Association Between Serum Bone Biomarker Levels and Therapeutic Response to Abatacept in Patients With Rheumatoid Arthritis (Ra): a Multicenter Ra Ultrasound Prospective Cohort Study in Japan

Background: To evaluate the effect of treatment on serum bone biomarkers and explore whether serum bone biomarkers are associated with therapeutic response in rheumatoid arthritis (RA) patients treated with abatacept. Methods: We enrolled 59 RA patients treated with abatacept from a multicenter prospective ultrasound cohort study of patients who received biologic or targeted synthetic disease-modifying antirheumatic drug (DMARD) therapy. We evaluated the patients' clinical disease activity and musculoskeletal ultrasound (MSUS) scores. The serum concentrations of ve bone biomarkers were evaluated (dickkopf-1 [Dkk-1], sclerostin [SOST], osteocalcin [OC], osteopontin [OPN], and osteoprotegerin [OPG]) by multiplex bead assays at baseline, 3, and 6 months: the change over 6 months was dened as the Δ value. 'Power Doppler (PD) responder' was dened as a patient whose Δtotal PD score over 6 months was greater than the median change. Results: Abatacept signicantly improved clinical disease activity as well as the MSUS score over 6 months. Serum OPG was signicantly elevated at 6 months after the introduction of abatacept (p=0.016). The ΔSOST and ΔOPG values were negatively correlated with the Δtotal PD score (rs= −0.31, p<0.05 and rs= −0.34, p<0.01, respectively). The serum Dkk-1 at baseline was signicantly lower in the PD responders (n=30) compared to the non-PD responders (n=29) (p=0.026). A multivariate logistic regression analysis showed that the serum Dkk-1 at baseline (odds ratio 0.50, 95% condence interval [CI] 0.23–0.91, p=0.043) was an independent predictor of PD responder status. Conclusion: Serum levels of bone biomarkers may be useful for predicting RA patients' therapeutic response to abatacept.


Introduction
Rheumatoid arthritis (RA) is characterized by persistent synovitis, systemic in ammation, and autoantibodies [1]. Uncontrolled active RA causes joint damage, disability, decreased quality of life, and comorbidities. The tight control of the disease activity of RA following the treat-to-target (T2T) strategy is thus recommended [2]. Advances in the treatment of RA, such as biological disease-modifying antirheumatic drugs (bDMARDs), have provided better clinical outcomes, including the achievement of clinical remission, the prevention of joint damage, and the preservation of function for individuals with RA.
Bone damage in RA is characterized by articular erosions, periarticular bone loss, and secondary osteoporosis caused by chronic in ammation [12]. In the bone metabolism of RA osteoclastogenesis/bone resorption is promoted and bone formation is inhibited. Among the underlying mechanisms, important roles are played by the interaction of immune responses mediated by proin ammatory cytokines such as tumor necrosis factor (TNF), interleukin (IL)-6 and IL-17, and bone biomarkers such as receptor activator of nuclear factor kappa-B ligand (RANKL), osteoprotegerin (OPG), dickkopf-1 (Dkk-1), and sclerostin (SOST) [12][13][14][15]. Effects of bDMARDs including TNF inhibitors [16,17] and IL-6 inhibitors [18][19][20] on bone biomarkers in patients with RA have been reported, but the effect of abatacept on bone biomarkers in patients with RA has not been investigated. Here, we measured the serum levels of ve bone biomarkers (Dkk-1, SOST, osteocalcin [OC], osteopontin [OPN], and OPG) using a multiplex bead assay (bone panel) in RA patients treated with abatacept.
Dkk1 and SOST inhibit the Wnt signaling pathway regulating bone formation [12,14,15]. OC is an important component of bone matrix protein, synthesized mainly in osteoblasts [12,21], and it is a biomarker of bone formation. OPG is a soluble decoy RANKL receptor that inhibits RANKL function, and it is highly involved in in ammatory bone resorption by interfering with RANK-RANKL binding [12,14,15].
OPN is strongly expressed in bone, where it promotes the adhesion of osteoclasts to the mineralized matrix regulating bone resorption and formation [22]. OPN has also been recognized as an important proin ammatory mediator which participates in pathological processes such as in ammatory reactions, vasospasm formation, and bone damage [23,24].
To achieve the goal of the T2T strategy, the adequate management of disease activity requires a sensitive and accurate assessment of arthritis. Imaging plays an important role in this assessment. Musculoskeletal ultrasound (MSUS) has been widely applied in clinical settings as an imaging modality for patients with rheumatic diseases [25,26]. Compared to clinical and radiographic examinations, MSUS provides a straightforward and more accurate detection of both in ammation and damage at the joint level [25,26]. We have conducted a multicenter prospective observational cohort study of patients with active RA who received bDMARDs or targeted synthetic DMARDs (tsDMARDs) therapy at 27 participating rheumatology centers from the Kyushu region of Japan since June 2013[Kyushu Multicenter Rheumatoid Arthritis Ultrasound Prospective Observational Cohort Study (KUDOS)] [27][28][29]. We evaluated the therapeutic e cacy of the bDMARDs and tsDMARDs by clinical measurements, MSUS, and biomarker assessments. A multicenter collaborative study that prospectively evaluates disease activity by using MSUS standardized at a high level is rare, even worldwide.
In the present study, we evaluated the effect of abatacept treatment on bone biomarkers and explored whether bone biomarkers are associated with the therapeutic response in RA patients treated with abatacept using the KUDOS.

