Evidence for predictive validity of remission on long-term outcome in rheumatoid arthritis: A systematic review

Authors


Abstract

Objective

Remission is rapidly becoming a key end point in rheumatoid arthritis (RA) clinical trials, but its definition is not satisfactory. Although it is generally believed that achieving a state of remission will lead to better structural outcome, this has not been studied systematically. As part of an undertaking to redefine remission, the current review describes the relationship between remission and long-term structural outcome.

Methods

A systematic literature search of PubMed, EMBase, and The Cochrane Library intersected 3 groups of terms: RA, remission, and long-term outcome. The search identified 1,138 records, of which 14 were relevant to the research question.

Results

All of the studies included in this review showed a relationship between remission and long-term structural damage or disability. Patients that achieved a state of remission, defined in various ways, showed less deterioration of function and radiographic progression compared with patients who did not reach a state of remission.

Conclusion

Patients who achieved a state of remission were less likely to show deterioration of function and radiographic progression compared with patients who did not reach a state of remission.

INTRODUCTION

Treatment of rheumatoid arthritis (RA) has improved substantially in the past 20 years. Although RA is a chronic disease without a known cure, a state of remission or low disease activity is achieved by a substantial number of patients participating in clinical trials (1–3), and even patients in practice (4, 5). In the Quantitative Patient Questionnaires in Standard Monitoring of Patients with Rheumatoid Arthritis project, remission occurred frequently, but this depended strongly on the definition used: percentages of patients in remission varied from 9% for the American College of Rheumatology (ACR; formerly the American Rheumatism Association) definition of remission (6) to 20% for the Disease Activity Score in 28 joints (DAS28) definition of remission. Likewise, using the Care for RA database (7), a recent study showed differences when different scores were employed, with the lowest proportions of remission observed with the Simplified Disease Activity Index (SDAI) followed by the ACR and DAS28 remission criteria; in DAS28 remission, patients had up to 9 swollen joints, whereas in SDAI remission 1–2 swollen joints were seen in only 5% of the patients (8). DAS28 remission has often been seen to exceed ACR 70% improvement (ACR70) rates (1, 9, 10), and sometimes even ACR50 rates (11), indicating significant residual disease activity in many patients in DAS28 remission. The findings of these papers highlight the need for a uniform definition of remission in RA.

In a collaborative project, the ACR and the European League Against Rheumatism (EULAR) have constituted a committee to redefine remission in RA. In its first publication, the committee concluded that a new remission definition should be strict, based on no or very low disease activity, and should be validated against long-term outcomes, specifically physical function and radiographic progression (12). The committee agreed that the new definition of remission should be able to predict patients who achieve and remain in a state of remission, who will have no or little progression of joint damage, and who will have less deterioration or more improvement in physical function. But is it possible to predict this favorable long-term outcome based on a definition of remission? What is the current body of knowledge that proves that being in a state of remission for a certain period will lead to better outcomes compared with not being in remission?

This systematic review aims at collecting and reviewing all available evidence on the ability of remission (defined variously) to predict long-term outcome (i.e., the stability of radiographic damage or the stability of physical function). In other words, what is known about the predictive validity of the currently available remission definitions?

MATERIALS AND METHODS

For this review, we sought all articles that studied the relationship between remission (by any definition) and long-term outcome in terms of function and radiologic progression.

Literature search.

A systematic literature search of PubMed (on October 6, 2008; primary search), EMBase (on October 20, 2008), and The Cochrane Library (on November 5, 2008) identified studies addressing the longitudinal relationship between any kind of remission definition and long-term RA outcome. The search did not include limitations for language. Search terms included medical subject headings or keywords when appropriate. The search intersected 3 groups of terms. The first group described the disease RA, the second group described the condition of remission according to any available definition, and the third group described the concept of long-term outcome in terms of function and radiologic progression. To be included in the review, the population studied must have been patients with RA, remission as well as long-term outcome must have been defined clearly, and the measurement of remission had either to be concurrent with or precede the interval over which the long-term outcome was assessed. Two authors (LHDvT and MB) composed the search strategy in consultation with medical information specialists. For the exact strategy, see Table 1.

Table 1. Search strategies*
Database, date accessedTopicResult, no. studies
  • *

    MeSH = medical subject headings; :NoExp = not exploded; tiab = title and abstract (PubMed); /de = descriptor, EMTree-term not exploded; /exp = EMTree term exploded; :ti,ab = title and abstract (EMBase); :ab,ti = abstract and/or title (Cochrane Library); /lim = limits; :ti,ab,kw = title and/or abstract and/or keywords.

