Modulation of RANKL and osteoprotegerin expression in synovial tissue from patients with rheumatoid arthritis in response to disease-modifying antirheumatic drug treatment and correlation with radiologic outcome

Authors


Abstract

Objective

To demonstrate the effect of treatment with disease-modifying agents on the expression of osteoprotegerin (OPG) and RANKL in the synovial tissue from rheumatoid arthritis (RA) patients and to correlate these changes with radiologic damage measured on sequential radiographs of the hands and feet.

Methods

Synovial biopsy specimens were obtained at arthroscopy from 25 patients with active RA (16 of whom had a disease duration <12 months) before and at 3–6-month intervals after starting treatment with a disease-modifying agent. Immunohistologic analysis was performed using monoclonal antibodies to detect OPG and RANKL expression, with staining quantitated using computer-assisted image analysis and semiquantitative analysis techniques. Serial radiographs of the hands and feet were analyzed independently by 2 radiologists and a rheumatologist using the van der Heide modification of the Sharp scoring method.

Results

Thirteen patients achieved a low disease state as defined by a disease activity score <2.6 while 19 patients achieved an American College of Rheumatology response >20% after disease-modifying antirheumatic drug (DMARD) treatment. Successful DMARD treatment resulted in an increase in OPG expression and a decrease in RANKL expression at the synovial tissue level, which correlated with a reduction in erosion scores measured on annual radiographs of the hands and feet.

Conclusion

Successful treatment-induced modulation of OPG and RANKL expression at the synovial tissue level, resulting in a reduction in the RANKL:OPG ratio, is likely to have a significant impact on osteoclast formation and joint damage in patients with active RA.

INTRODUCTION

Rheumatoid arthritis (RA) is characterized by inflammation of the synovial membrane leading to invasion of synovial tissue into the adjacent cartilage matrix with degradation of articular cartilage and bone as a consequence. This results in erosion of bone, which is often observed as marginal joint erosions radiographically and is predictive of a poorer prognosis (1). Although the pathophysiologic mechanisms for cartilage and bone destruction in RA are not yet completely understood, it is known that matrix metalloproteinases, cathepsins, and mast cell proteinases can contribute to cartilage and bone destruction in RA (2–4). However, it is now clear that osteoclast formation and activation at the cartilage–pannus junction is an essential step in the destruction of bone matrix in patients with RA (5–8). A number of inflammatory cytokines found in the synovial tissue of patients with RA (interleukin-1α [IL-1α] and IL-1β, IL-6, tumor necrosis factor α, macrophage colony-stimulating factor [M-CSF]) have the potential to promote osteoclast formation and bone resorption (9–11). However, recent evidence indicates that the interaction between RANKL and RANK has an essential role in osteoclastogenesis (5, 6, 12, 13). RANKL is expressed on osteoblasts, fibroblasts, and T cells, whereas RANK is mainly expressed on preosteoclasts, possibly of macrophage lineage. There is a naturally occurring inhibitor of the RANKL interaction with RANK, called osteoprotegerin (OPG), which binds RANKL with high affinity, preventing RANKL from interacting with RANK (14).

We have previously published reports on the expression of both RANKL (15) and OPG (16) in the synovial tissue of patients with inflammatory arthritis and osteoarthritis as well as normal synovial tissue (17) and demonstrated a lack of OPG expression with significant RANKL expression in the synovial tissue from patients with active RA. We hypothesized that successful treatment with disease-modifying antirheumatic drugs (DMARDs) would reduce RANKL expression and increase OPG expression, altering the RANKL:OPG ratio and suppressing osteoclast formation in the synovial tissue of patients with RA. The goal of the present study was to test this hypothesis in a cohort of RA patients with active disease initiating DMARD treatment and to attempt to correlate the changes in synovial tissue expression of RANKL and OPG with radiologic outcomes in this patient cohort.

PATIENTS AND METHODS

Patients.