Patients
This study is part of an ongoing non-randomized multicenter prospective cohort study [KUDOS] of patients with active RA who received bDMARD or tsDMARD therapy at 27 participating rheumatology centers from the Kyushu region of Japan since June 2013 [27][28][29]. We evaluated therapeutic e cacy by determining the patients' clinical disease activity, MSUS score, and serum biomarkers at baseline and at 3, 6, 9, 12, 18, and 24 months starting from the initiation of new bDMARD or tsDMARD. For the present study, we enrolled the 59 consecutive Japanese patients with active RA who were treated with abatacept and had continued the treatment for > 6 months at 10 participating rheumatology centers during the period from December 2013 to March 2016. All patients were required to satisfy the 1987 American College of Rheumatology [30] and/or 2010 American College of Rheumatology (ACR)/EULAR criteria for RA [31]. Abatacept was administered as recommended by the manufacturers: 125 mg via subcutaneous injection weekly or 500-750 mg via intravenous infusion every 4 weeks. We excluded patients who have newly introduced an oral bisphosphonate during the study period or treated with intravenous bisphosphonates, anti-RANKL antibodies, or parathyroid hormone (PTH) agents.
The study is registered with the University Hospital Medical Information Network Clinical Trials Registry (http://www.umin.ac.jp/ctr/, UMIN 000012524) and was approved by the Institutional Review Board of Nagasaki University (approval no. 13102866). All patients gave their signed informed consent to participate in accordance with the Helsinki Declaration.

Musculoskeletal ultrasound assessment
The MSUS examination of each patient was performed by JCR-certi ed sonographers. At all of the participating institutions, a trained MSUS expert examined the patient in a situation recommended by the JCR guidelines, paying attention to factors that can affect PD results, including the room temperature, the last use of a nonsteroidal anti-in ammatory drug (NSAID), and hand position. Medium-level to high-level ultrasound machines were used (Toshiba AplioXG and Aplio300, GE Logic series 7 and 8 or Hitachi Ascendus, Avius, Noblus, and Hi Vision Preirus) with high-frequency (12-18.5 MHz) linear transducers. The Doppler parameters were adjusted according to the device used (range of pulse repetition frequency 500-1000 Hz; Doppler frequency 6.1-10.0 MHz). There was no change in MSUS settings during the study.
Twenty-two joints including the metacarpophalangeal (MCP), proximal interphalangeal (PIP), and wrist joints of the bilateral hands were assessed by MSUS at baseline and at 3 and 6 months of treatment. The 22 joints were scanned on the dorsal aspect. Standardized joint and probe positions were used, based on a guideline published by the JCR. Each grayscale (GS) synovial hypertrophy and PD signal was scored semi-quantitatively on a scale from 0 to 3 [32]. The sum of the GS or PD scores was used as the indicator of US disease activity, described as the total GS score or total PD score. The total scores ranged from 0 to 66. We de ned a PD responder at 6 months as a patient whose change in total PD score (Δtotal PD score) over 6 months was greater than the median change (i.e., a Δtotal PD score less than or equal to − 4) in all patients. We de ned PD remission as a total PD score of 0 at 6 months. Interobserver reliability was con rmed in a previous investigation [27].