PubMed, 10/6/2008  
 1Disease: rheumatoid arthritis 
  MeSH((“Arthritis, Rheumatoid”[MeSH:NoExp] OR “Caplan's Syndrome”[MeSH]) OR “Felty's Syndrome”[MeSH]) OR “Rheumatoid Nodule”[MeSH] 
 OR 
  Free Textarthriti*[tiab] OR rheuma*[tiab] OR caplan[tiab] OR felty[tiab]154,812
 2Remission: by any definition 
  MeSH(“Remission Induction”[MeSH] OR “Remission, Spontaneous”[MeSH]) 
 OR 
  Free Textremission[tiab] OR ((absence[tiab] OR minimal[tiab] OR low[tiab] OR no[tiab]) AND (disease activit*[tiab]))87,877
 3Long term RA outcome: function and radiographic progression 
  MeSH“Health Status Indicators”[MeSH] 
 OR 
  Free Texthaq[tiab] OR aims2[tiab] 
 OR 
  Free Text((sharp[tiab] OR heijde[tiab] OR vanderheijde[tiab] OR larson[tiab] OR genant[tiab]) AND (score[tiab] OR scoring[tiab] OR scores[tiab])) OR ((damage[tiab] AND radiograph*[tiab]) OR (radiologic*[tiab] AND progression[tiab])) OR disabilit*[tiab] OR “physical function”[tiab]182,241
 4Combination 
 #1 AND #2 AND #3643
EMBase, 10/20/2008  
 1‘rheumatoid arthritis’/de OR ‘felty syndrome’/exp OR ‘rheumatoid nodule’/exp OR ‘pneumoconiosis’/exp95,296
 2arthriti*:ti,ab OR rheuma*:ti,ab OR caplan:ti,ab OR felty:ti,ab156,926
 3‘remission’/exp37,423
 4remission:ab,ti OR ((absence:ab,ti OR minimal:ab,ti OR low:ab,ti OR no:ab,ti) AND (disease:ab,ti AND activit*:ab,ti))118,041
 5haq:ab,ti OR aims2:ab,ti OR ((sharp:ab,ti OR heijde:ab,ti OR vanderheijde:ab,ti OR larson:ab,ti OR genant:ab,ti) AND (score:ab,ti OR scoring:ab,ti OR scores:ab,ti)) OR ((damage:ab,ti AND radiograph*:ab,ti) OR (radiologic*:ab,ti AND progression:ab,ti)) OR disabilit*:ab,ti OR ‘physical function’:ab,ti86,823
 6‘structural damage’1,884
 7‘health status’/exp72,863
 8‘international classification of functioning, disability and health’/exp318
 9#5 OR #6 OR #7 OR #8156,976
 10(#1 OR #2) AND (#3 OR #4) AND #9 AND [embase]/lim727
Cochrane Library, 11/5/2008  
 1(arthriti* OR rheuma* OR caplan OR felty):ti,ab,kw and (remission OR ((absence OR minimal OR low OR no) AND (disease AND activit*))):ti,ab,kw and (haq OR aims2 OR ((sharp OR heijde OR vanderheijde OR larson OR genant) AND (score OR scoring OR scores)) OR (damage AND radiograph*) OR (radiologic* AND progression) OR disabilit* OR (physical AND function*) OR (health AND survey*) OR (structural AND damag*) OR (health AND status) OR (international AND classification AND functioning)):ti,ab,kw187

Data extraction.

The titles and abstracts of citations resulting from the search were screened for relevance by LHDvT. This resulted in an initial set from which LHDvT and MB independently selected the full articles to be retrieved. The same reviewers independently assessed these for final eligibility (Figure 1). Screening the references for articles identified other potentially relevant articles, which were subsequently retrieved and assessed. Lastly, all members of the ACR/EULAR committee charged with defining remission were asked to review the final list of relevant articles and to identify other work (published or not) that could be relevant to the research question.

Figure 1.

Flow chart of literature selection.

Data analysis.

Narrative summaries of the articles were compiled that highlighted the following characteristics:

the number of patients included in the study, their disease duration at inclusion, the type of intervention/treatment that the patients received during the study, the type of remission definition(s) used to classify patients as in or not in remission, the percentage of patients achieving remission, the duration of remission, the followup time of the study, and the assessment of outcome (lower radiologic progression and/or lower Health Assessment Questionnaire [HAQ] score).

