FULL PAPER Pharmacology

Comparison of Oral Robenacoxib and Carprofen for the Treatment of Osteoarthritis in Dogs: A Randomized Clinical Trial

Kazuya EDAMURA1)*, Jonathan N. KING2), Wolfgang SEEWALD2), Nobuhiro SAKAKIBARA3) and Masahiro OKUMURA4)

1)Laboratory of Veterinary Surgery, Department of Veterinary Medicine, College of Bioresource Sciences, Nihon University, 1866 Kameino, Fujisawa, Kanagawa 252–0880, Japan 2)Clinical Development, Novartis Animal Health Inc., Werk Rosental, Schwarzwaldalle 215, Basel, CH-4058, Switzerland 3)Technical Service, Novartis Animal Health K.K., Tokyo 106–0031, Japan 4)Laboratory of Veterinary Surgery, Department of Veterinary Clinical Sciences, Graduate School of Veterinary Medicine, Hokkaido University, Hokkaido 060–0818, Japan

(Received 25 November 2011/Accepted 23 April 2012/Published online in J-STAGE 17 May 2012)

ABSTRACT. The efficacy and tolerability of robenacoxib for the treatment of osteoarthritis in dogs were evaluated in a prospective, multi- center, randomized, noninferiority design clinical trial. A total of 32 dogs presenting with osteoarthritis were allocated randomly to receive, orally once daily for 28 days, either 1–2 mg/kg robenacoxib (n=21) or 3.5–5 mg/kg carprofen (n=11). Dogs were assessed by clinicians and owners using numerical rating scale scores at baseline and days 14 and 28. The primary efficacy endpoint was the global functional disability score, which was the sum of clinician scores for standing posture, lameness at walk, lameness at trot, willingness to raise the contralateral limb and pain at palpation. There was a good to excellent level of efficacy in both treatment groups. Differences between days 14 and 28 compared to day 0 were significant for all 11 clinician and owner scores for robenacoxib, and for 6 of 11 scores for carprofen. The efficacy of robenacoxib was numerically superior to carprofen for all 13 endpoints, but differences were not statistically significant. For the global functional disability score, the estimated efficacy of robenacoxib was 1.244 (95% confidence interval 0.555–2.493) relative to carprofen. The tolerability of both treatments was good as assessed from adverse events, clinical signs, and hematology and serum biochemistry variables. In conclusion, once daily administration of robenacoxib tablets had noninferior efficacy and tolerability compared to carprofen for the treatment of the clinical signs of osteoarthritis in dogs. KEY WORDS: canine, carprofen, COX-2 inhibitor, osteoarthritis, robenacoxib. doi: 10.1292/jvms.11-0529; J. Vet. Med. Sci. 74(9): 1121–1131, 2012

Canine osteoarthritis is a common disorder affecting dogs dosages [3, 12]. of all ages, with an incidence as high as 73% in some breeds Robenacoxib is a new coxib NSAID, developed solely [16]. It is characterized by articular cartilage lesions, under- for companion animal use [8]. It is highly selective for the lying bone remodeling with osteophytes, inflammation and cyclo-oxygenase (COX)-2 enzyme in dogs. In whole blood pain [7]. The most frequently involved joints are the stifle, assays, robenacoxib was 129 fold more potent for COX-2 elbow and hip. At present, there are no drugs proven capable relative to COX-1 in vitro, and ex vivo a dosage of 1–2 mg/ of reversing the pathological changes of osteoarthritis, and kg administered orally significantly inhibited COX-2 but therefore the main aim of therapy remains the reduction of had no effect on COX-1 [9]. Moreover, robenacoxib was pain and inflammation with nonsteroidal anti-inflammatory well tolerated and had a high safety index in healthy dogs drugs (NSAID). NSAIDs maintain mobility and quality of [10]. life [7]. In Japan, several NSAIDs are registered for use In a European clinical trial, robenacoxib had noninferior in canine osteoarthritis (carprofen, firocoxib, ketoprofen, efficacy compared to carprofen in dogs with osteoarthritis meloxicam and tepoxalin), but older NSAIDs are associated when administered once daily for 12 weeks [19]. In the with gastrointestinal side-effects. Recently, experimental present study, robenacoxib was administered once daily for investigations in healthy Beagle and mixed breed dogs have 28 days, and was compared to carprofen as a positive con- confirmed the propensity of carprofen, etodolac, flunixin, trol in a noninferiority clinical trial in Japan. The objective ketoprofen and meloxicam to cause lesions of the upper of this study was to evaluate the efficacy and tolerability of gastrointestinal tract, when administered at recommended a tablet formulation of robenacoxib for the treatment of the clinical signs of osteoarthritis in dogs in Japan. Since breed distribution and management of dogs in Japan may differ *Correspondence to: Edamura, K., Laboratory of Veterinary Sur- from Europe, the Japanese regulatory authorities require gery, Department of Veterinary Medicine, College of Bioresource Sciences, Nihon University, 1866 Kameino, Fujisawa, Kanagawa that new medicines are evaluated in local field studies. Rep- 252–0880, Japan. etition of field studies in different geographical locations e-mail: [email protected] also provides more confidence in the results. ©2012 The Japanese Society of Veterinary Science 1122 K. EDAMURA ET AL.