Twenty-five RA patients with active synovitis, including an involved knee joint, were recruited for the study. All RA patients fulfilled the American College of Rheumatology (ACR; formerly the American Rheumatism Association) criteria for RA (18). The mean age of the patient group was 68.4 years (range 47–87 years) with 14 men and 11 women. Eighteen of the 25 patients were seropositive for rheumatoid factor, while 2 of the 7 seronegative patients had radiologic evidence of erosions at study entry. The mean disease duration was 4.3 years (range 0.1–24 years); 16 patients had a disease duration <1 year. Decisions about which DMARD to use to treat an individual patient were made by the treating rheumatologist and were not influenced by participation in this study. DMARDs used included methotrexate (6 patients), intramuscular (IM) gold (6 patients), methotrexate and IM gold (9 patients), sulfasalazine (2 patients), cyclosporin A (1 patient), and hydroxychloroquine (1 patient). No anti–tumor necrosis factor (anti-TNF) agents were used to treat this patient group because the study preceded the general availability of these therapeutic agents in Australia. Failure to respond to treatment with the original DMARD along with clear evidence of radiologic progression led to withdrawal from the study. All patients were followed up at regular intervals (3–6 months) for clinical (tender and swollen joint counts, visual analog scales for pain, patient and physician global assessments, and a modified Health Assessment Questionnaire [HAQ]), laboratory (C-reactive protein, erythrocyte sedimentation rate, and rheumatoid factor), and radiologic (radiographs of the hands and feet taken annually) parameters. Response to DMARD treatment was assessed by calculating a disease activity score in 28 joints (DAS28) (19) and ACR response (20). All patients gave informed consent, and the study protocol was approved by the research and ethics committee of the Repatriation General Hospital.

Synovial tissue.

A small-bore arthroscopy (2.7-mm arthroscope; Dyonics, Andover, MA) was performed with patients under local anesthesia as previously described (21) at baseline and at 3, 6, 12, 18, 24, and 36 months after starting DMARD treatment. Biopsy specimens of synovial tissue were obtained from all accessible regions of the knee joint, but mainly from the suprapatellar pouch. The samples were separately snap-frozen in Tissue-Tek OCT (Miles Diagnostics, Elkhart, IN) and stored at −80°C until used. Cryostat sections (6 μm) were mounted on glass slides (Superior Marienfeld, Baden-Wurttemberg, Germany). The glass slides were boxed and stored at −20°C until immunohistologic analysis.

Immunohistochemistry.

Serial sections were stained with the following mouse monoclonal antibodies (mAb): anti-human OPG antibodies (mAb 805 and mAb 8051; R&D Systems, Minneapolis, MN); anti-human TRANCE (mAb 626; R&D Systems); anti-CD68 (clone EBM11; Dako Australia, Botany Bay, New South Wales, Australia) to detect macrophages; mAb 67 (Serotec, Kidlington, Oxford, UK), which recognizes CD55, to detect fibroblast-like synoviocytes; anti-CD3 (BD Biosciences, San Jose, CA) and anti-CD45RO to detect T cells and memory T cells, respectively; anti-CD22 (Central Laboratory of the Netherlands Red Cross Blood Transfusion Service, Amsterdam, The Netherlands) to detect B cells; anti-CD38 (BD Biosciences) to detect plasma cells; and anti–granzyme B (Novo Castra Laboratories, Newcastle upon Tyne, UK). Endogenous peroxidase activity was inhibited using 0.1% sodium azide and 1% hydrogen peroxide in Tris–phosphate buffered saline. Staining for cell markers was performed as described previously (15–17). Following a primary step of incubation with mAb, bound antibody was detected according to a 3-step immunoperoxidase method. Horseradish peroxidase (HRP) activity was detected using hydrogen peroxide as substrate and aminoethylcarbazole (AEC) as dye. Slides were counterstained briefly with hematoxylin solution and mounted in Gurr Aquamount (BDH, Poole, UK). Affinity-purified HRP-conjugated goat anti-mouse antibody was obtained from Dako, affinity-purified HRP-conjugated swine anti-goat Ig from Tago (Burlingame, CA), and AEC from Sigma (St. Louis, MO). The specificity of the antibodies against RANKL (15) and OPG (16) has previously been demonstrated by absorption studies using purified RANKL and OPG.