RESULTS

The PubMed, EMBase, and Cochrane Library databases identified 1,138 possible articles referring to remission in RA in relation to long-term outcome. The process by which the final selection of 14 articles was made is shown in Figure 1. Searches in EMBase and The Cochrane Library did not provide additional articles. The expert committee provided 1 additional article and several relevant background articles. Checking the references of all of the relevant articles provided 2 additional articles.

Of the 14 studies, 5 used the DAS28 <2.6 remission definition; the SDAI <3.3, DAS <1.6, and a modified version of the ACR criteria were used 3 times each; the full ACR criteria were used 2 times; and the following definitions were each used 1 time: DAS28 <2.4, the rheumatologists' judgment, patient-reported remission, time-averaged DAS <1.6, DAS28 <2.6 for ≥1 year, DAS28 <3.2, DAS28 <2.84, and Clinical Disease Activity Index (CDAI) ≤2.8. Heterogeneity was also present in outcome measures. Specific details and sources of heterogeneity for all articles included in the review are presented in Table 2. Furthermore, all studies included in this review are summarized below, with articles discussing a similar research question grouped together into 1 of 3 themes: 1) studies reporting on the influence of remission on physical function in patients using disease-modifying antirheumatic drugs (DMARDs), 2) studies reporting on the influence of sustained remission on radiographic progression, and sometimes also physical function, in patients using DMARDs, and 3) studies reporting on the influence of sustained remission primarily on radiographic progression in patients using DMARDs and biologic agents.

Table 2. Description of the 14 included studies, all finding a positive relationship between remission and outcome*
Study author, year (ref.)No.Disease duration, mean yearsStudy typeInterventionRemission definitionPatients in remission, %Remission durationFollowup, yearsRemission beneficial for outcome measureContrast
Radiologic progressionHAQ scores
  • *

    HAQ = Health Assessment Questionnaire; RA = rheumatologist; DMARDs = disease-modifying antirheumatic drugs; GCs = glucocorticoids; ACR = American College of Rheumatology; mACR = modified ACR criteria (often excluding fatigue); ESR = erythrocyte sedimentation rate; DAS = Disease Activity Score; SvH = Sharp/van der Heijde; SDAI = Simplified Disease Activity Index; DAS28 = DAS in 28 joints; RCT = randomized controlled trial; pred = prednisone; SSZ = sulfasalazine; MTX = methotrexate; HCQ = hydroxychloroquine; anti-TNF = anti–tumor necrosis factor; etan = etanercept; inflix = infliximab; CDAI = Clinical Disease Activity Index; ada = adalimumab.

  • Median years.

