Sustained DMARD-free remission in rheumatoid arthritis − about concepts and moving towards practice

Sustained DMARD-free remission (SDFR) is the best possible outcome in RA. It is characterized by sustained absence of clinical arthritis, which is accompanied by resolution of symptoms and restoration of normal physical functioning. Therefore it’s the best proxy for cure in RA. The mechanisms underlying SDFR-development are yet unidentified. Hypothetically, there are two possible scenarios. The first hypothesis is based on the concept of regaining immune-tolerance, which implies that RA-patients are similar at diagnosis and that disease-processes during the disease-course shift into a favorable direction, resulting in regaining a state in which arthritis is persistently absent. This could imply that SDFR is theoretically achievable for all RA-patients. The alternative hypothesis is that RA-patients who achieve SDFR are intrinsically different from those who cannot. This would imply that DMARD-cessation could be restricted to a subgroup of RA-patients. Since the 1990s, DMARD-discontinuation and SDFR have been increasingly studied as long-term-outcome in RA. In this review, we describe hitherto results of clinical, genetic, serological, histological and imaging studies and looked for arguments for the first or second hypothesis in both auto-antibody-positive and auto-antibody-negative RA. In auto-antibody-negative RA, SDFR is pre-sumably restricted to a subgroup of patients with high serological-markers of inflammation at diagnosis and a rapid and sustained decrease in inflammation after treatment-start. Identifying these RA-patients could be helpful in realizing personalized-medicine. In auto-antibody-positive RA, only few patients achieve SDFR and no definite conclusions can be drawn, but data could suggest that SDFR-patients might be a subgroup with relatively low inflammation from disease-presentation onwards.


Introduction
Sustained DMARD-free remission (SDFR), the sustained absence of clinical arthritis after discontinuation of all DMARD-therapy, is increasingly achievable in rheumatoid arthritis (RA) [1,2].As SDFR is accompanied by resolution of symptoms like pain and fatigue, and by restoration of normal physical functioning, it can be regarded the best proxy for cure in RA [2].Nevertheless, mechanisms underlying SDFR-development are poorly understood and the concept of 'cure' contrasts the general perception that RA is a chronic disease requiring life-long disease-modifying treatment.In this review, we propose two concepts that may explain diseaseresolution in RA.We evaluated hitherto scientific evidence of clinical, serological, histological, imaging and genetic studies on SDFR in RA and studied these to substantiate the two hypotheses, with the ultimate aim to increase understanding of permanent diseaseresolution in autoantibody-positive and autoantibody-negative RA.Hypotheses which could explain sustained DMARD-free remission in rheumatoid arthritis.RA-patients achieving sustained DMARD-free remission escape the ongoing process of inflammation.For disease-resolution, it could either be proposed that processes shift into a favorable direction in which markers of disease chronicity -yet unidentified -are lost over time, allowing to achieve SDFR (hypothesis 1).Or, SDFR could be determined by specific characteristics of a subgroup of RA-patients; then patient achieving SFDR are intrinsically different (hypothesis 2).LDA: low disease activity; HDA: high disease activity; DMARD: disease-modifying anti-rheumatic drugs; RA: rheumatoid arthritis.
the first hypothesis the likelihood of achieving SDFR is more or less equal for RA-patients at diagnosis and differentiation occurs over time, whereas hypothesis 2 assumes that the differences in the likelihood of SDFR are already determined at disease-presentation or very early in the disease.