Quantitation of immunohistochemical labeling.

After immunohistochemical staining, sections stained for OPG (mAb 805 and 8051) and RANKL were analyzed in a random order by computer-assisted image analysis, analyzing 6 high-power fields for each section as previously described (22, 23). In addition, these sections were scored by a semiquantitative method on a 5-point scale by 2 independent observers in a random order, as described previously (15, 16, 24).

Real-time polymerase chain reaction on synovial tissue.

Complementary DNA (cDNA) samples were prepared from synovial biopsy specimens from 5 RA patients using the initial synovial tissue obtained before starting DMARD treatment and again using the biopsy specimens obtained when there was significant improvement in disease activity as defined by the DAS28. As controls for these patients, cDNA was prepared from synovial tissue obtained at arthroscopic biopsy in 5 patients with osteoarthritis of the knee joint. Real-time polymerase chain reaction (PCR) was performed using Platinum SYBR Green quantitative PCR Supermix-UDG (Invitrogen Life Technologies, San Diego, CA) as per the manufacturer's recommendations. Amplification was carried out in a Rotor-Gene 3000 (Corbett Life Science, Mortlake, New South Wales, Australia) with SYBR Green detection and melt curve analysis. Oligonucleotide primers that were used have been described previously, and are specific for OPG and RANKL (25). The endogenous reference gene hARP (26) was used to normalize threshold cycle (Ct) data obtained from the genes investigated. Reaction mixtures contained 1 μl of 1:5 diluted cDNA, 7.5 μl Platinum SYBR Green quantitative PCR Supermix-UDG, 300 nM each of forward and reverse primer, and diethyl pyrocarbonate–treated water to a final volume of 15 μl. All samples were investigated in triplicate and the melting curves obtained after each PCR amplification confirmed the specificity of the SYBR Green assays. Optimization of forward and reverse primer concentrations between 50 nM and 900 nM was evaluated to obtain the combination of primers with the lowest Ct value. For each target gene (OPG and RANKL) and endogenous reference gene (hARP), a concentration of 300 nM of both forward and reverse primers yielded the lowest Ct values, with the highest increase in fluorescence. Validation experiments were performed to demonstrate that amplification efficiencies of the target genes and the endogenous reference gene were approximately equal. Complementary DNA was prepared from pooled RNA samples and diluted in a 2-fold dilution series over 6 orders of magnitude. Target and reference genes were then amplified in separate tubes using this cDNA dilution series and Ct were values obtained. The difference in Ct (ΔCt) for each sample was then calculated as outlined below and data were plotted against the log cDNA dilution. If the absolute value of the slope is <0.1, the efficiencies of the target and reference genes are similar, and the ΔΔCt calculation for the relative quantification of target may be used (27). The slope for OPG hARP was 0.078 and for RANKL hARP was 0.0125. Relative expression of the target genes in the studied samples was obtained using the difference in the Ct (ΔΔCt) method. Briefly, for each sample, a value for threshold (Ct) was determined, defined as the mean cycle at which the fluorescence curve reached an arbitrary threshold. The ΔCt for each sample was then calculated according to the formula Ct target gene − Ct hARP; ΔΔCt values were then obtained by subtracting the ΔCt of a reference sample (the average ΔCt for osteoarthritis samples) from the ΔCt of the studied samples. Finally, the levels of expression of the target genes in the studied samples as compared with the reference sample were calculated as 2math image.

Grading of serial radiographs.