Laborie et al, 2002 (13)8820Retrospective cohortRA's choice: DMARDs and GCs1) Patient-reported1) 29Mean 8.5 monthsMedianPatients in vs. not in remission
2) Patient-reported ACR criteria2) O7
Lindqvist et al, 2002 (14)1680.9CohortRA's choice: DMARDs and GCsmACR79Minimum 1 time point10Mean for patients after 10 yearsPatients in relapsing– remitting disease vs. continuously active disease
Molenaar et al, 2004 (15)1877CohortRA's choice: DMARDs1) mACR + ESR1) 100 with 52 sustained2.5 years2% patients with Δ SvH ≥5Patients in sustained remission vs. with exacerbations
2) mACR2) 59 with 59 sustained
3) DAS <1.63) 81 with 42 sustained
Aletaha et al, 2005 (18)56?Inception cohortUnknown1) SDAI ≤3.3?Never vs. 1) 3–18 months3Change in Larson score between baseline and 3 years3 disease activity groups
2) DAS28 <2.42) 21–36 months
Cohen et al, 2007 (19)1340.7CohortDMARDsDAS <1.622 at 3 and 5 years2 years5Change in SvH score at baseline, 3 years, and 5 yearsContinuous and dichotomousPatients in remission vs. not in remission
Brown et al, 2008 (20)907CohortDMARDs1) RA's judgment1) 100Median 2 years before inclusion1Change in Genant-modified Sharp score over 1 yearPatients in RA-determined remission vs. different assessment technique
2) ACR criteria2) 54
3) DAS28 <2.63) 56
Häfstrom et al, 2008 (21)1500.5RCT extensiona) DMARDsDAS28 <2.655 at 2 years, 40 at 4 yearsMinimum 1 time point4Change in SvH and absolute scores at 0, 2, and 4 yearsAt 2 and 4 yearsPatients in remission vs. not in remission at 2 years
b) DMARDs + pred
Mäkinen et al, 2007 (22)169<2RCTa) SSZ1) mACR1) 8 sustained vs. 92 not sustained1.5 years2Change in Larson score between 0, 6, 12, and 24 monthsPatients not in remission, in remission, and in sustained remission
2) DAS28 <2.62) 32 sustained vs. 68 not sustained
Machold et al, 2007 (23)55<0.3CohortRA's choice: DMARDs + GCs + (in 4 patients) anti-TNF1) DAS28 <2.6 for ≥1 year1) 291 year3Development of erosions yes/no, score, Larson/ScottPatients in 4 different disease activity categories
2) DAS28 <3.22) ?
Landewé et al, 2006 (24)4447RCTa) MTXTime-averaged DAS (12 months <1.6)9Mean 1 year1Annual change in SvH score, calculated from 0-, 1-, and 2-year dataPatients in 4 different disease activity categories
b) Etan
c) MTX + etan
Van der Heijde et al, 2008 (25)3327RCTa) MTX1) DAS <1.61) 1 year: 28, 2 years: 37, 3 years: 38Minimum 1 time point3Change in SvH scorePatients reaching DAS remission vs. not reaching this state
b) Etan2) DAS28 <2.6 
c) MTX + etan 2) 1 year: 30, 2 years: 38, 3 years: 41
Smolen et al, 2006 (26)1,0040.9RCTa) MTXDAS28 <2.8430 at week 14Minimum 1 time point1Change in SvH between 0 and 54 weeksPatients in different disease activity states categorized by DAS28 tertiles
b) MTX + inflix 3 mg/kg
c) MTX + inflix 6 mg/kg
Smolen et al, 2008 (27)1,0490.9RCTa) MTX1) SDAI ≤3.31) 9 at week 14, 20 at week 54Minimum 1 time point1Change in SvH between 0 and 54 weeksPatients reaching remission at 14 weeks, at 54 weeks, or no remission
b) MTX + inflix 3 mg/kg2) CDAI ≤2.8
c) MTX + inflix 6 mg/kg3) DAS28 <2.6
Aletaha et al, 2009 (28)5100.7RCTa) MTXSDAI ≤3.3231 year vs. preceding 3, 6, or 9 months2Change in total Sharp score during the second year of study while in remissionPatients with another 3, 6, or 9 months in remission
b) Ada
c) MTX + ada

All studies demonstrated that patients who achieved a state of remission, defined in various ways, showed less deterioration of function and radiographic progression compared with patients who did not reach a state of remission.

Influence of remission on physical function in patients using DMARDs.

Laborie et al (13) performed a cross-sectional questionnaire study to assess very long-term clinical and functional outcomes in patients with RA. Of 88 patients, 26 reported a current subjective remission, with a mean ± SD duration of 8.5 ± 5.9 months. Of these patients, 20% met the ACR remission criteria. The mean ± SD HAQ score for all patients was 1.11 ± 0.84. The median HAQ score was somewhat lower in the subgroup reporting remission, although considerable overlap occurred with the other patients.

Lindqvist et al (14) studied the 10-year outcome in a prospective cohort of patients with early RA. They showed that 133 patients (79%) had relapsing–remitting disease (no active joints on examination at ≥1 followup visit during 10 years). These patients had a significantly lower HAQ score (P = 0.013) after 10 years than patients with continuous active disease (data not shown). At the 10-year followup visit, 30 patients (18%) were in remission for a median of 13 months.

Influence of sustained remission on radiographic progression, and sometimes also physical function, in patients using DMARDs.