SDFR definition is essential to ensure sustainability
Although DMARD-discontinuation in RA has been studied as early as the 1990s, definitions for DMARD-free remission (DFR) vary widely between studies and several different remissionmeasures have been used to define DMARD-free remission; e.g.DAS, CDAI, SDAI or Boolean remission [3].Some of these measures have been criticized, as they allow a certain level of residual inflammation which is not in line with the concept that DFR is a proxy for cure [4].A strict definition, allowing no clinical signs of disease activity, is therefore preferred when defining DFR.Moreover, there is a distinct difference between "DFR" and "sustained DFR" (SDFR) [3].DFR refers to the state in which RA-patients are able to discontinue their DMARD-treatment but does not impose a duration in which remission is maintained after DMARD-stop (Fig. 2).A recent systematic literature review demonstrated that most flares occur in the first-year after DMARD-cessation; the flare rate after successful DMARD-tapering was on average 45-50% in the first year, whilst this significantly decreased to 5-18% after the first year [3].Hence, achieving DFR is a status that does not provide information on sustainability (Fig. 2).In order to approximate cure, the sustainability of remission needs to be ensured.There is no consensus on which duration of remission is sufficient to define this sustainability.However, since it has been shown that flares are less frequent once remission has been maintained for minimally 1-year after DMARD-cessation [3], a minimal time-frame of at least 1-year in the definition of SDFR seems warranted.Several (observational) studies described that SDFR was achieved approximately 3-4 years after disease-presentation, which means that DMARDs were discontinued 2-3 years after DMARD-initiation (Fig. 2).
In this review, we specifically focus on SDFR, i.e. sustained remission for minimally 1-year after complete DMARD-discontinuation (including corticosteroids).A systematic literature-search was carried out in PubMed to identify studies reporting on SDFR in RA (see Online material supplementary S1 for used terms).References of these studies were used to find additional studies by hand-search.Of the 141 identified studies, 81 reported on DMARDdiscontinuation in RA of which 39 studies reported on sustained DMARD-free remission (Online material supplementary S2).Clinical, serological, imaging, histologic and genetic factors studied in these articles in relation to SDFR are discussed here and reviewed in the light of the two hypotheses.Identified factors were stratified for the moment they were measured: at baseline (Online material supplementary S3), during the disease course (Online material supplementary S4) or at the moment of DMARD-discontinuation (Online material supplementary S5).

Autoantibody positive and negative RA: two disease subsets
Growing evidence supports the idea that RA can be classified into two distinct disease-entities; auto-antibody-positive and Fig. 2. Disease course of rheumatoid arthritis and subsequent development of (sustained) DMARD-free remission.RA-patients could either have persistent disease (upper timeline) or achieve remission and are subsequently able to discontinue their DMARD-therapy (lower two timelines).The middle timeline illustrates the situation in which RA-patients were able to discontinue their DMARD-treatment but were not able to maintain remission after DMARD-stop.In the lower timeline, patients were not only able to discontinue their DMARD-treatment but were also able to sustain remission for minimally one year after DMARD-stop (and the subsequent follow-up thereafter).

Genetic factors associated with SDFR
Genetic-variants have been linked to RA-susceptibility and disease-severity and associations have been observed with radiological-damage and mortality in RA [16][17][18][19][20][21].If specific genetic-variants can determine a trait which makes SDFRdevelopment more accessible in RA-patients with these variants, this would naturally fit the subgroup-hypothesis as geneticvariation is determined long before disease-onset.However, it can also be proposed that specific genetic-domains contribute to more general processes leading towards restoration of immunetolerance once markers of chronicity are turned off over time.Of the 5 genetic studies published with SDFR as outcome (Online material supplementary S3) [22][23][24][25][26], 3 studies focused on shared epitope (SE) which presence was associated with a decreased risk of SDFR [22][23][24][25][26].However, this association disappeared after correction for ACPA-positivity, suggesting that SE and ACPA are part of the same pathway towards SDFR.One study also evaluated other genes in relation to SDFR; nine radiographic-progression-associated genetic variants (SE in HLA-DRB1, rs1896368/rs1896367/rs1528873 in Dickkopf-1, rs2104286 in IL2RA, rs26232 in C5Orf30, rs11908352 in MMP-9, rs451066 at chromosome 14, and rs1485305 in osteoprotegerin), and was auto-antibody-stratified [25].
In auto-antibody-positive RA, none of the 9 genes were associated with SDFR-development, although the number of patients achieving SDFR was low which limited statistical-power (Fig. 3) [25].Interestingly, in auto-antibody-positive RA, shared epitope was not independently associated with SDFR which suggests that SE might have no other role in disease-persistency than via ACPA.
In auto-antibody-negative RA, specific genetic-variance of IL2RA (rs2104286) was associated with increased SDFR-prevalence [25].IL2RA encodes the ␣-chain of the high-affinity IL-2 receptor, which is involved in homeostasis and function of regulatory T-cells which in turn are essential for self-tolerance [27].This specific IL2RAvariant also associates with lower IL-2 serum-levels, and lower IL-2 serum-levels during the disease-course were also associated with more SDFR.Genetic variants in IL2RA were still significantly Crosses indicate that findings are suggestive that the hypothesis is not plausible.A question mark indicates that there is no data on this topic which can be related to one of the two hypotheses.Fig. 4. Scientific substantiation for both hypotheses to understand disease-resolution in auto-antibody-negative-RA. Scientific evidence on biomarkers related to SDFR in auto-antibody-negative RA, summarized from the perspective of both hypotheses.A tick symbol indicates that findings on this topic are suggestive for the specific hypothesis.Crosses indicate that findings are suggestive that the hypothesis is not plausible.A question mark indicates that there is no data on this topic which can be related to one of the two hypotheses.associated with SDFR after correction of serum-Il-2 levels, suggesting that part of its effect exerts via a different pathway than via serum IL-2-levels.Since RA-patients who did not have this IL2RAvariant achieved SDFR less often, the subgroup-hypothesis may be felt most plausible here (Fig. 4).Since most genes that conferred risk for radiographic-damage genes were not associated with SDFR, mechanisms driving joint-damage progression and diseasepersistence might be mostly different.