Radiographs of the hands and feet were obtained for all patients included in this study ∼12 months apart for the duration of the study using a standard technique. The radiographs were graded by 2 radiologists and a rheumatologist, using the van der Heide modification of the Sharp technique (28), after first standardizing against each other using a series of hand and feet radiographs from 4 RA patients not included in this study. The radiographs were graded separately by the 3 assessors without knowledge of the clinical outcome, but with knowledge of the timing of the radiographs. The scoring system results in a maximum score of 210 for joint space narrowing and 280 for erosions. A disagreement between scorers of up to 5% of the maximum score was allowed, with the final score being an average of the 3 scores. Any disagreement above this level was settled by consensus, with all 3 assessors scoring all radiographs for a single patient together in the same session. This was necessary for 4 of the 25 patients included in this study.

Statistical analysis.

Results are presented as the mean ± SD. Tests for normality were applied to the data. Semiquantitative scores were treated as nonparametric data. Data generated by computer-assisted digital image analysis (integrated optical density [IOD]) were generally normally distributed. Changes within the groups were analyzed using Wilcoxon's signed rank test for data that were not normally distributed (semiquantitative scores) and Student's paired t-test for normally distributed continuous data (IOD). Separate linear regressions of each parameter versus time in months were performed. To account for the correlation between repeated observations on the same patient in the analyses, generalized estimating equations, assuming an exchangeable correlation structure, were used in the regressions. The analyses were also carried out for the differenced data, in which the first synovial biopsy result was subtracted from each subsequent synovial biopsy result for each of the variables.

RESULTS

Clinical and demographic features.

All patients included in this study had active RA, with a mean tender joint count of 17.7 (range 5–26), mean swollen joint count of 13.8 (range 3–27), mean modified HAQ score of 2.3 (range 1.3–3.0), and mean DAS28 of 5.9 (range 4.4–7). There was a significant decrease in DAS28 with treatment (P = 0.000). Thirteen patients attained a low disease activity state as defined by a DAS28 score <2.6 following DMARD treatment. Nineteen patients attained a significant ACR response to treatment, defined as a >20% ACR response to treatment, but 6 of these patients did not achieve a low disease activity score as defined by a DAS28 <2.6.

Radiologic outcomes for the patient group.

Despite the predominantly short disease duration of this patient group, patients had evidence of erosions (mean erosion score 16.2, range 0–211) and joint damage (mean joint space narrowing 16.2, range 0–69) at the time of study entry. The patients who failed to respond to DMARD treatment had significant increases in joint damage over time, whereas the patients who responded to DMARD treatment had no real change in radiologic damage over time (Figure 1). Patients who were clearly not responding to treatment and had radiographic evidence of progressive joint damage were removed from the study and offered alternative treatment. For this reason, there was a longer radiologic followup in the responder group than the nonresponder group (7 years versus 3 years).

Figure 1.

Graphic representation of erosion and joint space narrowing scores (JSN; using the van der Heide modification of the Sharp score [28]) over time for A, patients who responded well to disease-modifying antirheumatic drug (DMARD) treatment and B, patients with no response to DMARD treatment, based on the Disease Activity Score in 28 joints. Solid circles = erosion score hands; solid squares = erosion score feet; open squares = total erosion score; open triangles = JSN hands; solid triangles = JSN feet; open circles = total JSN.

Immunohistochemistry results.

