Reexamining Remission Definitions in Rheumatoid Arthritis: Considering Disease Activity Score in 28 Joints, C-Reactive Protein, and Patient Global Assessment.

The editors of the five American College of Rheumatology (ACR) and European League Against Rheumatism (EULAR) journals have been reminded by this article by Felson et al that the ACR and EULAR have jointly agreed on various classification criteria, definitions, recommendations, or points to consider that do not always find reflection in manuscripts submitted to the journals. Consequently, in the future, the editors will enforce the use of the products obtained in the course of joint ACR/EULAR activities in all respective articles. For rheumatoid arthritis, this would mean to use the ACR/EULAR classification criteria, remission definitions, and recommendations on what to report in clinical trials and others as pertinent. The same applies to other diseases. There are valid and important reasons why these activities have been set forth by the ACR and EULAR, and therefore the various task forces’ conclusions, which have been endorsed by the ACR and EULAR, should be respected by investigators and administrators. This does not mean other methods could not be used in a study or article, but at the least, the reports should address the methods agreed on by the two organizations. Maintaining uniformity across major publications regarding rheumatoid arthritis remission or other definitions not only allows for more appropriate comparison across analyses but also enhances readers’ ability to interpret results. Author instructions across the five journals will more strongly reflect this requirement.

Over the last 30 years, treatment for rheumatoid arthritis (RA) has improved dramatically. By the early 2000s, disease remission had become a realistic goal, although the definition of remission varied widely, making it hard to compare treatment strategies and gauge how often remission occurred. In 2009, the American College of Rheumatology (ACR) and European League Against Rheumatism (EULAR) created a joint committee charged with recommending a definition of remission. Members of the committee suggested a large number of candidate definitions and, using a data‐driven consensus process, statisticians and programmers tested these candidates in a bank of RA trial data to see which definitions performed best in predicting long‐term good function and lack of radiographic progression. The committee endorsed a stringent definition using measures from the validated core set of outcome measures.

After reviewing analysis results, the committee selected two definitions of remission that were approved by the ACR and EULAR (1). The first was a Boolean version in which, to attain remission, a patient had to have tender and swollen joint counts of less than or equal to 1, a C‐reactive protein (CRP) level of less than or equal to 1 mg/dl, and a patient global assessment of arthritis activity of less than or equal to 1 (on a 0‐10 scale). The second recommended definition was a Simple Disease Activity Index (SDAI) score of less than or equal to 3.3, a score based on the same core set of outcomes. Although designed and validated in trials, these definitions could help assess treatment success in both trials and clinical practice, and in practice, they could serve as a treat‐to‐target goal for some patients.

Like all developed criteria, the ACR/EULAR 2011 RA remission criteria were labeled as provisionally approved and awaited validation in an independent sample for final approval. Many concerns have arisen since their publication. Among them is the continuing use in trials of Disease Activity Score in 28 Joints (DAS28) thresholds to define remission, questions about the use of the CRP level as an element of remission definitions, and questions about the appropriateness of including patient global assessment in defining RA remission. This editorial will address each of these issues.

Using the DAS28: When “remission” is often not remission

The DAS28 is a widely used measure of disease activity. An ACR committee that critically evaluated RA disease activity measures for use in the clinical setting found that the DAS28 met predefined criteria, including providing a score that stratified patients into at least three disease activity states, being measurable in the clinical setting, and having adequate psychometric properties. The DAS28 was one of four recommended RA disease activity measures (2).

