Background
It is now widely accepted that an ultrasound examination is superior to a clinical examination for the detection of joint inflammation [
1‐
3]. The European League Against Rheumatism (EULAR) task force has published its recommendations for the use of imaging of the joints in the clinical management of rheumatoid arthritis (RA), and the recommendations state that ultrasound is very helpful for identifying synovitis and bone erosions and thus for making accurate diagnoses, predicting outcomes and responses to treatment, and monitoring disease progression [
4]. In addition, ultrasound can be used to assess the persistent inflammation that predicts subsequent joint damage in patients achieving clinical remission [
4].
Highly sensitive modalities such as ultrasound and magnetic resonance imaging (MRI) have revealed the presence of residual or subclinical synovitis in patients with RA in clinical remission [
3,
5,
6]. A meta-analysis of reports of residual synovitis detected by ultrasound revealed that the prevalence of synovial hypertrophy with a power Doppler (PD) signal was 32% to 44% [
6]. In addition, PD signal positivity of residual synovitis strongly predicted the risk of structural progression and relapse in clinical remission [
3,
6].
The use of aggressive treatment, including biologic disease-modifying antirheumatic drugs (bDMARDs), has greatly improved outcomes for patients with RA. Although some patients showing persistent clinical remission may achieve bDMARD- and drug-free remission, a considerable proportion of the patients cannot maintain the clinical benefit [
7‐
11]. There are no definitive predictive markers that enable the identification of such patients, and additional challenges have been to determine (1) how remission is maintained over time, (2) how subclinical disease can be detected and evaluated, and (3) how undetectable active or progressive disease can be distinguished from true remission of inflammation [
12].
Another issue is the definition of imaging remission. Apart from mere clinical remission in terms of joint manifestations, imaging remission may have to be considered in RA cases in which synovial inflammation has also completely ceased [
4,
12]. The lower relapse rates observed in patients with normal ultrasound results support the biological relevance of this concept [
12]. Recent studies indicated that the presence of synovitis with a PD signal before the discontinuation or dose reduction of a bDMARD is an independent predictor for a subsequent relapse in patients in clinical remission [
13,
14]. Therefore, if a discontinuation or dose reduction of a patient’s treatment with a bDMARD is being considered, it may be preferable to confirm that PD activity is absent or low by an ultrasound examination. However, relapses occur even in patients with RA without PD activity. These data suggest that the identification of ultrasound indices other than synovitis, especially PD synovitis, is necessary to improve the accuracy of ultrasound imaging remission. The objectives of this study were to explore potential risk factors of relapse, especially ultrasound indices, after the discontinuation of bDMARD therapy in patients with RA whose ultrasound PD synovitis activity and clinical disease activity were well controlled.
Discussion
Our study results show that the presence of bone erosion detected by ultrasound at bDMARD discontinuation was the only independent risk factor of relapse during the 12 months after discontinuation of bDMARD therapy. No other factors, including clinical variables at initiation, clinical and ultrasound variables at discontinuation, and biomarkers, were associated with relapse.
Several randomized control trials of bDMARD strategies addressing the discontinuation or dose reduction of bDMARDs have been reported [
7,
8]. The PRESERVE trial, as a representative study, was undertaken to determine whether LDA could be sustained with discontinuation or dose reduction of etanercept (ETN) in patients with established RA [
9]. After treatment with 50 mg of ETN plus methotrexate (MTX) for 36 weeks, patients were randomized to three groups: 50 mg of ETN plus MTX (maintenance), 25 mg of ETN plus MTX (reduction), and placebo plus MTX (discontinuation). At 12 months after randomization, LDA had been maintained in only 42.6% of the discontinuation group but in 82.6% of the maintenance group and 79.1% of the reduction group [
9]. Similarly, on the basis of results of registries in the United States [
10] and Japan [
11], clinical benefit could not be maintained in a considerable portion of the patients after discontinuation of bDMARD therapy.