Clinical characteristics associated with SDFR
When specific clinical characteristics can be linked to SDFRdevelopment this will not only contribute to further understanding of disease-resolution in RA, but it makes personalization of DMARD-tapering more accessible for clinical practice.As several patient characteristics (age, gender, BMI) were not associated with SDFR, studies focused on more disease-specific clinical characteristics like disease-activity, symptom-duration and physical functioning (Online material supplementary S3).Intuitively, SDFR might appear more likely in RA-patients with less clinical inflammation (less involved joints, lower acute-phase reactants) at disease-presentation and in those who attain clinical remission during DMARD-treatment, as switches of chronicity might be more easily turned off and immune-tolerance regained.However, we know that not all RA-patients in clinical remission can successfully taper their DMARD-treatment and other patient-characteristics seem required to attain SDFR (subgrouphypothesis).In total, 19 articles reported on the relation between various clinical-markers and SDFR in the total RA-population (Online material supplementary S2).Although a wide range of clinical-characteristics at disease-presentation was studied, results appeared conflicting (Online material supplementary S3).With a lack of association between baseline clinical-characteristics and SDFR, also clinical markers during the disease course (Online material supplementary S4) and at time of DMARD-discontinuation have been studied (Online material supplementary S5) [28,29].Only four studies were stratified for the presence of auto-antibodies (1 study under-review) [28,[30][31][32].
In auto-antibody-positive RA, a recent study in two separate auto-antibody-positive RA-populations reported that patients achieving SDFR had lower disease-activity-scores (DAS) and less swollen-joints at disease-presentation (data under review).In one of these cohorts, not only baseline-SJC was lower, but also the other individual DAS-components (CRP/TJC/global health).A study focusing on DAS-trajectories after DMARD-initiation, demonstrated that subsequent DAS-trajectories during follow-up were not different in patients achieving SDFR, but statistical power was limited since only few auto-antibody-positive RA-patients achieved SDFR [28].The finding that auto-antibody-positive RA-patients achieving SDFR have already less inflammation from disease-presentation onwards might intuitively suggest that the inflammatory response in these patients is less developed and therefore more easily suppressed by DMARD-treatment after which immune-tolerance may be easier restored.If so, a shorter symptom-duration in SDFRpatients would be expected.Although one study reported that a shorter symptom-duration was associated with SDFR [31], this was not replicated [33].Thus, that auto-antibody-positive RApatients achieving SDFR have a lower disease activity already from diagnoses onwards seems to be better explained by the subgrouphypothesis (Fig. 3).
In auto-antibody-negative RA, no differences were found in baseline DAS, nor in the individual DAS-components (SJC/TJC/ESR/GPA) at baseline [28].However, auto-antibodynegative RA-patients achieving SDFR did have a significantly stronger decline in DAS in the first 4-months of DMARD-treatment [28].In patients with early DAS-remission (DAS < 1.6 after 4months) the likelihood of achieving SDFR increased up to 75%, compared to 40% in other auto-antibody-negative RA-patients.Of the DAS-components, predominantly SJC and ESR declined stronger in SDFR-patients.Moreover, higher baseline CRP-levels are associated with SDFR in auto-antibody-negative-RA [30,32], which might appear counterintuitive.However, a follow-up study demonstrated that auto-antibody-negative RA-patients with high baseline CRP-levels also have a subsequent stronger decline in CRP-levels during the first year of DMARD-treatment, stronger than patients who do not achieve SDFR [30].Thus, patients achieving SDFR are characterized by high baseline CRP-levels and a subsequent strong decline in DAS and CRP upon DMARD-initiation.The combination of both high baseline CRP-levels and early DASremission even resulted in 85% SDFR, suggesting that patients who share these favorable characteristics define a subgroup of patients most likely to attain disease-resolution (Fig. 4).
In auto-antibody-positive RA-patients in both studies, neither baseline levels of these biomarkers, nor the change in levels of these markers in the first 2-years of DMARD-treatment, were associated with later on development of SDFR (Fig. 3) [30,32].Yet, the number of auto-antibody-positive-RA-patients achieving SDFR was low, limiting statistical-power.
In auto-antibody-negative RA, one study demonstrated that auto-antibody-negative-RA-patients achieving SDFR had not only higher baseline CRP-levels, but also baseline levels of MMP-3 and SAA were higher [32].Although also this initially appeared counterintuitive from a disease-resolution perspective, also here a follow-up study demonstrated that the levels of these serological markers (and MMP-1) declined significantly stronger in the first year of DMARD-treatment in auto-antibody-negative RA-patients achieving SDFR [30].Interestingly, this decline in serological levels in the first year was strongest in SDFR-patients achieving early DAS-remission, suggesting that the combination of these favorable patient-characteristics define a subgroup most likely to achieve SDFR (Fig. 4).