Molenaar et al (15) studied the radiographic progression in patients with RA in clinical remission for at least the preceding 6 months. Clinical remission was defined as a normal erythrocyte sedimentation rate and meeting 3 of the 4 remaining ACR remission criteria (fatigue was not measured). In the study period, the ACR remission definition (meeting 5 of the 5 ACR remission criteria) and the original DAS definition (DAS <1.6) were applied. Radiographs of hands and feet were obtained at baseline, 1 year, and 2 years, and were scored according to the Sharp/van der Heijde method (16, 17). Of the 187 included patients, remission persisted in 52% of the total cohort, in 59% of the patients in ACR remission, and in 42% of the patients in DAS remission. Compared with patients with an exacerbation during followup, patients in persistent clinical remission were less likely (7% versus 23%; P = 0.05) to have relevant radiographic progression (change in Sharp/van der Heijde ≥5). For the ACR criteria, these percentages were 7% for the remission group versus 10% for the patients with an exacerbation (P = 0.053) (3), and for the DAS were 6% versus 17% (P = 0.017). Of the patients in persistent clinical remission (as defined by the authors) for 2 years, 15% developed erosions in a previously unaffected joint. The authors concluded that radiologic progression could occur during all 3 defined states of persistent clinical remission.

Aletaha et al (18) studied, among other things, the construct validity of the SDAI and the DAS28 by evaluating the association between the time spent in SDAI- and DAS28- defined remission, and radiographic progression over 36 months in a cohort of 56 patients with RA. In this study, remission was defined as an SDAI score ≤3.3 or a DAS28 <2.4. These cutoff values were calculated based on the expert opinion of 35 rheumatologists who judged the disease activity of paper patients (from 32 different patient profiles). Three groups were compared: patients never in remission, patients 3–18 months in remission, and patients 21–36 months in remission. They found that the longer the patient remained in SDAI remission or DAS28 remission, the less likely the patient was to progress radiologically (P < 0.01). A similar relationship was observed with patients in high disease activity: the longer that patients remained in high disease activity, the more likely they were to progress radiologically.

Cohen et al (19) studied the radiographic progression over 5 years of patients with early RA who reached sustained remission (DAS <1.6 at the 3-year and 5-year evaluations). Radiographs and HAQ scores were obtained at baseline, 3 years, and 5 years. A total of 134 patients were included, of whom 36% achieved remission at 3 years, 28% achieved remission at 5 years, and 30 patients (22%) were in sustained remission at both 3 and 5 years. This last group was compared with the 104 patients who did not achieve remission at either 3 or 5 years. The total Sharp/van der Heijde scores (with interquartile range [IQR]) at baseline in the remission and nonremission groups were 0.5 (IQR 0–7) and 2 (IQR 0–7), respectively. At 5 years they were 2.5 (IQR 0–14) and 13 (IQR 3–29), respectively. The progression of damage at 5 years was significantly higher in the group with persistent disease activity (Δtotal Sharp/van der Heijde score 7 [IQR 1–19]) compared with the sustained remission group (Δtotal Sharp/van der Heijde score 1.5 [0–5]; P = 0.0014). Still, 10 patients (33%) in sustained remission had a significant increase in radiographic damage between baseline and 5 years. This proportion was 55% (57 patients) for those who were not in remission. Furthermore, 5 patients (17%) in sustained remission had significant radiographic progression between the third and fifth year, and 6 patients developed erosions between the third and fifth year in a previously unaffected joint. Mean ± SD HAQ scores were already better at baseline in the remission group: 1.1 ± 0.7 versus 1.4 ± 0.7 for the nonremission group. At 3 and 5 years, the mean ± SD HAQ scores were 0.2 ± 0.4 and 0.1 ± 0.3, respectively, for the remission group, as opposed to 0.7 ± 0.6 at both times for the nonremission group. A total of 28 patients (93%) in sustained remission had a HAQ score ≤0.5 at 5 years versus 39 patients (38%) with persistent disease activity (P < 0.001). The authors concluded that sustained clinical remission according to the DAS <1.6 criteria was associated with stability of radiologic damage in most patients and a clear improvement of functional capacity over 5 years in a cohort of patients with early RA.

Brown et al (20) evaluated the long-term significance of subclinical synovitis measured by imaging techniques in patients who experienced clinical remission with conventional DMARD therapy, in relation to structural outcome. A group of 90 patients with RA in remission at baseline were followed for 12 months, and radiographs, magnetic resonance imaging (MRI), and ultrasonography were available at both time points. Remission was defined by physician's clinical judgment and had to be stable for ≥6 months; 54% fulfilled the ACR remission criteria and 56% fulfilled the DAS28 remission criteria (DAS28 <2.6). A control group of 17 healthy subjects underwent the same imaging procedures. At 12 months, 17 (19%) of 90 patients experienced a significant deterioration in their total radiographic joint damage scores that was larger than the smallest detectible change. Likewise, 11% of the patients in ACR remission and 12% in DAS28 remission also showed radiographic progression (data not shown). In a binary logistic regression analysis, the authors found that ACR remission protected for adverse structural outcome radiographically (odds ratio 0.33, P = 0.054). Furthermore, of 25 patients without painful, tender, or swollen joints, 4 experienced radiographic progression. The authors concluded that structural progression occured in patients with RA treated with conventional DMARD therapy despite the presence of clinical remission.