Regarding the significance of subclinical synovitis in patients with RA [
3,
5,
6], we reported that moderate or severe PD activity was present in approximately 30% of patients who were free of physical synovitis [
5]. In these physical synovitis-free patients, the use of bDMARDs was low and the presence of ultrasound-detected bone erosion was high in the group with moderate or severe PD activity [
5]. A strategy combining clinical and ultrasound assessments may better select individuals for sustained discontinuation or dose reduction of DMARD therapy. In this regard, three prospective studies that included ultrasound indices reported predictive factors of relapse after discontinuation or dose reduction of bDMARD therapy in sustained clinical remission [
13,
14,
24]. Among these factors, higher clinical disease activity and a higher PD synovitis score are thought to be important indicators of the failure of bDMARD discontinuation or dose reduction.
The difference between these studies and our present data is that neither clinical disease activity nor ultrasound synovitis activity was associated with discontinuation of bDMARDs. Compared with these studies, the clinical disease activity and the residual ultrasound synovitis activity at entry in the present study were quite low, even in the patients who experienced subsequent relapse [
13,
14,
24]. Therefore, these factors would not predict relapse in the present patients. However, although no statistical significance was found between the relapse patients and nonrelapse patients, the frequency of PD grade 1 synovitis was numerically high in the relapse patients compared with nonrelapse patients. Thus, no PD signal, instead of PD grade at each joint being ≤1, might be crucial to maintaining clinical remission after discontinuation of bDMARD therapy. As stated in the discussion of limitations below, a study including a larger number of patients is necessary to confirm this speculation.
The novel aspect of the present study was the evaluation of bone erosion by ultrasound in patients with RA whose ultrasound PD synovitis activity was well controlled. The presence of ultrasound bone erosion at discontinuation was the only independent predictor of relapse during the 12 months after bDMARD therapy. We speculate that ultrasound-detected bone erosion is important for the prediction of relapse for the reasons outlined below.
First, we and other investigators have found a close association between the presence of bone erosion and residual PD synovitis at the joint level in patients with RA in clinical remission or with LDA [
5,
25]. Thus, bone erosions are the natural consequence of persistent joint inflammation. In this regard, MRI-detected bone edema (osteitis), which is bone inflammation that is strongly predictive of further radiographic progression [
26,
27], coexisted with PD synovitis in patients with RA [
28]. We suspect that ultrasound can detect subclinical findings of joint inflammation in the areas most aggressively hit by the disease and/or where the inflammation has started earlier where bone erosion is detected. Alternatively, ultrasound-detected bone erosion could be derived from the integration of joint inflammation, whereas PD synovitis might always not be detected at bDMARD discontinuation.
Second, the presence of bone degradation elements as a result of joint damage can trigger immunological pathways, developing further joint inflammation, because it is well known that bone erosion detected by ultrasound or plain radiography can be used for the prediction of further joint destruction [
25,
29]. These two hypotheses do not conflict, and we thus speculate that there is a vicious cycle in which the joint inflammation leads to joint damage, with the consequent release of bone and cartilage fragments that sustain the joint inflammation, resulting in further relapse.
With respect to serum biomarkers, the best-studied predictor of relapse to date is ACPA positivity [
12]. In the REduction of Therapy in RA patients in Ongoing remission (RETRO) study, ACPA status clearly indicated a higher relapse risk with lower chances of maintaining remission when ACPAs are present [
12]. Data derived from other investigations, such as the Dutch Behandel Strategieen (BeSt) study and the High Induction Therapy with Anti-Rheumatic Drugs (HIT HARD) study, support this concept [
12]. In contrast, we did not find an association of positivity for ACPA or RF at the initiation or the discontinuation of bDMARDs with subsequent relapse. In the Remission Induction by Remicade in RA (RRR) study, the continuous presence of RF was reported to lower the likelihood of a successful withdrawal of TNF inhibitors [
12]. The serial changes of positivity for ACPA or RF after the discontinuation of bDMARDs should thus be studied further. In addition, the titer of autoantibodies might be important to predict RA relapses.
Some limitations of this study should be acknowledged. The RA population was relatively small and heterogeneous regarding the patients’ RA characteristics, bDMARDs, and dosages. The limited sample size of our study does not allow for subanalyses of differences between bDMARDs with different modes of action. A study with a larger number of patients and a longer follow-up period is needed to establish optimal ultrasound-based strategies to reduce the unnecessarily long use of bDMARDs.
Acknowledgements
Not applicable.