Imaging findings associated with SDFR
Advanced imaging-modalities have contributed significantly to the understanding of tissues inflamed in RA [34][35][36].Whereas conventional radiographs can only visualize erosive changes, ultrasound and MRI are sensitive measures of local joint-inflammation in synovium, tenosynovium and (for MRI) bone [37].This allows to study joint-inflammation in more detail in relation to SDFRdevelopment.
No ultrasound studies have been conducted in relation to SDFR in RA.Of the 6 imaging-studies [22,23,33,38], two studies that used MRI of hand and foot-joints were auto-antibody-stratified (1 under review) [33].
In auto-antibody-positive RA, less MRI-detected jointinflammation (especially synovitis and osteitis) and erosions at diagnosis, and during subsequent follow-up, were associated with SDFR (Online material supplementary S3 and S4) [39].This was validated in an external auto-antibody-positive RA-population [39].Subsequently, another study linked less MRI-detected synovitis and erosions at time of DMARD-discontinuation to more SDFR (Online material supplementary S5) [33].Thus, also here auto-antibody-positive RA-patients achieving SDFR had significantly less joint-inflammation from disease-presentation onwards.Importantly, the symptom duration was not shorter in patients that later achieved SFDR, suggesting that the lower levels of inflammation may be a patient-characteristic (subgroup-hypothesis, Fig. 3).Nevertheless, definite conclusions should not be drawn yet because of low number of auto-antibody-positive RA-patients achieved SDFR in these analyses.
In auto-antibody-negative RA, MRI-detected jointinflammation did not differ at baseline between patients who did and did not achieve SDFR.However, the decline in MRIdetected inflammation (especially tenosynovitis and osteitis) in the first-year of DMARD-treatment was significantly stronger in patients achieving SDFR [39].Synovitis, tenosynovitis and osteitis declined together, in contrast to auto-antibody-positive RA in which tenosynovitis did not play a role in SDFR.Thus, autoantibody-negative RA-patients achieving SDFR are comparable to other RA-patients at baseline but demonstrate a stronger decline in inflammation (measured with the DAS, serological marker and imaging of the joints) after treatment-start.This suggests that underlying disease-processes might be inherently different and respond differently to DMARD-therapy which in turn appears favorable for attaining disease-resolution (subgroup-hypothesis, Fig. 4).

