective validation has been tested against multiple outcomes, including the development of new erosions, starting/continuing MTX (or other DMARDs), fulfilling the 1987 criteria and the opinion of the rheumatologist. Van der Linden et al. [12] studied 972 patients with UA from three separate cohorts (Leiden Cohort, Berlin Early Arthritis Cohort and the Norwegian Very Early Arthritis Cohort) for RA development. They calculated different test characteristics (sensitivity, specificity, positive likelihood ratio and negative vi11

3 Vivian P. Bykerk and Elena M. Massarotti likelihood ratio) in this patient group. Depending on which cohort they applied the new 2010 criteria to, the likelihood of developing RA ranged from 3.45 to The presence of high-titre RF and/or ACPAs increased this likelihood. They noted that the sensitivity of the new 2010 criteria in these cohorts ranged from 0.50 to 0.60, whereas the specificity ranged from 0.88 to Thus, in these patients, underclassification or underdiagnosis may have occurred. Furthermore, patients who were RF or ACPA positive were more likely to develop erosions and less likely to achieve spontaneous remission, highlighting the important role of biomarkers in the classification of RA. The 2010 ACR/EULAR criteria have been validated in several cohorts of patients and been found to have excellent discriminative ability to assess the likelihood of progression to various forms of RA outcomes. Alves et al. [13] tested the diagnostic properties of the new 2010 criteria in a very early arthritis cohort of patients that presented with joint pain/loss of range of movement in at least two joints or synovitis of at least one joint of <1 year s duration, unexplained by any other illness. In patients with UA in their cohort, a large area under the curve (AUC) was demonstrated when the investigators used the new 2010 criteria that was comparable to using two other diagnostic algorithms. They showed that in the patients with UA (n = 231), the AUC for MTX use with overlapping 95% CIs was 0.79 (95% CI 0.73, 0.83); for the ACR/EULAR 2010 criteria it was 0.80 (95% CI 0.74, 0.87); for the van der Helm algorithm [14] and for the Visser algorithm it was 0.84 (95% CI 0.77, 0.88) [15]. The sensitivity and specificity for the 2010 criteria were both high (0.74 and 0.66, respectively, when using the gold standard of needing MTX, and 0.69 and 0.72 when using persistent arthritis, defined as still having active synovitis or needing DMARDs at 1 year). They point out that one requires not only high sensitivity to prevent undertreatment of disease, but also high specificity to prevent overtreatment. It is not unusual for a high specificity to be accompanied by low to moderate sensitivity and vice versa. Given that they observed a sensitivity and specificity of 70% for the ACR/EULAR 2010 criteria, Alves et al. [13] recommended these criteria for use in real-world patients. Britsemmer et al. [16] also evaluated the 2010 criteria in 455 patients with early arthritis and showed similar results to those of Alves et al. [13]. In their analysis, the discriminative abilities of the three criteria sets (2010 ACR/ EULAR, 1987 ACR criteria and Visser algorithm) were similar with AUCs of (gold standard: MTX), (gold standard: expert opinion RA) and (gold standard: erosive disease after 3 years). They observed that the sensitivity of the 2010 criteria was highest at 0.85 for the gold standard of treating with MTX [16]. Cader et al. [17] highlighted the issue of potential overdiagnosis in their validation cohort. They studied 265 patients with very early symptoms of <3 months duration who were followed for 18 months. Of these 265 patients, 60 had alternative diagnoses at baseline. They applied the criteria to the remaining 205 patients and noted that 20% fulfilled both 1987 and 2010 criteria, 3% fulfilled only 1987 criteria and 22% fulfilled only 2010 criteria at baseline. They noted that more patients who met the 2010 criteria eventually required DMARDs within the first 18 months when the criteria were applied at baseline than did those who did not meet these criteria (62% vs 38%, respectively; P < 0.001) for DMARDs and (68% vs 42%, respectively; P < 0.01) for MTX. However, they also noted that more patients whose disease eventually resolved without ever requiring DMARDs were classified at baseline as having RA according to the 2010 criteria than those classified with the 1987 criteria (8% vs 2%, respectively; P = 0.01) [17]. This study raises the issue whether the increased sensitivity of the 2010 criteria comes at the expense of reduced specificity. Hazlewood et al. [18] outlined a recent algorithm that should be considered in all patients with new-onset UA to whom one may wish to apply the new 2010 ACR criteria (Fig. 1). These authors highlight the importance of carefully investigating the patient to confirm or rule out a diagnosis. In particular, health care providers need to rule out psoriasis and exposure to agents that will cause reactive arthritis, and to identify infectious causes of arthritis pertinent to the geographic location of the patient. Also, it is important to ensure that the patient has active inflammatory synovitis [18]. Other researchers have not shown that there is an increase in the number of patients diagnosed with RA by using the new 2010 criteria