Bone biomarker measurements
We measured the concentrations of the following biomarkers by using serum stored on the same day as the patient's clinical evaluation. Rheumatoid factor (RF) was measured by the latex agglutination turbidimetric immunoassay (LATIA) (LZ test 'Eiken' RF). Anti-cyclic citrullinated peptide antibody (ACPA) was measured by a chemiluminescent immunoassay (CLEIA) (STACIA MEBLux test CCP). We performed the multiplex bead assays using diluted serum supernatants and a Milliplex MAP Human Bone Panel analyzed with a Bio-Plex® MAGPIX™ Multiplex Reader (Bio-Rad, Hercules, CA) according to the manufacturer's instructions. The bone biomarkers that could be measured by the bead panel included Dkk-1, SOST, OC, OPN, and OPG using a multi-suspension array. These bone biomarkers were also measured in 18 healthy age-and gender-matched volunteers (median age 70 years, 67% females). The volunteers were recruited at medical check-ups in the town of Saza, Japan. The protocol was approved by the Nagasaki University Ethics Committee for Humans Subjects (approval no. 14051404).

Statistical analyses
Statistical analyses were performed using JMP Pro statistical software, ver. 11.0 (SAS, Cary, NC).
Quantitative variables are presented as medians and interquartile ranges (IQRs). Categorical variables are presented as percentages. We used the Mann-Whitney U-test for comparisons between independent medians, and we used the Chi-square test for the evaluation of the associations between categorical variables. Correlations were assessed with Spearman's correlation coe cient. The changes in clinical disease activity indices, MSUS scores, or serum concentrations of bone biomarkers over 6 months were analyzed using the Wilcoxon signed ranks test. We attempted to identify any variables that were independently predictive of the PD responder status at 6 months from the patients' baseline characteristics by performing a multivariate logistic regression analysis. All variables with a p-value < 0.1 in a univariate analysis were used in the multivariate models, but the total GS score was excluded as a confounding factor of the total PD score. P-values < 0.05 were considered signi cant.

The serum concentrations of the bone biomarkers
The results of our comparison of the serum concentrations of the ve bone biomarkers between the healthy volunteers and the RA patients and the changes of the biomarkers over the 6-month period after the introduction of abatacept in the RA patients are shown in Supplementary Table S1. Serum OPN was signi cantly higher (p < 0.0001) and serum OC tended to be lower (p = 0.099) in the RA patients compared to the healthy volunteers. The other serum bone biomarkers were not signi cantly different between the RA and healthy groups. Serum OPG was signi cantly elevated at 6 months after the introduction of abatacept (p = 0.016, Suppl. Fig. S1), but the other serum bone biomarkers did not change after treatment. The oral corticosteroids and bisphosphonates did not affect the serum concentrations of bone biomarkers or their changes.
The correlations between the bone biomarkers and disease activity Table 2 shows the correlations between the serum levels of the bone biomarkers and the disease activity at baseline and the changes (Δ value) over the 6 months of abatacept treatment. At baseline, the serum level of Dkk-1 was positively correlated with the CRP level (rs = 0.276, p < 0.05), and that of OC tended to be negatively correlated with the CRP level (rs = − 0.254, p < 0.1). The serum level of OPG was positively correlated with the SDAI value (rs = 0.264, p < 0.05) and total GS score (rs = 0.276, p < 0.05), and it tended to be positively correlated with the total PD score (rs = 0.232, p < 0.1). The results were obtained using Spearman's correlation coe cient. *p < 0.1, *p < 0.05, ***p < 0.01. OC: osteocalcin, OPG: osteoprotegrin, OPN: osteopontin, SOST: sclerostin.

