Comparable Clinical and Radiographic Outcomes among DAS28-ESR-based Remission Criteria and SDAI, CDAI, and ACR/EULAR Boolean definitions in patients with Established Rheumatoid Arthritis in Clinical Practice. Observational Study of 5 years of Follow-Up


 Objective To compare long-term clinical and radiographic outcomes among five sets of remission criteria [four clinical and one Ultrasound (US)-based] in a cohort of RA patients in a clinical care setting. Methods RA patients in remission (DAS28-ESR <2.6) were selected. Hand US assessments were performed, and serum levels of inflammation/angiogenesis biomarkers were determined at baseline. Changes in baseline treatment and radiographic progression, defined as the variation in the modified Sharp van der Heijde score (mSHS) at 5 years, were analyzed. To define remission, five different concepts were used, as follows: DAS28-ESR<2.6, SDAI<3.3, CDAI<2.8, Boolean criteria and score Power Doppler (PD)=0.Results Eighty-seven patients with DAS28-ESR<2.6 were included. One third fulfilled SDAI (33.3%), CDAI (31%) and Boolean (35.6%) remission criteria and 25.3% had no PD signal in the US evaluation. 26 patients (29.9%) changed the therapy, ranging from 13.6% (PD remission) to 33.3% (CDAI remission) (p=0.11). Serum levels of ANG (p=0.015) and TNFa (p=0.025) were significantly lower in patients with Boolean remission, whereas IL-18 levels were significantly lower in those with PD remission (p=0.049). Patients without PD in the US assessment had significantly lower mSHS erosion progression (p=0.014) at 5 years. Conclusions Patients with established RA in DAS28-ESR remission had comparable clinical and radiographic outcomes than SDAI, CDAI and Boolean definitions in a clinical care setting. US remission remained as the closest to structural damage abrogation.


Introduction
The main goal in the treatment of rheumatoid arthritis (RA) is to suppress the in ammatory process and achieve remission. Remission may be de ned as a state with no or, at least, low disease activity. Ideally, remission should be maintained to abrogate joint damage progression [1].
Several sets of remission criteria have been proposed. The original American Rheumatism Association (ARA) remission criteria are infrequently used today since all components of the criteria are not included in the current core set of variables [2]. A Disease Activity Score (DAS) less than 1.6 was found to correspond well to the ARA remission criteria and was proposed as a remission criterion. Later, DAS remission was modi ed by a 28-joint count to the DAS28-ESR less than 2.6 criterion, which has been widely accepted [3][4]. However, DAS28-ESR criterion has been criticized as being not stringent enough, as many patients under DAS28-ESR remission can still have low levels of disease activity. Therefore, more stringent criteria have been developed, for example, the Simpli ed Disease Activity Index (SDAI) less than 3.3 remission criterion and the Disease Activity Index less than 2.8 remission criterion (CDAI), which lacks acute phase reactants [5]. Recently, the American College of Rheumatology (ACR) and the European League Against Rheumatism (EULAR), have jointly proposed that remission in RA may be de ned either according to the SDAI or to the new Boolean-based set of criteria [6]. The Boolean Criteria have been shown to perform well in clinical trials, but their utility in long-term observational studies remains to be demonstrated [6].
We have previously demonstrated that around 60% of RA patients in DAS28-ESR remission have Power Doppler (PD) signal [7]. This subclinical ultrasound (US) activity has been related to higher structural damage progression after 12 months of follow-up [8].
The aim of this study was to compare clinical and radiographic outcomes among four different clinical sets of remission criteria and one remission criteria based on imaging (US) in a cohort of established RA patients in a clinical setting with 5 years of follow-up.

Patients And Methods
Observational study. Consecutive patients from the Rheumatology Department (Hospital Clinic, Barcelona, Spain) meeting ACR/EULAR 2010 criteria for RA [9] in clinical remission (de ned as DAS28-ESR < 2.6) for > 6 months assessed by two rheumatologists were included. Clinical, epidemiological and serological data were analyzed. An ultrasound assessment of hands was performed, and serum biomarkers of in ammation and angiogenesis were determined at baseline. Patients were followed up for 5 years. Clinical and radiological data (radiographs of hands and feet) were collected.
The study was approved by the Ethics Committee of the Hospital Clinic of Barcelona (HCB-2017-0562). Signed informed consent was obtained from all patients.
Treatment change was de ned as the change in any DMARD therapy (including synthetic or biologic DMARD but not glucocorticoids), across the observational period of 5 years.
Radiographic progression was de ned with both, quantitative, as the difference between the last and rst reading, and a qualitative concept based on the small detectable change, as it is explained below.
The proportion of patients with treatment change and radiographic progression throughout 5 years was calculated and compared across the ve sets of remission criteria.

Ultrasound assessment
All sonographic assessments were performed using high-sensitivity ultrasound equipment (Acuson Antares®, Siemens AG, Erlangen, Germany). Sonographic assessments were made using a frequency range from 8 to 12 MHz. We used the same protocol detailed in previous works [7][8].
Intra-rater agreement was 0.81 for Synovial Hypertrophy and 0.92 for PD. This index was calculated as the percentage of agreement between these scores at two time points.