As previously described (16), 2 distinct patterns of staining for OPG in synovial tissue were seen: mAb 805 stained exclusively endothelial cells, while mAb 8051 stained mainly the lining layer of the synovial membrane and weakly stained endothelial cells. As shown in Tables 1 and 2, the expression of OPG in both blood vessels (Figure 2) and synovial lining (Figure 3) was either low or absent in patients with active RA, whereas the expression of OPG increased as a result of response to DMARD treatment (P = 0.005). In contrast, RANKL expression was high in the synovial tissue of patients with active RA (Figure 4) and decreased as a result of successful DMARD treatment (P = 0.003). The major changes in synovial tissue inflammatory infiltrate were reductions in macrophage and T cell content as a result of DMARD treatment, with little change in B lymphocytes, plasma cells, or granzyme B–positive cells, as previously shown (29, 30). We assessed whether the change in T cell content of synovial tissue from patients treated with DMARDs could explain the decrease in RANKL expression in these patients by fitting regression models using generalized estimating equations that included time and CD3 content as covariates with an exchangeable correlation structure. The effect of DMARD treatment on CD3 content of synovial tissue did not explain the changes seen in RANKL content of synovial tissue from patients treated with DMARDs. There was an increase in fibroblast lining cell (CD55 positive) content of the synovial tissue so that patients who responded to drug treatment had a synovial lining layer structure approaching that of normal synovial tissue (17), as we have previously demonstrated (29). Although OPG content did increase with DMARD treatment, we have previously shown that OPG is produced mainly by endothelial cells and macrophages (16) rather than fibroblast lining cells, therefore changes in these cells in synovial tissue from patients treated with DMARDs (both decreased) could not explain the increase in OPG content of synovial tissue from these patients.

Table 1. Results of immunohistochemical labeling of synovial tissue from rheumatoid arthritis patients with a good clinical response to disease-modifying antirheumatic drug treatment based on a DAS28 <2.6*
Time of synovial biopsy since baseline, yearsIOD 8051IOD 805IOD RANKLArea CD68Area CD55IOD B cellsIOD plasma cellsIOD T cellsIOD memory T cellsIOD granzyme B cellsDAS28
  • *

    Values are the mean ± SD. DAS28 = Disease Activity Score in 28 joints; IOD = integrated optical density, measured by image analysis; 8051 = monoclonal antibody that measures monomeric osteoprotegerin (OPG) produced mainly by macrophage-like synovial lining cells; 805 = monoclonal antibody that measures dimeric OPG, mainly expressed by endothelial cells; CD68 = monoclonal antibody that detects macrophage lineage cells; CD55 = monoclonal antibody that detects decay activator factor, a marker for synovial lining fibroblasts.

Baseline398 ± 657406 ± 7806,432 ± 8,22041,354 ± 34,19815,152 ± 14,045492 ± 68612,748 ± 12,3865,467 ± 4,0659,801 ± 8,54623 ± 155.5 ± 0.6
0.25514 ± 716963 ± 1,0465,935 ± 8,09637,911 ± 25,54116,459 ± 13,694371 ± 71210,656 ± 12,0754,407 ± 4,3708,744 ± 6,78930 ± 264.2 ± 1.9
0.51,597 ± 2,3082,483 ± 1,1664,980 ± 7,26227,383 ± 16,89223,254 ± 21,715786 ± 5474,335 ± 5,3833,911 ± 4,0718,170 ± 5,669278 ± 2342.5 ± 2
1.02,657 ± 3,3162,359 ± 1,2293,246 ± 6,66918,210 ± 15,46431,411 ± 25,813734 ± 5907,359 ± 6,5433,591 ± 4,4245,911 ± 5,390192 ± 2051.8 ± 1.5
1.52,862 ± 3,4022,141 ± 1,3423,109 ± 5,91914,392 ± 13,04139,339 ± 26,838764 ± 6902,952 ± 3,9631,899 ± 2,1284,793 ± 4,975158 ± 2071.5 ± 1.2
23,934 ± 4,7472,227 ± 1,1562,118 ± 4,82715,495 ± 13,24142,376 ± 32,341864 ± 5604,674 ± 4,3294,172 ± 3,0314,419 ± 3,59343 ± 211.5 ± 1.2
34,023 ± 4,7142,118 ± 9701,097 ± 2,80814,082 ± 14,83243,329 ± 25,6781,014 ± 8754,390 ± 3,3784,020 ± 2,3424,082 ± 4,77963 ± 701.5 ± 1.3
Table 2. Results of immunohistochemical labeling of synovial tissue from rheumatoid arthritis patients who demonstrated no clinical response to disease-modifying antirheumatic drug treatment, based on a DAS28 <2.6*
Time of synovial biopsy since baseline, yearsIOD 8051IOD 805IOD RANKLArea CD68Area CD55IOD B cellsIOD plasma cellsIOD T cellsIOD memory T cellsIOD granzyme B cellsDAS28
  • *

    Values are the mean ± SD. See Table 1 for definitions.