compared with the 1987 criteria. Varache et al. [19] studied patients with arthritis with a duration of <1 year who were followed over 2 years for their eventual diagnosis. They applied the gold standard definition of RA in their study as diagnosis of RA by an office-based rheumatologist after 2 years combined with [a DMARD] or glucocorticoid treatment [19]. These authors compared the performance characteristics of five criteria sets including the 1987 ACR criteria and the 2010 ACR/EULAR criteria against their gold standard definition of RA. They concluded that there was no significant difference between the diagnostic accuracy of the 2010 ACR/EULAR criteria and those classified as RA by the 1987 ACR criteria [19]. Can or will the criteria be used in clinical practice? As has been noted, the new 2010 criteria were not developed for use in clinical practice. Some argue that the scoring is too complicated and that although the sensitivity is high, in some cohorts the specificity is too low. The validation studies to date show that on average the sensitivity varies in the range of 60 85% (using the initiation of MTX as the gold standard), whereas the specificity ranges between 50% and 80%. In those studies that looked at the discriminative ability of the new 2010 criteria, consistent AUCs of 80% were reported despite differences in patient recruitment, symptom duration and time of the study [13, 16, 17]. vi12

4 The new ACR/EULAR RA classification criteria FIG. 1Algorithm for identification of undifferentiated peripheral arthritis [18]. P/E: physical examination; DDx: differential diagnosis; UPIA: undifferentiated peripheral inflammatory arthritis. Reproduced from ref. [18], with permission from the Journal of Rheumatology. General considerations in applying the new 2010 criteria to clinical practice The presence of ACPAs has, in part, enabled earlier and more specific diagnosis of RA. No doubt, new biomarkers will be discovered over the coming years that will further enhance the ability to diagnose early RA. These will need to be taken into account in future modifications of the classification criteria. What criteria should be used in clinical practice? Should the older criteria be applied first because these criteria embody the paradigm of RA? Should the new criteria be applied only if the 1987 criteria are not met? Some remain skeptical about implementing the new classification criteria in practice. One author quite rightly points out that, in clinical practice, we need to give our patients a diagnosis and also comments that, ideally, sensitivity and specificity should be 100%. He questions at vi13

5 Vivian P. Bykerk and Elena M. Massarotti what point classification criteria can become diagnostic criteria and that, in reality, diagnosis and classification are the same but that neither can be universal in other words, good for all purposes [20]. Who should be applying these classification criteria in practice? We are still left with the issue of how non-arthritis health care practitioners diagnose RA. The new criteria require expertise in the identification of synovitis. Clinicians who are less than expert in treating RA may be less able to identify swollen joints due to inflammatory arthritis and would then be likely to underestimate the number of joints with synovitis. Conversely they might overestimate joints with synovitis in the case of OA with inflamed joints. Thus the non-arthritis practitioner would be left with the conundrum of being able to diagnose RA in the presence of erosions on radiographs, assuming they are recognized, and the presence of high RF/ACPA levels and elevated acute-phase reactant concentrations. The combination of elevated inflammatory marker concentrations and the presence of joint symptoms for >6 weeks would yield a score of 2 with the new criteria, below the threshold for classification of RA. Thus, in order to confirm a diagnosis of RA, examining clinicians would need to identify swelling in at least one joint in order to use the criteria. In addition, they would need to identify swelling or tenderness in more than 10 joints (which gives a score of 5) in order to raise the score to at least 6 if the RF or ACPA serology is negative and in 1 3 small joints if the serology were mildly positive. It is only in the case of highly positive serology (RF or ACPA 53 times the upper limit of normal) along with a symptom duration of at least 6 weeks AND elevated acute-phase reactant concentrations that clinicians need identify only one small or two large joints to make the diagnosis of RA. This situation underscores the importance of reliably diagnosing synovitis on the clinical examination. The role of US in this setting may play a greater role in the future, but this remains to be validated. Conclusion The use of biologic agents for the treatment of RA over the last decade has resulted in great benefit for patients affected by this chronic illness. Once considered a progressively disabling disease, patients and rheumatologists can now, with reasonable confidence, embark upon a treatment course that is generally well tolerated, long lasting and effective in preventing long-term disability and irreversible joint destruction. These therapeutic advances for patients with well-established disease, along with the new classification criteria for