Radiographic assessment
All patients underwent postero-anterior X-rays of the hands and feet at baseline and at 5 years.
Radiographic changes were graded anonymously by 2 trained readers (JR and AC), according to the modi ed Sharp van der Heijde score (mSHS) [10]. Intra-observer and inter-observer reliability were considered moderate to good; We analyzed radiographic progression using a double approach.
First, we quantitatively measured the radiographic progression as the difference between the nal and baseline score using both the total mSHS and the erosion sub-index score.
Secondly, using a qualitative approach, radiographic progression was de ned as the change in the mSHS at 5 years > 10.47 [mdc (minimum detectable change)]. Radiographic evidence of structural progression was de ned as a variation in the mSHS greater than the mdc, which was calculated using the following formula [11]: mdc = 2 × SD ([obs a followup-obs a baseline] -[obs b followup -obs b baseline]) Once de ned the value of mdc, the value for each patient was calculated. If this value was higher than mdc we considered this patient as progressor. Using the same formula, we calculated the radiographic erosion progression, nally de ned as 6.98.

Quanti cation of biomarkers of in ammation/angiogenesis
Cytokines and angiogenic mediators were analyzed using Quantibody® Human Custom Array (RayBiotech, Norcross, GA, USA). Each sample was diluted 2-fold and prepared in quadruplicate. An Axon scanner 4000B with GenePix software was used to collect uorescence intensities. Detection limits for cytokines are displayed on the manufacturer's website [RayBiotech [http://www.raybiotech.com]. After sample dilution, the effect of Rheumatoid Factor (RF) on the nal results was estimated to be around 1% [12].

Statistical analysis
Categorical and quantitative variables were described as frequencies, percentage and mean ± standard deviation (SD). Clinical and radiographic outcomes were compared among the ve remission criteria sets in a direct head-to-head comparison by the non-parametric Mann-Whitney U test, T-student or X 2 test, as appropriate. Subsequently, we compared serum levels of biomarkers in a direct head-to-head comparison between the ve groups using non-parametric tests. A value of p ≤ 0.05 was considered statistically signi cant. Statistical analysis was performed using SPSS V. 20.0 (SPSS. Chicago, Illinois, USA).

Angiogenic and In ammatory biomarkers
Serum levels of ANG (p = 0.015) and TNFa (p = 0.025) were signi cantly lower in patients in SDAI remission and IL18 serum levels were signi cantly lower in patients in PD remission (p = 0.049) (Fig. 1).
No other remarkable differences were found in serum levels of angiogenic and in ammatory biomarkers across de ve sets of remission criteria. Of remark, serum levels of in ammatory biomarkers such as calprotectin, IL-6 or IL17F were balanced across the ve sets of remission criteria.

Radiographic Progression
Globally, radiographic progression was low in all groups, with a mean mSHS progression slightly above 1 unit per year. Patients in PD remission had the lowest mSHS progression across groups. Remarkably, we found that patients in PD remission had signi cantly lower mSHS erosion progression (mean 1.36, SD 1.94, p = 0.014) after 5 years (Fig. 2).
Using a qualitative approach [mSHS threshold of 10.47(mdc)], radiographic progression ranged from 4.5% of patients in PD remission to 14.8% of patients in CDAI remission. Erosion progression above 6.98 (mdc) ranged from 4.5% for patients in PD remission to 16.1% for patients in Boolean remission. No signi cant differences were found among them. When a less stringent cut-off was used for determining radiographic progression (> 5, 1 unit/year), range raised to 47.1-61.3% for mSHS progression (p-value non-signi cant) and 4.5-38.7% for mSHS erosion progression (p = 0.005 for PD remission) ( Table 3). Changes of baseline therapy and radiographic progression were numerically lower in patients with sustained remission either at 12 month or at 5 years (data not shown) although the reduced population hampered comparisons between groups.