Hafström et al (21) studied whether remission (DAS28 <2.6) induced by low-dose glucocorticoid treatment in the first 2 years of the Better Anti-Rheumatic Farmacotherapy study had a sustained effect on radiologic damage after 4 years. At 2 years, 35 patients (55%) in the prednisolone group and 26 (30%) in the no prednisolone group were in remission. At 4 years, 40% of all 150 patients were in remission. Patients in remission at 2 years had significantly less radiographic damage and less change in radiologic scores after 4 years compared with those not in remission. Longitudinal analysis showed that DAS28 remission at 2 years was significantly associated with lower Sharp scores during the entire course of observation. The HAQ scores at both 2 and 4 years were significantly lower for the patients who were in remission at 2 years, irrespective of intervention group.

Mäkinen et al (22) reported the effect of sustained remission on radiographic outcome over 2 years in the Finnish Rheumatoid Arthritis Combination Therapy trial. Remission was defined by a modified ACR definition of remission, excluding fatigue, and by the DAS28 <2.6. Sustained remission was defined as remission at 6, 12, and 24 months; radiographs were obtained at baseline, 6 months, and 24 months. At 6 months, 31 patients were in modified ACR remission and 72 patients were in DAS28 remission. At 6, 12, and 24 months, 14 patients were in sustained modified ACR remission, compared with 54 patients who were in sustained DAS28 remission. The authors compared the radiographic progression of the patients in sustained remission, of the patients in remission at 6 but not at 12 or 24 months, and of the patients not in remission at 6 months. For the modified ACR definition of remission, they found a median increase in Larsen score of 0 (95% confidence interval [95% CI] 0–2) in patients in sustained remission, of 4 (95% CI 0–10) for patients in remission at 6 months but losing it later, and of 4 (95% CI 2–8) for patients not in remission at 6 months. For the DAS28 definition of remission, the Larson scores were 1 (95% CI 0–2), 4 (95% CI 2–16), and 6 (95% CI 2–10), respectively. The authors concluded that sustained remission protects against radiographic joint damage.

Influence of sustained remission on radiographic progression in patients using DMARDs and biologic agents.

Machold et al (23) performed a study to determine the frequency of erosive disease and the pace of radiologic progression in an inception cohort of patients with very early RA, treated at the discretion of the treating physician. In a subgroup of 55 patients who completed 3 years of followup, 64% developed erosive disease. Longstanding DAS28 remission was seen in 16 patients and was associated significantly with the absence of erosive disease (5 [31%] of these patients had erosive disease, 2 at baseline and 3 developed erosions during the first year and did not progress any further). Time in remission/low disease activity (DAS28 <3.2) correlated negatively with progression in Larsen score. In a stepwise regression model, time in remission/low disease activity showed a significant contribution to the overall outcome of damage progression, although the factors with the greatest influence were rheumatoid factor, anti–citrullinated peptide, C-reactive protein level, and cumulative swollen joint count. DMARDs were used by all patients except 2; 4 patients (all in the erosive group) received tumor necrosis factor α (TNFα) antagonists. The authors concluded that joint damage occurred despite very early treatment.

Landewé et al and van der Heijde et al (24, 25) described the radiographic progression in patients included in the Trial of Etanercept and Methotrexate with Radiographic Patient Outcomes study receiving either methotrexate (MTX) monotherapy, etanercept monotherapy, or a combination of both. Landewé and colleagues studied the relationship between disease activity and 2-year radiographic progression in this data set, and found that patients in remission (defined as a time-averaged DAS <1.4 during 1 year) showed less radiologic progression than patients with a time-averaged DAS >3.7. They demonstrated that the relationship most often seen (remission or low disease activity leading to less radiographic progression) was not present in the subgroup of patients who were treated with a combination of MTX and etanercept. They suggested that inflammation is not the only cause of radiographic progression, and thus that remission or low disease activity do not necessarily lead to less radiographic damage. Van der Heijde and colleagues studied the radiographic progression in patients reaching DAS <1.6 and DAS28 <2.6 remission. Although 682 patients started the study drug, 350 patients were withdrawn during the 3 years. DAS/DAS28-defined remissions were reached by 145 patients (28%)/152 patients (30%) in year 1, by 154 patients (37%)/157 patients (38%) in year 2, and by 117 patients (38%)/124 patients (41%) in year 3 (completers only). In patients who achieved DAS remission, radiographic remission (change in total Sharp score ≤0.5) was also attained in 85%, 80%, and 56% of those in the combination, etanercept, and MTX groups, respectively. Conversely, fewer patients in the group not achieving DAS remission showed stability of radiographic progression (change in total Sharp score >0.5): 34%, 45%, and 58% of patients in the respective treatment groups.