The association between the ultrasonographic response and the bone biomarkers
Because the ΔSOST and ΔOPG were negatively correlated with the Δtotal PD score as described above, we compared these two parameters between the PD responders and non-PD responders. Both the ΔSOST and ΔOPG were signi cantly greater in the PD responders compared to the non-PD responders (p = 0.0041 and 0.0073, respectively, Fig. 2).
The results of our comparison of the patient characteristics at baseline between the PD responders and non-PD responders at 6 months are provided in Table 3. The patient age and disease duration tended to be higher in the PD responders compared to the non-PD responders. RF positivity (p = 0.030), the SDAI (p = 0.0081), the total GS score (p = 0.0005), and the total PD score (p < 0.0001) were signi cantly higher in the PD responders compared to the non-PD responders. The serum level of Dkk-1 was signi cantly lower in the PD responders compared to the non-PD responders (p = 0.026). The serum level of SOST tended to be signi cantly lower in the PD responders compared to the non-PD responders (p = 0.058).  We compared the changes in the SDAI and the total PD score between the patients with a low Dkk-1 value (i.e., < the median Dkk-1 value at baseline) and the patients with a high Dkk-1 value (≥ the median Dkk-1 at baseline) (Suppl. Fig. S2) and between the patients with a low SOST value (< the median SOST at baseline) and those with a high SOST value (≥ the median of SOST at baseline) (Suppl. Fig. S3). The SDAI score improved in the patients with a low Dkk-1 as well as the patients with a high Dkk-1 and in the patients with a low SOST as well as the patients with a high SOST. However, the improvement in the total PD score was better in the patients with a low Dkk-1 compared to those with a high Dkk-1.

