In juvenile idiopathic arthritis (JIA) the temporomandibular joints (TMJs) are affected in more than 60% of the patients, depending on the examination method used 123. TMJ involvement can lead to serious mandibular growth alterations 456 because of a unique intracapsular localization of the mandibular growth zone in the condylar cartilage 7. The use of intra-articular (IA) corticosteroids in the TMJ shows a favourable response in more than 50% by improving jaw movement and reducing pain 8. However, the clinical long-term effects on growth of IA corticosteroid therapy in juvenile TMJs are unknown. In experimental TMJ arthritis we demonstrated that IA corticosteroid reduced inflammation but accentuated mandibular growth disturbances 91011. Therefore a search for alternative local anti-inflammatory treatment is warranted.

The efficacy and safety of etanercept, a soluble tumor necrosis factor (TNF)-α receptor fusion protein, has been proven in a randomized controlled withdrawal-study in children with severe polyarticular JIA 1213. In a small clinical trial, administration of IA etanercept and IA corticosteroid shows comparable results in knee joints of rheumatoid arthritis patients 14. The effects in this study were measured using questionnaires and ultrasonography. To our knowledge no data exist on histology in IA etanercept. The use of IA etanercept is a promising candidate of local inflammation control in the TMJs, since no side effects are expected in reducing the mandibular growth.

In this study we investigated the effect of IA and systemic administration of etanercept in ovalbumin antigen-induced arthritis in the TMJ in young growing rabbits. The effects on inflammation control will be discussed in the present article and the aspects on mandibular growth will be discussed in Part II: Mandibular growth 15.

Forty-two female 10-week-old New Zealand white rabbits (Oryctolagus cuniculus) were used. The rabbits were housed at the animal facilities at Aarhus University, Denmark, and had constant access to food and water. The animal-welfare was supervised daily by the staff at the housing-facilities. All procedures in this study were approved by the Danish ethical committee for experimental animal research.

The study design is shown in figure 1. All rabbits were sensitized to ovalbumin using the method described by Kapila 16, using 1 ml of ovalbumin (1 mg/ml) (Sigma Chemicals, Steinheim, Germany) dissolved in physiological saline and Freund's complete adjuvant^® (IFA; Sigma Chemicals, St. Louis, MO, USA). The procedure was repeated two weeks later using Freund's incomplete adjuvant^®. All animals were tested for sensitivity one week after by subcutaneous injection of 1 ml (1 mg/ml) ovalbumin on a shaved spot (3 × 3 cm) on the back. Injection-site inspection was done twelve hours later for signs of local inflammation. An animal was considered sensitized if the erythema was more than 1 × 1 cm, and the induration was more than 0.5 × 0.5 cm.

Arthritis was induced in both TMJs one week after sensitation by an IA injection with sterile filtered 0.1 ml ovalbumin (10 mg/ml) dissolved in physiological saline. Arthritis was induced a total of four times, with 3-week intervals in all animals. At the fourth arthritis induction the ovalbumin dosage was reduced to 5 mg/ml because of anaphylactic reactions at the 10 mg/ml dose. All IA injections in this study were done under anesthesia with 16 mg ketamine, 0.8 mg xylazine and 0.17 mg acepromazine given subcutaneously. The IA injections were done as described by Kapila 16. The mandible was manipulated under general anesthesia and the condylar head was easily palpated. When inside the lower joint chamber the injection is done slowly.

Animals were randomly assigned to receive either IA saline (2 × 0.1 ml), IA etanercept (Enbrel^®, Wyeth Pharmaceuticals Inc., Collegeville, PA, USA) bilaterally (2 × 0.1 mg/kg) or systemic etanercept (0.8 mg/kg) treatment.

The etanercept dose is calculated as follows:

Desired concentration ( m g / k g ) = Amount of injected fluid ( m l ) ⋅ Injected etanercept concentration ( m g / m l ) Average weight of animals ( k g ) MathType@MTEF@5@5@+=feaagaart1ev2aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacPC6xNi=xI8qiVKYPFjYdHaVhbbf9v8qqaqFr0xc9vqFj0dXdbba91qpepeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=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@CEE3@

IA saline or IA etanercept were administered one week after every arthritis induction. We administered etanercept systemically in the last group once weekly beginning one week after the first arthritis induction. A total of ten animals died before study completion because of anaphylactic shock against the ovalbumin at the 10 mg/ml dose, which prompted dose reduction as mentioned.