In 2006, Smolen et al (26) reported on data from the Active-Controlled Study of Patients Receiving Infliximab for the Treatment of Rheumatoid Arthritis of Early Onset (ASPIRE) trial, with disease activity divided into tertiles with the lowest tertile a DAS <2.85. In the group that was treated with MTX monotherapy, patients with a higher DAS at 14 weeks had more radiologic progression between baseline and week 54 than those with a lower DAS (P < 0.05). There was little progression in the MTX + infliximab group, even in patients in the highest DAS28 tertile. Analyses using SDAI tertiles to define disease activity produced similar results. The authors concluded that higher disease activity was associated with greater radiographic damage in the MTX group, whereas little progression occurred in the combination group.

In 2008, Smolen et al (27) reported on data from the ASPIRE trial, studying the effectiveness of MTX alone compared with MTX + infliximab. Disease activity was classified into 4 states, with an SDAI score ≤3.3 defined as remission. Of the 1,049 patients who started the study, 155 patients (15%) reached SDAI remission at 54 weeks, and the radiographic progression of these patients was lower than that of the patients in the other disease activity states. In general, patients treated with MTX + infliximab showed less radiographic progression across all disease activity states, but in patients using MTX monotherapy, joint damage increased with every disease activity category. At 14 weeks, 78 patients (9%) achieved SDAI remission, and these patients had very limited radiographic progression at week 54, even with MTX. This study showed that sustained remission even with MTX (if seen already at week 14) is associated with no radiographic progression on the group level.

In 2009, Aletaha et al (28) studied the rate of radiologic progression in patients in sustained remission in a post hoc analysis of the PREMIER trial (comparing effectiveness of either MTX monotherapy, adalimumab monotherapy, or a combination of both). They studied the radiographic damage in a subgroup of patients who were in sustained remission defined by an SDAI score ≤3.3 during months 12 and 24 of the PREMIER study, and compared the groups with an additional preceding remission period of either 3, 6, or 9 months. They found that the shorter the period of remission, the more likely it was for some mild progression to be found, which was deemed a consequence of a carry-over effect of past periods of inflammation. The authors concluded that preceding disease activity is a relevant determinant for radiographic progression between 2 time points.

DISCUSSION

This systematic review identified only 14 studies reporting on the longitudinal relationship between a definition of remission and the long-term outcome of RA in terms of function and/or radiographic progression. Fortunately, all studies showed that patients who achieved a state of remission, defined in various ways, showed less deterioration of function and radiographic progression compared with patients who did not reach a state of remission. It is remarkable that there are only 14 studies that report on this subject, although these data are probably available in many followup studies of clinical trials.

Many studies had to be excluded because they reported on the relationship between remission and long-term outcome at the same time point, studied the relationship between disease activity and long-term outcome without defining remission, or collected all of the relevant variables but did not look at the relationship between them. Although no limitations for year of publication were used, the oldest articles included in this review are from 2002, by Laborie et al and Lindqvist et al (13, 14). We expected to find older articles because the first official remission definition was proposed by Pinals et al in 1981 (29). However, remission did not occur often before the introduction of combination DMARD therapies and anti-TNF strategies in the last decade.

Although all of the reviewed articles examined remission and stability of longer-term outcomes, the temporal relationship of remission to those longer-term outcomes varied a bit from study to study. Most of the reviewed studies assessed remission during the period when longer-term stability was also being assessed. Because change in longer-term outcomes was occurring at the time that remission was assessed, these studies might not be considered longitudinal. We are not sure whether any of these studies evaluated remission preceding the period in which long-term outcomes were assessed (e.g., remission at 6 months; long-term outcomes from 6 months to 2 years).