Baseline547 ± 835696 ± 5137,372 ± 7,26022,665 ± 19,70223,172 ± 22,329434 ± 4538,904 ± 9422,332 ± 2,3848,498 ± 9,00468 ± 786.3 ± 0.7
0.25882 ± 901774 ± 6297,115 ± 7,50821,555 ± 19,75423,648 ± 24,015427 ± 4238,722 ± 7591,874 ± 1,1788,432 ± 9,97360 ± 776.0 ± 1.2
0.51,293 ± 994992 ± 7785,145 ± 5,70822,826 ± 19,87122,478 ± 27,290135 ± 2099,091 ± 8,0851,951 ± 1,3066,721 ± 6,77959 ± 775.5 ± 1.0
1.01,167 ± 9271,059 ± 9995,123 ± 6,23926,909 ± 19,46022,646 ± 26,192115 ± 2165,989 ± 5,4101,908 ± 1,3245,318 ± 6,96164 ± 804.7 ± 1.1
1.51,280 ± 9221,174 ± 8394,912 ± 6,37418,678 ± 16,54822,950 ± 26,095157 ± 2265,887 ± 5,4101,929 ± 2,3425,329 ± 5,96162 ± 804.8 ± 1.1
21,219 ± 1,1061,032 ± 9394,925 ± 6,36217,967 ± 20,15626,060 ± 26,670160 ± 2256,019 ± 7,3931,896 ± 1,3295,333 ± 6,54361 ± 804.7 ± 1.3
31,484 ± 1,0901,051 ± 9255,667 ± 3,66418,018 ± 15,29127,863 ± 29,001200 ± 2487,362 ± 7,1262,252 ± 2,6966,309 ± 5,71863 ± 464.5 ± 1.5
Figure 2.

Immunohistochemical labeling of sequential synovial biopsy samples from the same knee joint in a patient with rheumatoid arthritis obtained at A, baseline, B, 3 months, C, 6 months, and D, 12 months after starting methotrexate. Biopsy samples are labeled with monoclonal antibody 805, detecting osteoprotegerin in blood vessels, with aminoethylcarbazole as the chromogen (red color). (Original magnification × 400.)

Figure 3.

Immunohistochemical labeling of sequential synovial biopsy samples from the same knee joint in a patient with rheumatoid arthritis obtained at A, baseline, B, 3 months, C, 6 months, and D, 12 months after starting methotrexate. Biopsy samples are labeled with monoclonal antibody 8051, detecting osteoprotegerin in the synovial lining, with aminoethylcarbazole as the chromogen (red color). (Original magnification × 400.)

Figure 4.

Immunohistochemical labeling of sequential synovial biopsy samples from the same knee joint in a patient with rheumatoid arthritis obtained at A, baseline, B, 3 months, C, 6 months, and D, 12 months after starting methotrexate. Biopsy samples are labeled with a monoclonal antibody detecting RANKL, with aminoethylcarbazole as the chromogen (red color). (Original magnification × 400; insets, original magnification × 800.)

Real-time PCR measurements on cDNA from synovial tissue.

Real-time quantitative PCR confirmed the results demonstrated using immunohistochemical labeling of RA synovial tissue. There was a decrease in RANKL detection and an increase in OPG detection at the messenger RNA (mRNA) level when the followup synovial biopsy was compared with the initial synovial biopsy in RA patients started on DMARD treatment (data not shown). Average ΔΔCt for OPG was 0.57 for active RA patients and 1.11 for inactive RA patients, while the average ΔΔCt for RANKL was 1.31 for active RA patients and 0.18 for inactive RA patients. This confirmed the results for the immunohistochemical labeling studies at the mRNA level.