Discussion
The advent of biologic agents has allowed to reach new and ambitious outcomes in RA. Consistently, the cornerstone of the Treat To Target recommendations was the de nition of an achievable treatment target, i.e. remission or at least low-disease activity [1].
DAS28-ESR < 2.6 has been the de nition most widely used for remission in clinical practice. However, it has been suggested that disease activity could still be present in patients ful lling that criterion. New ACR/EULAR de nitions are more stringent and fewer patients are classi ed as being in remission using these concepts. Accordingly, barely half of patients with DAS28-ESR < 2.6 ful lled SDAI or ACR/EULAR Boolean remission criteria in clinical practice [14]. However, whether or not more stringent remission criteria provide better clinical and radiographic outcomes in the long term is still under debate. Selecting more stringent remission criteria implies, on one hand, fewer joint counts and less subclinical activity [15][16]. On the other hand, the more stringent is our target, the more we will step up the therapy to achieve the goal, with a subsequent higher risk of adverse events [17].
In this observational study of clinical practice, DAS28-ESR criterion identi ed remission three-fold as often as SDAI, CDAI, ACR/EULAR Boolean and PD remission criteria.
The number of patients classi ed in remission (27-31%) and the rate of change of therapy (31-33.3%) were not essentially different among SDAI, CDAI and Boolean criteria. Even radiographic progression, no matter the threshold used, was quite balanced among them. Therefore, ACR/EULAR remission de nitions could be interchangeable, as it has been previously highlighted [18].
Remarkably, no signi cant differences were reached among SDAI, CDAI, Boolean criteria and DAS28-ESR criterion with regard to change of baseline therapy after 5 years of follow-up. Even radiographic progression was not substantially different. Previous studies also showed that the DAS28-ESR remission criterion had similar radiographic damage and disability than that seen with other criteria, and patientreported outcomes were similar [19][20]. The global low level of radiographic progression, barely above one unit of mSHS per year, in part explained by the high use of biological therapy at baseline (60%), de nitely in uenced the low structural damage progression and hampered the comparison among groups. According to our results, patients in remission, whatever the concept used to de ne it, have no relevant radiological progression in the long-term.
The two main drawbacks of DAS28-ESR-based remission are, on one hand, the excessive weight in the formula for acute phase reactants, giving an advantage to IL-6-inhibiting agents [21]. In this particular case, indexes such as CDAI or RAPID-3, not taking into account acute phase reactants, should be of rst election [22][23]. On the other hand, a substantial amount of subclinical in ammation persists in patients in DAS28-ESR remission [24]. It is important to highlight that clinical remission, no matter the de nition employed, does not entirely correspond to imaging remission and some patients might experience structural damage progression [25][26][27]. However, very few patients of our cohort had signi cant radiographic progression as measured by mSHS (less than one unit per year), as it has seen in previous studies [19].
In order to discern which remission criterion is closer to arrest structural damage in our patients with established RA, we took different thresholds for radiographic progression. We observed that PD remission was related to better radiographic outcomes than exclusively clinical remission de nitions, something already known for early rheumatoid arthritis [28]. Ideally, achieving US remission and the complete abrogation of subclinical activity should be the ultimate goal for our RA patients. However, recent data from TASER [17], ARTIC [29] and IMAGINE-RA [30] trials showed that speci cally targeting subclinical in ammation did not yield better clinical outcomes. This target could imply a risk of overtreating, with a higher use of biological therapy and more adverse events. No clear explanation exists for this. Maybe, Treat to Target strategies are not effective for treating subclinical in ammation or, alternatively, persistent subclinical synovitis does not imply worse clinical outcomes in the long-term. We previously have reported that, despite half of patients in remission still have US subclinical activity, the rate of radiographic progression was relatively low (10%) and, most probably, clinically irrelevant [31]. Our results suggest that targeting remission whatever the concept selected, should be appropriate to achieve good clinical and radiological outcomes in a clinical care setting for patients with established RA.
Attending to our results, DAS28-ESR is not inferior to more stringent remission concepts in a clinical care setting after a long follow-up. SDAI, CDAI and ACR/EULAR Boolean criteria are de nitely more appropriate for clinical trials, where comorbidities are speci cally avoided, and patients' assessments and acute phase reactants are related exclusively to RA and not to other extra-conditions seen in a real world-setting [20]. For routine clinical practice, DAS28-ESR could be acceptable, achievable in a high percentage of patients and more suitable for "real" patients with comorbidities. Similar levels of serum calprotectin found across the ve sets of remission, a biomarker strongly related to subclinical in ammation and radiological progression [13,32], reinforces the idea of comparable outcomes for all remission de nitions in the long-term. US remission yielded signi cantly better radiographic outcomes than clinical criteria.
Due the uncertain value of structural damage associated with subclinical synovitis, this should not be an important limitation for the use of non-imaging criteria in clinical practice, especially in established RA.
If aiming for a state closer to normality (true remission), a concept of multi-dimensional remission (MDR) should be used, involving the achievement of different depths of remission, using clinical, imaging and additional serological, immunological or histological parameters [33]. MDR has been shown to be related to better patient-reported outcomes, although prospective data is necessary to de nitely address this issue.
This study has some limitations. The sample size was not big, and patient's data was retrospectively collected from medical records, although the long-term follow-up and the objective outcomes selected, such as change of therapy and radiographic progression partially outweigh these drawbacks. Second, patients' global assessment but neither pain assessment nor HAQ were recorded along the follow-up, since these two patient-related outcomes are not included in our routine monitoring of RA in clinical practice. In addition, since the current study was conducted in a clinical setting, we did not use a preestablished treatment protocol. Finally, few data on sustained remission was collected due to the reduced number of patients. This concept of persistent remission could be of relevance, since the continual ful llment of any remission criteria has been strongly effective in preventing progression of functional disability [34].

Conclusions
DAS28-ESR remission criterion, in routine clinical setting, provided similar outcomes in the long-term than most recent and stringent clinical criteria for patients with established RA. Therefore, taking DAS28-ESR < JR and JDC had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of data analysis. JR, JDC and RS were responsible for the study design. JR, JIM, VRE, RCM, AP, RC, AC, JGP performed data acquisition, analysis, interpretation, and nal approval of the manuscript. Manuscript preparation was by JR, JDC and RS. All authors read and approved the nal manuscript.