A 28-day course of Robenacoxib proved effective in reducing the lameness scores of dogs with stifle osteoarthritis secondary to failure of the cranial cruciate ligament. The reduction in lameness scores was statistically significant. It is accepted that a deficiency of the study was the use of a subjective assessment of lameness although all the lameness assessments were carried out by the same individual who was not aware of the initial score when re-evaluating the animal a month later. More objective measurements of lameness such as force plates and pressure mats were not available for this study. The use of activity monitors was not feasible since part of the case management involved restricted exercise during the course of the NSAID. The analgesic properties of NSAIDs for joint pain are well known and clinical improvement has been reported with many different compounds  Robenacoxib preferentially inhibits the COX-2 enzyme which is involved in the pathophysiology of joint pain and central sensitization . It is possible that some of the clinical improvement was due to the restricted exercise regime used, although all of the dogs had already been placed on a controlled exercise plan by the referring veterinarian.
Although in 58% of cases the radiographic scores did not change, overall there was a statistically significant increase in the radiographic score over the period of the study indicating that the radiographic changes had progressed. This is to be expected since once initiated, OA is a gradually progressive disease and there is little evidence that NSAIDs can influence the structural pathological changes that occur within the joint. The radiographic score that was used in this study is very much biased towards assessing osteophyte formation and changes in bone opacity and the only study which has shown an effect of a NSAID on such bony changes is an experimental study in the dog [38, 39]. These studies involved only small numbers of dog and the protocol meant that the NSAID was given from day one when the arthritis was initiated (by sectioning the cranial cruciate ligament), a situation very different from the clinical one, where the disease is generally well advanced before clinical signs are evident and treatment required.
The synovial fluid score based on total nucleated cell counts was not significantly affected by the course of Robenacoxib. Although it is generally accepted that the total nucleated cell count (neutrophils, lymphocytes and macrophages) does reflect the level of inflammation within the joint , there have been no studies validating the use of nucleated cell counts as an outcome measure in the treatment of osteoarthritis.
A major finding in this study was the statistically significant decrease in the concentration of CRP in synovial fluid following the month’s course of Robenacoxib. There was no statistically significant difference in the concentration of serum CRP between the pre- and post-treatment samples. This supports the view of Hurter et al.  who concluded that serum CRP was not a useful marker of OA in the dog. Our study is the first to have reported CRP concentrations in the synovial fluid of dogs and the results support the possibility that synovial fluid CRP is a useful biomarker of joint inflammation and a useful monitor of response to anti-inflammatory treatment. The concentrations of CRP detected in the synovial fluid were much lower than the blood levels. Although the concentrations of synovial fluid CRP in the OA dogs were statistically significantly higher than found in normal synovial fluid, in approximately a third of cases the levels fell within the normal range even though they still decreased after therapy. The measurement of these relatively low levels of CRP does require a very sensitive assay such as the ELISA used in this study. Immunoturbidimetric assays, used in previous reports [41, 42] had a limit of detection of 1 mg/L and are thus not sensitive enough to provide meaningful results, whereas the ELISA used in our study had a lower limit of detection of 0.05 mg/L. The assay was validated for use in measuring this protein in synovial fluid and demonstrated equivalent precision to that for serum CRP while the accuracy of the assay was also acceptable with a recovery rate of 113% from spiked samples. Whilst the antibody used in the Western blot was not the same as used in the ELISA, the immune blot does confirm that CRP is present in canine synovial fluid.
It is possible that drugs used in inducing and maintaining general anaesthesia can cause alterations in the blood (and therefore perhaps also in synovial fluid) concentration of CRP in both humans [43–45] and dogs  although in all these studies the patients also underwent surgery and this is much more likely to be the stimulus for elevation of CRP levels. In our study, both pre- and post-treatment blood samples were collected from conscious dogs but the 2 synovial fluid samples were collected whilst the dogs were anaesthetized, within minutes of the dog losing consciousness and before any surgical stimulus. The fact that both synovial fluid samples were collected under general anaesthesia means that comparison of CRP concentrations in the pre-and post-treatment samples should not be compromised, even if the anaesthetic agents do have an effect since the same anaesthetic protocol was always used and any effect of the anaesthetic is likely to take several hours .