Correlation of immunohistochemical labeling for RANKL and OPG with changes in clinical and radiologic parameters.

Although there was an increase in synovial tissue expression of OPG with DMARD treatment, the correlation with change in DAS28 failed to achieve statistical significance (r = 0.402, P = 0.052). The decrease in synovial tissue expression of RANKL (measured by either image analysis or semiquantitative score) did correlate with the reduction in DAS28 with DMARD treatment (r = 0.448, P = 0.037). There was also a significant correlation between changes in DAS28 and changes in joint space narrowing score (r = 0.454, P = 0.023), with the least change in joint space narrowing score seen in patients with the greatest reduction in DAS28. Similarly, there was a significant correlation between changes in RANKL expression in synovial tissue (measured with either image analysis or semiquantitative score) and changes in erosion score (r = 0.538, P = 0.007), with the smallest changes in erosion score seen in patients with the greatest reduction in RANKL expression in synovial tissue following DMARD treatment. There was no correlation of changes in either RANKL or OPG expression in the synovial tissue with changes in joint space narrowing with DMARD treatment. We did not demonstrate any effect of disease duration on changes in OPG or RANKL seen with DMARD treatment. We attempted to calculate RANKL:total OPG, RANKL:endothelial OPG, and RANKL:synovial lining OPG ratios, but the data were skewed in distribution and neither logarithmic nor inverse transformation of the data led to a meaningful analysis.

DISCUSSION

Osteoclasts are responsible for the resorption of bone during normal bone metabolism and the destruction of bone seen in a variety of pathologies such as RA. It is now clear that M-CSF and RANKL are essential factors required for the development of osteoclasts. Osteoclasts form from cells isolated from the RA joint, with large numbers rapidly forming from cells isolated from the pannus region (8). The cartilage–pannus junction in RA contains many types of cells that produce inflammatory cytokines reported to stimulate osteoclast differentiation and bone resorption, including IL-1α and IL-1β, IL-6, IL-11, and TNFα (9). The end result of the production of inflammatory cytokines, such as IL-1β and TNFα, in the inflamed joint is likely to be the up-regulation of RANKL (produced by T cells, fibroblasts, and osteoblasts) (8, 13) and RANK (expressed by preosteoclasts, T cells, and dendritic cells) (13). OPG is an alternative, high-affinity decoy receptor for RANKL that blocks the interaction between RANKL and RANK and significantly inhibits osteoclastogenesis (14, 16). Similar to RANK and RANKL, OPG production is stimulated in vitro by proinflammatory cytokines, such as IL-1β and TNFα. RANK, RANKL, and OPG are expressed in synovial tissue from the RA joint (5, 6, 8, 15, 16).

We have previously demonstrated the expression of OPG in synovial tissue (both lining and endothelial expression) from patients with both inflammatory arthritis and osteoarthritis, as well as normal synovial tissue (16). The notable exception to this was in RA synovial tissue: little or no OPG was expressed in the synovial tissue from patients with active RA, whereas OPG was expressed both in the synovial lining and on endothelial cells in the synovial tissue from patients with inactive RA. In contrast, we have also demonstrated the expression of RANKL mainly in the synovial tissue of patients with active RA (15). This leads to a local synovial environment in the active RA joint where osteoclast formation would be likely, whereas in the inactive RA joint osteoclast formation would be suppressed. The current study was designed to test this hypothesis by examining the expression of RANKL and OPG at the synovial tissue level in patients with active RA started on treatment with DMARDs and by attempting to correlate changes in synovial expression of RANKL and OPG with radiologic outcomes for this RA patient cohort. Because it was not predictable if and when an individual patient would respond to DMARD treatment and when any change in synovial membrane expression of RANKL or OPG would occur, synovial biopsies were performed up to 36 months after initiation of treatment, if the patient remained in the study. Our study is the first to demonstrate that successful DMARD treatment of patients with RA can result in the reduction of RANKL expression and an increase in OPG expression at the synovial tissue level, with a resultant reduction in the RANKL:OPG ratio, which would reduce the likelihood of osteoclast formation and, theoretically, joint erosion and joint space narrowing. We have confirmed these results at the mRNA level using a subset of RA patients included in this study. Our study is also the first to demonstrate that there is a correlation between changes in DAS28 and both synovial expression of RANKL and OPG as well as joint erosion measured radiologically, as a result of successful DMARD treatment. Although we had a high rate of response to conventional DMARD treatment (52% with low disease activity based on a DAS28), it should be noted that 72% of the patients included in this study had a disease duration <1 year at study entry. The early-disease RA population has previously been shown to have a higher and similar response rate to conventional DMARDs and biologic treatments (31, 32). Similarly, our patient population had a relatively high erosion rate at study entry, but previous studies have shown that this is not unusual, with most erosions being detectable within the first 2 years of disease (33, 34). Our RA patient population is also unusual in that there were more men than women. Although sex may have some effect on clinical and radiologic outcomes, it is small compared with the effect that the presence of rheumatoid factor (or anti–cyclic citullinated peptide antibodies) and erosions on radiograph at disease onset has on these outcomes. Despite the above caveats, we believe that our results are generalizable to the global rheumatology population.