Because the topic of this review is often not the primary goal of a study, our search strategy had to be sensitive enough to find the eligible articles, even when the topic was not mentioned in the abstract. Although we used a sensitive search strategy, it is possible that relevant articles were missed because these findings were reported in a subanalysis. Nevertheless, we believe that we were comprehensive in our search, especially by including a round with the experts currently designing a new definition for remission in RA, who could not identify any additional relevant studies. The consistency of the results of the included articles makes it very likely that this is the true direction of the signal, with those patients achieving a state of remission showing less deterioration of radiographic progression and function.

It is important to report on this specific research question because increasing numbers of trials use remission as an important outcome, assuming that this leads to favorable long-term outcomes. The current review confirms that there is evidence for a longitudinal relationship between remission and long-term RA outcome, but more research is needed to confirm this finding.

Another interesting development in the relationship between remission and radiographic progression is the so-called disconnect between disease activity and radiologic damage as hypothesized by Kirwan et al (24, 26, 27, 30, 31). According to this hypothesis, disease activity (or inflammation) only partially explains long-term radiologic damage. The articles by Landewé et al (24) and Smolen et al (27) included in this review strongly suggest that most patients treated with biologic therapies experience a virtual halt of radiographic progression, regardless of disease activity, whereas in patients treated with monotherapy there was increasing joint damage with increasing disease activity. In fact, all articles reporting some residual radiologic progression in patients in remission lend support to the disconnect hypothesis. However, even with biologic agents, a slight progression of joint damage can be observed with increasing disease activity, and especially with high disease activity. Moreover, methods for identifying progression are insensitive in the setting of very low progression rates, being that when the progression is close to zero, the method is unreliable to measure this, since the smallest detectable difference is ∼5 Sharp/van der Heijde points (32). Furthermore, clinical methods are also unreliable in the setting of very low disease activity. The study by Brown et al (20) included in this review shows that radiographic progression in patients in clinical remission is mostly explained by subclinical joint inflammation detected by ultrasound and MRI. It therefore remains possible that the reduction in association is in fact explained by subclinical synovitis occurring less frequently in patients in remission induced by biologic or combination therapies than in patients in remission induced by monotherapy. However, the studies by Aletaha et al (28) and Smolen et al (27) suggest that it is primarily the sustained remission that will allow joint damage to halt, indicating that either a carry-over effect occurs or that subclinical disease activity may still be present, driving joint damage before it is ultimately arrested. One way or the other, the validation of a remission definition based on clinical measurement must take into account the reduced association between disease activity and structural damage demonstrated by these studies.

Although all remission definitions lead to reduced radiologic progression, some definitions still allow significant residual disease activity, whereas others are more strict, allowing no swollen or tender joints. This was presented at the first meeting of the ACR/EULAR remission workgroup (13), where it was summarized that remission definitions can roughly be categorized as either strict (ACR, CDAI/SDAI, Patient Activity Scale/Routine Assessment of Patient Index Data 3) or lax (the modified ACR criteria and the DAS28 definition), with the latter being very similar to the Outcome Measures in Rheumatology Clinical Trials definition for minimal disease activity (9). It has been shown that the exclusion of the ankles and feet in reduced joint counts can give different results and should be used with caution in clinical practice (33, 34). This might explain at least a proportion of the joint damage that occurs in such states of remission.

As can be learned from this review, the current concept of remission is complex and the development of a new definition is a big challenge. This review, albeit of a very limited number of studies, offers consistent evidence that remission by all currently applied definitions is longitudinally related to radiologic progression and physical function.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be submitted for publication. Dr. van Tuyl had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Van Tuyl, Felson, Wells, Boers.

Acquisition of data. Van Tuyl, Smolen, Boers.

Analysis and interpretation of data. Van Tuyl, Wells, Zhang, Boers.

ROLE OF THE STUDY SPONSOR

Abbott, Schering-Plough, Wyeth, Bristol-Myers Squibb, and Roche had no role in the study design, data collection, data analysis, and writing of the manuscript, nor was their approval needed for the content of the submitted manuscript.

Acknowledgements

We would like to thank Hans Ket, medical information specialist at the medical library of the VU University Medical Center, Amsterdam, The Netherlands and Jessie McGowan, senior information scientist at University of Ottawa, Ottawa, Ontario, Canada for their help in formulating a good search strategy.

Ancillary