The origin of synovial CRP is not known. It could originate from the blood or be locally produced by the synovial membrane. There was no correlation between the concentrations of CRP in synovial fluid and matched serum sample suggesting that levels in the synovial fluid were not merely a reflection of the serum levels. However, it is possible that CRP in synovial fluid is solely a result of transfer from blood but that there is sufficient variation in the rate of transfer between dogs leading to the lack of correlation. Alternatively, blood derived CRP may be diluted to varying degrees by the synovial fluid. Arthritic joints tend to have much greater volumes of synovial fluid than normal joints, which can vary between cases, although even in normal joints synovial fluid concentrations of CRP tend to be lower than matched blood levels. Local production of CRP within the joint could also account for the lack of correlation of blood and synovial levels although to date, there have been no studies published to show local production of CRP within the joint although this has been shown with SAA in man [19, 21], dog  and horse . There have been several studies which have shown the local production of CRP in other inflamed tissues proving that the liver is not the only source although it is probably the major producer, particularly when the patient is suffering from a systemic illness. A study by Maekawa et al.  demonstrated a significantly elevated gene expression of CRP in human patients with gingivitis and periodontitis. The CRP gene was up-regulated in fibroblasts and endothelial cells rather than the epithelial cells of the oral mucosae. Another study demonstrated CRP gene expression in adipose tissue from human patients . Adipose tissue is commonly found within the joint capsule and, in addition, the stifle joint contains the infra-patellar fat pad and this coupled with the fact that fibroblasts are the commonest cell found within the joint capsule means that it is highly likely that CRP is produced locally within the joint. Further studies, particularly determining CRP gene expression within the synovium and looking for synovial isotypes of CRP are required.
Robenacoxib has been shown to persist within inflamed tissues for much longer than it remains in the circulation . This has lead to the concept of tissue selectivity  by which a drug concentrates in the inflamed tissues where it is required and has a prolonged anti-inflammatory effect. The fact that the half-life of the drug is reduced in the circulation means it is less likely to cause adverse effects on tissues such as the liver and kidney thus contributing to an improved safety profile. The fact that Robenacoxib appears to reduce synovial levels of CRP supports a local effect and the concept of tissue selectivity.
CRP production is mediated by the catabolic cytokines, particularly IL-6, TNFα and IL-1. It is known that PGE2 regulates the production of cytokines such is IL-6, IL-8, IL-11, macrophage colony stimulating factor and vascular endothelial growth factor [50–52], possibly involving the activation of the PGE2 receptors, EP2 and EP4 with an increase in cyclic AMP. By inhibiting PGE2 activity and thus cytokine production, NSAIDs will affect the acute phase response and also have other beneficial effects eg IL-6 is known to stimulate PGE2 production and matrix metalloproteinase production within the joint leading to articular cartilage damage, macrophage colony stimulating factor activates monocytes/macrophages to produce more catabolic cytokines and vascular endothelial growth factor plays an important role in angiogenesis and endothelial migration during the development of synovitis [53–56]. These effects would be additional benefits of using Robenacoxib (and other NSAIDs) in the treatment of osteoarthritis and help to explain why NSAIDs at the level of the joint may result in the slowing of disease progression [38, 39]. We found no correlation between CRP concentrations (in blood or synovial fluid) and lameness or radiographical (pre-treatment) scores.
It is worth noting that in dogs 8, 9 and 30 where the NSAID was stopped prematurely, there was a marked increase in the concentration of synovial fluid CRP in the post-treatment sample (18, 33 and 49 fold increases respectively). This was significantly greater than for any other case where there had been an increase after treatment; in all these other cases, the animal was still receiving the NSAID when the post-treatment sample was taken. This observation raises the question as to whether the NSAID should be gradually withdrawn after a therapeutic course rather than finished abruptly although further studies are necessary to investigate this.