Discussion
We evaluated the association between the serum concentrations of ve bone biomarkers and the therapeutic response to abatacept in RA patients, using our multicenter prospective ultrasound cohort study [KUDOS]. In the present study, since abatacept signi cantly improved the patients' clinical disease activity as well as their MSUS score over the 6-month treatment, bone destruction was expected to be prevented in this population.
First, we observed that both Dkk-1 and SOST (which are inhibitors of the Wnt signaling pathway) were associated with the therapeutic response, especially the ultrasonographic response evaluated by the total PD score. Low serum levels of Dkk-1 at baseline was an independent predictor of the PD responder status at 6 months. Low serum levels of SOST at baseline tended to be associated with the PD responder status at 6 months. Increased serum levels of SOST were signi cantly correlated with the improvement of disease activity after treatment with abatacept. The Wnt signaling pathway plays a key role in several biological processes such as cellular proliferation and tissue regeneration, and its dysregulation is involved in the pathogenesis of many autoimmune diseases [14]. In RA, Wnt signaling is implicated in systemic and localized bone loss. This process involves proin ammatory cytokines produced by the synovial membrane, which may increase bone resorption but also stimulate soluble antagonists of the canonical Wnt/β-catenin signaling pathway (including DKK-1 and sclerostin) and subsequently inhibit osteoblast proliferation, maturation, and progenitor differentiation [14,33,34].
In particular, the pro-in ammatory cytokines TNF and IL-1 induce Dkk-1 and SOST; IL-17 down-regulates the Wnt/β-catenin pathway indirectly, enhancing the production of TNF and IL-1; and IL-6 induces the differentiation of B cells into plasma cells which express Dkk-1 [14,35,36]. Moreover, Dkk-1 induces SOST [14]. It has been demonstrated that Dkk-1 promotes synovial angiogenesis, a critical process in the pathogenesis of RA [34]: vascular proliferation occurs during pannus formation in the affected joints [37,38], during which the synovium becomes locally invasive at the interface with cartilage and bone. The serum level of Dkk-1 was shown to be higher in patients with RA than in controls and to correlate with bone erosions and in ammation [34]. Increased serum levels of Dkk-1 and SOST may therefore indicate a poor prognosis and resistance to treatment in patients with RA. In the present patient series, the serum levels of Dkk-1 and SOST at baseline were closely associated with the responsiveness of PD activity, which re ects synovial angiogenesis [37,39] and predicts joint destruction [40].
Regarding the effect of bDMARD therapy on the Wnt signaling in RA, most of the relevant studies were conducted with TNFα inhibitors [34,41] or an IL-6 receptor (IL-6R) inhibitor, tocilizumab [18,19], and they showed a decrease in the serum level of Dkk-1 in RA patients undergoing these bDMARD treatments. A reduction in the serum level of Dkk-1 may result from the inhibition of the TNFα-and IL-6-dependent induction of Dkk-1 by TNFα inhibitors or anti-IL-6 monoclonal antibody. On the other hand, abatacept did not reduce the serum level of Dkk-1 in the present RA patients, perhaps because it does not directly inhibit TNFα and IL-6. There is no report regarding the effect of abatacept, except the present study, on the Wnt signaling in RA patients [14].
In healthy mice, abatacept promoted bone formation by inducing the Wnt ligand Wnt-10b in a T celldependent manner [42], but paradoxically it increased the expression of SOST [41]. Bone formation has been suggested to be moderated by a direct negative feedback loop involving a putative CTLA-4 Ig association with CD80/CD86 on the osteoblasts, causing the production of sclerostin [14,43]. In the present patient series, an increased serum level of SOST was signi cantly correlated with the improvement of disease activity after the introduction of abatacept.
We also observed that the serum level of OPG was signi cantly elevated after the introduction of abatacept, and this elevation was signi cantly correlated with the improvement of disease activity. The RANK-RANKL system is the major driver of bone destruction in in ammatory arthritis [12,14,15]. This system is promoted by proin ammatory cytokines such as TNFα and IL-6 [12,14,15]. OPG, a decoy receptor of RANKL, in uences bone erosions in RA [15,44]. A low OPG/RANKL ratio has thus been associated with increased radiographic damage in RA [15,45]. Wnt signaling is involved in osteoclastogenesis regulation; the canonical Wnt/β-catenin signaling pathway leads to an up-regulation of OPG and a down-regulation of RANKL [14,35]. TNF inhibitors [14,46] or IL-6R inhibitor [14,18,19] increased both the expression of OPG and the OPG/RANKL ratio, possibly due to a promotion of Wnt signaling following a decrease in DKK-1. Similarly, abatacept may elevate the serum level of OPG because it promotes Wnt signaling as described above. In addition, the serum levels of OPN among our patients were higher and those of OC tended to be lower compared to the healthy volunteers as in previous reports [21,23]. However, we did not observe any effect of abatacept treatment on these bone biomarkers.
Some limitations of our study should be mentioned. The limited sample size (n = 59) does not allow for subanalyses of differences due to the patients' heterogeneous characteristics. However, our results are valuable as a part of a multicenter collaborative study that prospectively and closely evaluates disease activity using MSUS. Second, we could not evaluate structural changes in joints. In the cohort study, we evaluated the patients' X-ray images at baseline, 6, 12, 18, and 24 months. We will explore whether bone biomarkers are associated with radiographic progression in RA patients treated with abatacept.

Conclusions
This is the rst study to evaluate the effects of abatacept treatment on bone biomarkers in RA patients and to explore whether bone biomarkers are associated with the patients' therapeutic response using the KUDOS. Abatacept may prevent bone destruction through the promotion of the Wnt signaling pathway. In addition, the measurements of the serum levels of bone biomarkers may be useful for predicting the ultrasonographic response to abatacept. The study was approved by the Institutional Review Board of Nagasaki University (Approval No. 13102866). All patients gave signed informed consent. Availability of data and materials:

Abbreviations
The datasets used and/or analyzed during the present study are available from the corresponding author on reasonable request.   Changes in the clinical disease activity and MSUS scores over the 6-month abatacept treatment period.