Histological characteristics associated with SDFR
Histological studies can reveal fundamental processes underling RA-chronicity and are therefore of great interest when aiming to understand disease-resolution.Despite extensive histological studies on the pro-inflammatory processes in RA, histological studies on disease-remission are scarce [40].Yet, it would be of interest whether local (inflammatory) processes are inherently different from diagnosis onwards in patients achieving SDFR (subgrouphypothesis) or whether normal joint-homeostasis can theoretically be restored in all RA-patients when disease-processes shift into a favorable direction (regaining immune-tolerance-hypothesis).Histological studies on SDFR in RA are absent.However, an interesting histologic study was published recently about RA-patients who were in clinical remission after DMARD-tapering [40].Using singlecell-sequencing of synovial tissue, synovial tissue macrophages (STMs) were identified which appeared to play a role in regulating joint inflammation.MerTK(pos) STMs were high in RA-patients in remission and low in active disease; and a low proportion of MerTK(pos) STMs in RA-patients in remission was associated with an increased risk of flare after DMARD-tapering.However, since histologic or molecular-studies during the development of SDFR are absent, it remains to be determined whether such innate cells in the joint tissue are involved in SDFR-development.

Auto-antibody response characteristics
Several different auto-antibodies, e.g.RF, ACPA, anticarbamylated protein (anti-CarP), anit-acetylated protein antibodies (AAPA), have been studied in relation to SDFR [41][42][43]; but only the presence of ACPA appeared to be independently associated with SDFR, irrespective of the co-existence of other auto-antibodies (Online material supplementary S3) [41,43].Nonetheless, it remains unclear whether ACPA plays an active role in persistent disease, or whether ACPA is solely a bystander [44].ACPA can be present for years before the disease-onset and maturation of the ACPAresponse appears to occur predominantly in the asymptomatic pre-arthritis phase of RA, and according to current knowledge is not the final step towards developing RA [45,46].In addition, in the pre-RA-phase of clinically-suspect-arthralgia, a matured autoantibody-response also occurs in ACPA-positive patients who do not transition to RA but achieve resolution [45].The question is to what extent ACPA-characteristics are related to achieving resolution of RA.It has been demonstrated that ACPA-levels are not associated with SDFR [47,48].Moreover, it has been hypothesized that the disappearance of auto-antibodies, defined as ïmmunological remission ¨, would occur before or at the time that disease-resolution is achieved [49].However, interestingly enough, it has been demonstrated that SDFR is not paralleled by the disappearance of ACPA or by a reduction in ACPA-levels [47].Also ACPA-IgM remained similarly present in RA-patients achieving SDFR, showing a disconnection between the ACPA-response and a beneficial clinical disease-course [38].Further studies are needed to learn how "immunological-remission" should be defined in order to explain disease-resolution in RA on immunological-level.Interestingly, whereas extensive glycosylation of the variable (Fab) domain is a remarkable molecular-feature of ACPA-IgG, a recent study demonstrated that levels of ACPA Fab-glycosylation are significantly lower at disease-onset in ACPA-positive RA-patients achieving SDFR, compared to age-and gender-matched RA-patients not achieving SDFR [46].Since these differences in auto-antibodycharacteristics were already present at disease-presentation, it appears suggestive for a subgroup of auto-antibody-positive RApatients in whom RA more easily resolves (Fig. 3), rather than a change in auto-antibody-characteristics over time that relates to regaining immune-tolerance.However, numbers of auto-antibodypositive patients achieving SDFR were low and further studies are warranted.