Four days after the last treatment, the animals were euthanized for retrieval of histological samples using intravenous pentobarbital (200 mg/ml).

Thirty-two animals were available for histological investigation after completion of the study. Ten animals were lost due to the antigen-induced anaphylactic reaction. Additionally, one animal from the systemic etanercept group was excluded from the calculations as an outlier because of a very high number of plasma cells in the SSCT of more than 3 standard deviations above the second highest. The high cellular count may be explained by an intra-articular infection. However, neither neutrophils nor bacteria were observed in the SSCT. Only animals completing the entire study were used in the statistical evaluation.

Data from the histomorphometry are shown in Figure 4. Neither the fractional surface nor the thickness of the synovial lining changed after treatment with either IA or subcutaneous etanercept.

The semi-quantitative scoring of inflammation in the joint revealed that the systemic etanercept group had significantly lower scores (p = 0.009) using an ANOVA test. When t-tests were done they showed significant differences to the saline (p = 0.013) and the IA etanercept group (p = 0.0017), respectively (Figure 5a). Figure 5b shows the synovial proliferation scores. The systemic etanercept group scores lowest with a p = 0.0025 against the saline group and p = 0.0061 against the IA etanercept group.

No differences in volume of the SSCT were seen between the groups (Figure 6a). The number of plasma cells in the SSCT was significantly lower in the systemic etanercept group compared to both the placebo and IA etanercept groups with an ANOVA test result of p = 0.03 and t-tests of p = 0.04 between the groups, respectively (Figure 6b).

Inflammation in the TMJs in young rabbits was induced and maintained for 12 weeks by a modified method originally described by Kapila et al. 16. The model causes inflammatory changes in the TMJ and contains both an acute and a chronic stage predominated by mononuclear cells 2425, which is characteristic of the chronic inflammation in patients with JIA 26 or rheumatoid arthritis 27. Thus, the histological features of this model resemble the conditions seen in both rheumatoid arthritis and JIA. The ovalbumin-induced arthritis has been thoroughly tested and used for more than 40 years, but as with other experimental models they rarely each reflect the full extent of the complex, multi-faceted inflammatory process seen in human autoimmune disease 2829. One must therefore always be careful when extrapolating data from animal studies to human studies.

TNF-α contributes to joint damage by its diverse actions on proinflammatory cells and cytokines including the receptor activator for nuclear factor-κB ligand (RANKL) pathway 30. TNF-α is elevated in both serum and synovial fluid of children with juvenile idiopathic arthritis 31. Also, high levels of TNF-α present in synovial fluid from the TMJ are related to increased TMJ pain experience in rheumatoid arthritis patients 32. Blocking TNF-α by using subcutaneous etanercept has proven its efficacy and safety for the treatment of polyarticular JIA 1213. Therefore, intra-articular blockade of TNF-α could be a possible therapy for patients with arthritis in the TMJ.

Intra-articular corticosteroids improves jaw movement and reduces symptoms in arthritic temporomandibular joints in children with juvenile idiopathic arthritis 833. However, the long-term effects of repeated IA corticosteroids on the mandibular growth are still unknown. In previous studies we have examined IA corticosteroids in antigen-induced TMJ arthritis in rabbits. Although reducing TMJ inflammation, IA corticosteroids aggravates the arthritis-induced inhibition of the mandibular growth 91011.

The systemically administered etanercept was given once weekly (0.8 mg/kg), which is equivalent to the human dose per week given for the treatment of JIA 3435. The etanercept dose given intra-articularly (0.1 mg/kg per joint) was adjusted from a human study 36. However, no strict regimen for IA etanercept treatment exists 37, and it can be argued that a higher dose should be given. There are currently no studies describing the dose/response of etanercept in rabbits. However, one study reports that even high doses of etanercept is well tolerated in healthy rabbit eyes where a dose of 1 mg/kg was administered 38. This implies that higher doses also can be given IA without toxic effects.