The committee on RA remission considered several DAS28 thresholds as candidate definitions of remission, including the popular threshold, a DAS28 with CRP (DAS28‐CRP) score of less than 2.6, and an even lower threshold of less than 2.0. The DAS28 formula weights swollen joint count half as much as tender joint count and also underweights it relative to the CRP level (or the erythrocyte sedimentation rate [ESR]). Therefore, a patient can achieve a low DAS28 score but still have a substantial number of swollen joints. The committee’s analyses showed that 10% of patients with a DAS28 score of less than 2.6 had swollen joint counts of greater than or equal to 4, and one patient had more than 20 swollen joints. When a lower DAS28 threshold of less than 2.0 was used, swollen joint counts of 2 or 3 were common, and scores up to 6 were possible. In fact, if the tender joint count is 0, values of the other components of the DAS28 become nearly irrelevant (Figure 1). Values up to 60 (of 100) for patient global assessment are consistent with DAS28 remission. Even if the tender joint count is 1, the DAS28 score can still be in the remission range when other core set measures show active disease. DAS28‐CRP thresholds differ substantially from DAS28 with ESR (DAS28‐ESR) thresholds (3), and the DAS28‐ESR is even more likely to label patients in remission when disease is active.

Figure 1

The contribution of each component to DAS28 remission (score of <2.6) when other components are in the range of remission. DAS28‐CRP is an index composed of four components: CRP level (A), tender joint count (B), swollen joint count (C), and patient global assessment (D). Red dashed line represents TJC = 0 and blue dashed line represents TJC = 1. For each panel, we assumed that three components other than the one depicted met the threshold for remission (swollen joint count = 0; CRP level = 0.5; patient global assessment = 1; tender joint count = either 0 or 1). Note that when the tender joint count is 0, most values of CRP and patient global assessment yield a DAS28 score of less than 2.6 (remission), and values of swollen joint count up to 10 yield DAS28 remission. CRP, C‐reactive protein; DAS28, Disease Activity Score in 28 Joints.

One other major criterion was that those who achieved remission at 6 or 12 months in a 2‐year trial were likely to have both good and stable functional and radiographic outcomes later in that same trial. Patients achieving DAS28 remission had worse radiographic outcomes than those achieving other definitions of remission (no change in Sharp or Sharp van der Heijde scores). Ultimately, the committee rejected DAS28 candidates as definitions of remission because swollen joint counts were too high to be consistent with clinical remission and because DAS28 remission, even when using stricter thresholds, did not predict good combined functional and radiographic outcomes as well as the remission definitions selected by the committee.

Other studies conducted since the publication of the ACR/EULAR remission criteria provided additional evidence that the DAS28 should not be used to define remission. Saleem et al (4) showed that among those in DAS28 remission, power Doppler on ultrasound showed considerable disease activity unless the patient was also in SDAI remission. Lee et al (5) reported that joint pain was present and persisted in patients with RA in DAS28 remission but was absent if patients were in Boolean remission. Analyses from the AGREE trial of abatacept versus a placebo confirmed that those who achieved DAS28 remission had worse mean Health Assessment Questionnaire (HAQ) scores later than those who attained SDAI remission (6). Schoels and colleagues reported from an analysis of three large multicenter RA trials that among those with DAS28 scores less than 1.9, patients not in ACR/EULAR remission still had an average of two to three swollen joints (7).

Given the problems with using the DAS28 to define remission, why is it so widely used? First, the DAS28 is a commonly used disease activity measure, and it is easy to apply a threshold in data already being acquired, although the requisite elements of the ACR/EULAR definitions of remission are also acquired. Another potential reason relates to industry‐sponsored RA trials. A DAS28 score of less than 2.6 yields remission rates far higher than definitions endorsed by the ACR/EULAR, and treatments appear more efficacious when using the DAS28. Furthermore, using a definition that yields a higher remission rate improves statistical power. The same absolute difference in remission rates between two drugs is more likely to reach statistical significance when remission rates are higher. Lastly, the DAS28 use is mandated by some regulatory agencies. Many articles do not even include data on other measures of remission.