A recent study by Catrina et al (35) also demonstrated that OPG expression was low in RA synovial tissue while RANKL expression was high in patients with active disease, and that treatment with TNF blockers (either etanercept or infliximab) altered the RANKL:OPG ratio in favor of a reduction in osteoclast formation. However, unlike our previous findings, these authors failed to find any correlation between changes in disease activity and changes in synovial tissue expression of RANKL and OPG, and suggested that the changes they observed were a direct result of TNF blockade. The patient group included in the study by Catrina et al was similar to that used in our study, but Catrina et al did not indicate the disease duration of their patient group, so it is not possible to establish whether the RA patients they studied had early RA, as did the majority of the patients included in our study. In addition, there was considerable coprescription of both corticosteroids and DMARDs (mainly methotrexate) with infliximab and etanercept in their study, so it is unclear whether the effects on OPG and RANKL expression in RA synovial tissue were solely due to TNF inhibitors. Finally, Catrina et al only followed patients for 8 weeks and the clinical responses seen in these patients (ACR20 responses and reduction in DAS28 scores to mean levels of 3.8 [etanercept] and 4.2 [infliximab]) were inferior to those seen in the patients included in our study. A further study by the same research group (36) demonstrated that intraarticular corticosteroids could significantly reduce the synovial content of CD3-positive T lymphocytes and the expression of RANKL by these cells without affecting synovial expression of OPG, potentially reducing the risk of erosions in the RA joint, although the researchers did not measure the effect on radiologic outcomes. While these authors did not demonstrate any effect of intraarticular corticosteroids on synovial tissue OPG expression, the antibody they used detects only OPG on endothelial cells and not on macrophages, as we have previously demonstrated (16).

The results of our study suggest that a deficiency in OPG expression may have a role in the pathogenesis of bone erosions, which characterize RA, and suggest that OPG may well have a therapeutic role in the future management of RA. Standard DMARD treatment of patients with RA can down-regulate RANKL expression in RA synovial tissue with significant implications for progression of erosive damage within joints over time. The development of treatments for RA that modulate both the inflammatory milieu of the synovial tissue and the mediators of osteoclast formation is likely to result in significant improvement in the clinical, functional, and radiologic outcomes for patients with RA.

AUTHOR CONTRIBUTIONS

Dr. Smith 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 design. Haynes, Crotti, Smith.

Acquisition of data. Haynes, Crotti, Weedon, Slavotinek, Au, Ahern, Smith.

Analysis and interpretation of data. Haynes, Crotti, Coleman, Ahern, Smith.

Manuscript preparation. Haynes, Crotti, Roberts-Thomson, Ahern, Smith.

Statistical analysis. Ahern.

Ancillary