The model is essentially based on antigenic stimulation by ovalbumin, which causes a systemic activation of the immune system, proliferation of B-cells which are transformed to plasma cells releasing antibodies in the synovial tissues. Therefore, because systemic administration with the highest accumulated dose has the greatest potential of suppressing the number of plasma cells systemically. Herein lays a bias in this model when comparing systemic and local drug administration when evaluating the number plasma cells. But what we can see is that local administration of etanercept does not lower the number of plasma cells compared to saline injections, meaning that the IA etanercept does not lower the migration of plasma cells to the inflamed joint. Using this model on rabbit knees Idogawa and co-workers showed that the TNF-production is high in the joint initially after ovalbumin injection and suggested that the inflammation is TNF-driven. However, our IA etanercept injections were performed seven days after ovalbumin injection where the inflammation is chronic. This may explain why the effects of the IA etanercept are low; we speculate that the effect on initial inflammation can be higher than in chronic antigen-induced arthritis.

The lack of effect of IA etanercept could also be explained by its pharmacokinetics. The distribution of etanercept is rapid, and it will therefore quickly diffuse out of the joint. With a half-life between 70.8 and 96.4 hours in the organism it is possible that the IA concentration is rather low after injection 35. The time the etanercept stays within the joint is not known. Also, the accumulated total dose given is twelve times higher in the systemic group and in combination with the rapid distribution of etanercept this could be the reason for the lack of effect of IA etanercept. Applying the etanercept intra-articularly too often could result in scar-tissue formation and is not advised.

The results of the fractional surface measurements did not reveal differences between the groups (Figure 4a). A large percentage of the synovial lining is normal in all groups, and a fractional surface measurement does not categorize the villous hyperplasia or the hyperplastic synovium according to degree, but merely as being present or not. However, when given scores, clear statistical evidence of lower scores in the systemic etanercept group compared to the other groups was seen (Figure 5b). Therefore, systemic etanercept has favorable immunomodulatory effects on the synovial lining in this model.

The thickness of the synovial lining was no different between the groups (Figure 4b). This result was also found in our previous study using IA corticosteroid 9.

Semi-quantitative scoring of the inflammation in the SSCT revealed reduced inflammation in the systemic etanercept group. S-Q scores (Figure 5a) and estimation of the number of plasma cells in the SSCT (Figure 6b) both indicate that the chronic inflammation is effectively reduced in the rabbit.

No difference between the saline and IA etanercept groups was present, neither in scoring of the synovial lining nor the SSCT. Previously, we have found that IA saline injections did not cause inflammation nor accumulation of plasma cells in the non-pathological SSCT 9. Therefore, because the S-Q scores and synovial lining measurements were similar for the IA saline and IA etanercept groups, we conclude that IA etanercept has no significant effect on the parameters studied in this study of chronic inflammation.

Although numerous cytokines are involved in the pathogenesis of JIA 31, TNF-α-blockade by the use of etanercept alone had significant anti-inflammatory effects on experimentally ovalbumin-induced TMJ arthritis. However, in the parameters used in this study IA etanercept (0.1 mg/kg) had no significant effects on the severity of inflammation in ovalbumin antigen-induced arthritis and was inferior to systemic administration of etanercept in management of chronic inflammation.

JIA: Juvenile Idiopathic Arthritis; TMJ: Temporomandibular Joint; IA: Intra-articular; TNF: Tumor Necrosis Factor; S-Q: Semi-quantitative; SSCT: Sub-synovial Connective Tissue; CE: Coefficient of Error; RANKL: Receptor Activator for Nuclear Factor-κB Ligand.

The authors declare that they have no competing interests.

KDK, PS, AK, TKP and TH designed and conducted the experiment. KDK, EH, AK and JRN decided upon the histological methods to be used. KDK did the histomorphometric, stereological data collection and statistics under supervision by AK and JRN. EH conducted the semi-quantitative assessments. KDK drafted the initial manuscript. All authors have made significant contributions to the manuscript regarding content and interpretation and read and approved the final manuscript.