Treatment & SDFR: can we induce SDFR with disease-modifying treatment?
The ultimate question would be whether we can induce diseaseresolution in RA, thus are we able to actively promote cure of RA?If disease-resolution can be explained with the subgrouphypothesis, SDFR is reserved for only a part of the RA-population and cannot be induced in the rest.Identification of this subgroup at disease-presentation or early in the disease can lead to personalized treatment-decisions to taper DMARDs once remission is achieved.However, from the perspective of the regaining-immune-tolerance hypothesis, if SDFR would be inducible, this could theoretically be possible in all RA-patients.Thus, understanding SDFR from the perspective of these hypotheses is essential for future management of RA.Several studies have explored whether different treatmentstrategies led to higher prevalence of SDFR in RA (Online material supplementary S6).Five of these studies were stratified for autoantibodies [1,2,23,33,50].
In auto-antibody-positive RA, SDFR-prevalence was higher among patients undergoing DAS-driven-therapy compared to those receiving non-DAS-driven therapy [23,50].In turn, treatment-steering at DAS < 1.6 led to more SDFR than treatmentsteering at DAS < 2.4 [50].In line with this, a more recent study demonstrated that early and intensive DMARD-treatment led to a higher incidence of SDFR in auto-antibody-positive RA [14].This suggests that SDFR might be inducible in this patient-population, which fits the regaining immune-tolerance hypothesis (Fig. 3).Nevertheless, SDFR is still infrequent in auto-antibody-positive RA and future studies have to point out whether these treatmenteffects are present in all auto-antibody-positive patients or confined to a subgroup of patients (for instance those presenting with less severe inflammation or less F(ab)-glycosylation of ACPA).
In auto-antibody-negative RA, the transition to early and intensive treatment-strategies over time did not increase SDFRprevalence [14,50], which suggest that achieving SDFR is determined by patient-characteristics rather than treatmentstrategies which support the subgroup-hypothesis (Fig. 4).

Discussion
In auto-antibody-positive RA, SDFR is infrequent which limited statistical-power to characterize these patients.Nevertheless, the available data could suggest that auto-antibody-positive RApatients achieving SDFR represent of subgroup of patients with specific characteristics at baseline (Fig. 3); less clinically apparent and MRI-detected joint-inflammation and different auto-antibodycharacteristics (less ACPA Fab-glycosylation).However, the fact that more intensive DMARD-strategies have increased SDFRprevalence in auto-antibody-positive-RA could also support the regaining-immune-tolerance hypothesis.Future studies on SDFR in auto-antibody-positive-RA, ideally with a larger number of patients achieving SDFR, remain warranted to draw more definite conclusions.
In auto-antibody-negative RA, patients achieving SDFR are characterized by a combined strong decline in both clinical inflammation, serological inflammatory-markers and MRI-detected joint-inflammation in the first year of DMARD-treatment.These findings suggest that auto-antibody-negative RA-patients achieving SDFR represent a subgroup with favorable characteristics leading to a relatively early and adequate response to DMARDtreatment and subsequent disease-resolution later on (Fig. 4).Since

Fig. 1 .
Fig. 1.Hypotheses which could explain sustained DMARD-free remission in rheumatoid arthritis.RA-patients achieving sustained DMARD-free remission escape the ongoing process of inflammation.For disease-resolution, it could either be proposed that processes shift into a favorable direction in which markers of disease chronicity -yet unidentified -are lost over time, allowing to achieve SDFR (hypothesis 1).Or, SDFR could be determined by specific characteristics of a subgroup of RA-patients; then patient achieving SFDR are intrinsically different (hypothesis 2).LDA: low disease activity; HDA: high disease activity; DMARD: disease-modifying anti-rheumatic drugs; RA: rheumatoid arthritis.

Fig. 3 .
Fig. 3. Scientific substantiation for both hypotheses to understand disease-resolution in auto-antibody-positive-RA. Scientific evidence on biomarkers related to SDFR in auto-antibody-positive RA, summarized from the perspective of both hypotheses.A tick symbol indicates that findings on this topic are suggestive for the specific hypothesis.Crosses indicate that findings are suggestive that the hypothesis is not plausible.A question mark indicates that there is no data on this topic which can be related to one of the two hypotheses.