RA incorporating the more specific ACPA antibodies, have allowed for the possibility of earlier identification of disease with the possibility of further improvements in disease outcome. The diagnosis of RA remains clinical, incorporating the traditional tenet of the interpretation of the history, examination and laboratory testing. As these new diagnostic criteria become increasingly incorporated into clinical practice, it is important for all diagnosticians to remember that making the diagnosis of early or established RA can not be relegated to a checklist of sorts, and requires clinical acumen in ascertaining synovitis and the consideration of the many causes of joint inflammation. Rheumatology Key messages. The ACR/EULAR RA classification criteria have now been tested in several clinical cohorts and settings.. The utility of the ACR/EULAR RA classification criteria for non-arthritis health care practitioners remains questionable. Supplement: This paper forms part of the supplement Reducing the toll of autoimmune disease: a focus on rheumatoid arthritis. This supplement was supported by unrestricted education grants from Centocor and Genentech. Disclosure statement: E.M.M. is an investigator for BMS, Astra Zeneca and Human Genome Sciences and a consultant for Sanofi, UCB and the Goodwin Group. V.B. is a consultant for Bristol-Myers Squibb, Roche and UCB. References 1 Aletaha D, Neogi T, Silman AJ et al Rheumatoid arthritis classification criteria: an American College of Rheumatology/European League Against Rheumatism collaborative initiative. Arthritis Rheum 2010;62: Neogi T, Aletaha D, Silman AJ et al. The 2010 American College of Rheumatology/European League Against Rheumatism classification criteria for rheumatoid arthritis: phase 2 methodological report. Arthritis Rheum 2010;62: Arnett FC, Edworthy SM, Bloch DA et al. The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis. Arthritis Rheum 1988; 31: van de Sande MG, de Hair MJ, van der Leij C et al. Different stages of rheumatoid arthritis: features of the synovium in the preclinical phase. Ann Rheum Dis 2011; 70: Funovits J, Aletaha D, Bykerk V et al. The 2010 American College of Rheumatology/European League Against Rheumatism classification criteria for rheumatoid arthritis: methodological report phase I. Ann Rheum Dis 2010;69: Bykerk V, Emery P. Delay in receiving rheumatology care leads to long-term harm. Arthritis Rheum 2010;62: van der Linden MP, le Cessie S, Raza K et al. Long-term impact of delay in assessment of patients with early arthritis. Arthritis Rheum 2010;62: Combe B, Landewe R, Lukas C et al. EULAR recommendations for the management of early arthritis: report of a task force of the European Standing Committee for International Clinical Studies Including Therapeutics (ESCISIT). Ann Rheum Dis 2007;66: vi14

6 The new ACR/EULAR RA classification criteria 9 Bykerk VP, Akhavan P, Hazlewood GS et al. Canadian Rheumatology Association recommendations for pharmacological management of rheumatoid arthritis with traditional and biologic disease-modifying antirheumatic drugs. J Rheumatol 2012;39: Thabet MM, Huizinga TW, van der Heijde DM, van der Helm-van Mil AH. The prognostic value of baseline erosions in undifferentiated arthritis. Arthritis Res Ther 2009;11:R Ostergaard M, Boyesen P, Eshed I et al. Development and preliminary validation of a magnetic resonance imaging joint space narrowing score for use in rheumatoid arthritis: potential adjunct to the OMERACT RA MRI scoring system. J Rheumatol 2011;38: van der Linden MP, Batstra MR, Bakker-Jonges LE et al. Toward a data-driven evaluation of the 2010 American College of Rheumatology/European League Against Rheumatism criteria for rheumatoid arthritis: is it sensible to look at levels of rheumatoid factor? Arthritis Rheum 2011;63: Alves C, Luime JJ, van Zeben D et al. Diagnostic performance of the ACR/EULAR 2010 criteria for rheumatoid arthritis and two diagnostic algorithms in an early arthritis clinic (REACH). Ann Rheum Dis 2011;70: van der Helm-van Mil AH, Detert J, le Cessie S et al. Validation of a prediction rule for disease outcome in patients with recent-onset undifferentiated arthritis: moving toward individualized treatment decision-making. Arthritis Rheum 2008;58: Visser H, le Cessie S, Vos K, Breedveld FC, Hazes JM. How to diagnose rheumatoid arthritis early: a prediction model for persistent (erosive) arthritis. Arthritis Rheum 2002;46: Britsemmer K, Ursum J, Gerritsen M, van Tuyl L, van Schaardenburg D. Validation of the 2010 ACR/EULAR classification criteria for rheumatoid arthritis: slight improvement over the 1987 ACR criteria. Ann Rheum Dis 2011;70: Cader MZ, Filer A, Hazlehurst J, de Pablo P, Buckley CD, Raza K. Performance of the 2010 ACR/EULAR criteria for rheumatoid arthritis: comparison with 1987 ACR criteria in a very early synovitis cohort. Ann Rheum Dis 2011;70: Hazlewood G, Aletaha D, Carmona L et al. Algorithm for identification of undifferentiated peripheral inflammatory arthritis: a multinational collaboration through the 3e initiative. J Rheumatol Suppl 2011;87: Varache S, Cornec D, Morvan J et al. Diagnostic accuracy of ACR/EULAR 2010 criteria for rheumatoid arthritis in a 2-year cohort. J Rheumatol 2011;38: Yazici H. A critical look at diagnostic criteria: time for a change? Bull NYU Hosp Jt Dis 2011;69: vi15