When remission definitions favor some treatments over others

An emerging concern is reliance on the CRP level to define remission (8). The CRP level is the second most heavily weighted variable in the DAS28 formula. Our armamentarium for RA includes effective biologic agents that have different effects on the CRP level: interleukin 6 and Janus kinase (JAK) inhibitors both directly reduce the CRP level, whereas abatacept and rituximab do not. If trials used the DAS28‐CRP to compare the efficacy of abatacept and JAK inhibitors, even if effects on joint counts and patient‐reported outcomes were the same, JAK inhibitors would score better, as seen in one recent trial (9). In another trial comparing biologic agents, authors acknowledged avoiding use of the DAS28‐CRP because of this bias (10). The ACR/EULAR provisional criteria made room for remission definitions that excluded acute phase reactants, using a three‐variable version of the Boolean definition and using the Clinical Disease Activity Index instead of the SDAI. Furthermore, although the full ACR/EULAR remission definitions include acute phase reactants, they are not weighted as heavily as in the DAS28‐CRP (or DAS28‐ESR).

Concerns about the inclusion of patient global assessment

Yet another concern about the provisional definitions of remission has been championed by Ferreira and colleagues (11). They point out that a patient’s global assessment of their arthritis activity often is based on considerations unrelated to disease activity, such as pain from joint damage, and that this measure should not be included in definitions of remission. The factors that most influence the patient global activity measure are pain and fatigue. Their analyses suggest that removing patient global assessment would not compromise the ability to predict later radiographic outcomes in RA, although they acknowledge that patient global assessment is a powerful predictor of function (as measured by the HAQ). High patient global assessment scores not only correlate with poor concurrent physical function but also identify patients who experience worsening physical function (12, 13). If patient global assessment is removed, remission criteria no longer predict future patient function well.

In addition to it being the only patient‐reported outcome measure included in remission definitions and the importance of including the patient perspective, there are other critical reasons to include patient global assessment as a component of remission. First, the patient global assessment reflects components of disease activity otherwise not captured, including fatigue and pain and inflammation in joints not counted in a 28‐joint count, such as the feet and ankles. That may be why high patient global assessment scores, even when 28‐joint counts are low, identify patients with a high risk of later function loss. Second, it is among the most sensitive, if not the most sensitive, outcome measures in RA (13). It improves much more on active RA treatment than on placebo, suggesting that it provides a window into disease activity related to systemic inflammation not detected by tender and swollen joint counts. Therefore, eliminating patient global assessments from RA trial outcomes would compromise our ability to distinguish the comparative efficacy of different treatments. This would occur at a time when, given the large armamentarium of treatments available, we particularly need to maximize our ability to differentiate their efficacy. In addition, inclusion of patient global assessment markedly increases the likelihood that patients attaining remission will have both good radiographic and functional outcomes later (Table 1) and it ensures that the definition of remission captures nonradiographic outcomes that are important to patients.

Table 1

Proportion of patients with good outcomes for both radiographic and functional outcomes in three multicenter rheumatoid arthritis trials

Candidate Remission Definition	Proportion of Patients With Good Outcomes a		Positive Likelihood Ratio (95% CI)
	In Remission	Not in Remission
TJC, SJC, and CRP all ≤1	46%	17%	3.2 (1.9‐5.3)
TJC, SJC, CRP, and patient global assessment ≤1	66%	17%	7.2 (3.5‐14.8)

Excluding patient global assessment compromises ability to predict good outcomes (1).

Abbreviations: CI, confidence interval; CRP, C‐reactive protein; SJC, swollen joint count; TJC, tender joint count.

Based on remission status at 6 months after baseline. Good radiographic outcome was defined as a change of 0 in Sharp/van der Heijde scores between 12 and 24 mo after baseline, and good outcome on the Health Assessment Questionnaire was defined as a change of 0 and a score of 0.5 at both the 12‐ and 24‐mo time points.

Conclusion

With remission achievable in RA, making the definition of remission stringent will ensure that patients benefit from comprehensive control of their disease. The DAS28 should not be used to define remission because, even when using low thresholds, many patients in remission will still have a number of swollen joints and active disease. Also, given its dependence on the CRP value, the use of the DAS28 makes it hard to differentiate efficacious treatments with dissimilar effects on acute phase reactants.

Defining remission without asking patients to provide any information about their disease activity risks losing valuable information on treatment efficacy, not to mention failing to collect any patient‐reported outcomes.

AUTHOR CONTRIBUTIONS

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

Disclosure Form

Click here for additional data file.