Table 1. Trial schedule Examination Visits to veterinarian Dog at home Visit number V1 V2 VFinal Day (range) 0 14 (12-16) 28 (24-32) 1-28 Animal history X Owner consent and interview X Inclusion/exclusion criteria X Randomization/allocation to treatment group X Physical examination X X-ray X Administration of treatment X X Clinician investigator assessment of efficacy X X X Owner assessment of efficacy X X X Owner notification of adverse eventsa) X X Clinician investigator assessment of adverse eventsb) X X X Blood sample for serum biochemistry and hematology X X X Control of owner compliance X X a) The owner was required to report as soon as possible all adverse events to one of the clinician investigators. The owner was also asked specifically at each visit if adverse events had been observed. b) The clinician investigator performed a physical examination of the animal and asked the owner if any adverse events had been observed.

MATERIALS AND METHODS fected assessment of response to treatment; dogs which had received prior to inclusion local or systemic NSAIDs within Animals and study design: The study was a prospective, 14 days, corticosteroids within 30 days (systemically) or 90 randomized, parallel-group design noninferiority clinical days (intra-articularly); and dogs known not to be available trial with 2 treatment groups, robenacoxib and carprofen. It for the full duration of the study. was multi-centered, involving 17 veterinary practices from Randomization and blinding procedures: After selection several geographic locations in Japan, and was conducted for the trial, dogs were allocated randomly to one of the in compliance with Good Clinical Practice for Veterinary treatment groups in a 2:1 (robenacoxib:carprofen) ratio us- Pharmaceutical Products (Ordinance No. 75, Ministry of ing randomization lists prepared using a computer random- Agriculture, Forestry and Fisheries, Japan, 1997). The study number generator for each trial site. Because robenacoxib protocol was approved by the Japanese regulatory authori- and carprofen tablets have a different appearance, masking ties and internal Novartis reviews taking into account scien- of clinician assessments was achieved by the “blinding by tific, ethical, and animal welfare guidelines. All dog owners function” method, using 2 investigators. One (the clinician) provided written consent for their animal to participate in was responsible for clinical assessments, and the 2nd (the this study. The schedule is provided in Table 1. dispenser) was responsible for dosing on first administration, Inclusion and exclusion criteria: A detailed investigation dispensing tablets to the owners and control of treatment of each dog was performed at baseline on day 0 by the cli- compliance. Owners were not formally blinded. nician investigator. Inclusion criteria were dogs: ≥6 weeks Drugs and dosing procedures: Dogs received orally of age, body weight 2.5 to 80 kg, of both sexes and any once daily either robenacoxib or carprofen. The dosage of breed, presenting with both clinical signs and radiographic robenacoxib was the registered dosage of 1–2 mg/kg. The evidence (e.g. articular surface erosion, presence of osteo- dosing scheme was one tablet containing 5, 10, 20 and 40 mg phytes) of osteoarthritis; lameness and/or pain in affected robenacoxib (Onsior tablets, Novartis Animal Health K.K., joints present on palpation. If several joints were affected by Tokyo, Japan), respectively for dogs weighing 2.5–5.0, 5.0– osteoarthritis, the most severely affected joint was selected 10.0, 10.0–20.0 and 20.0–40.0 kg, and two 40 mg tablets for for efficacy assessment. The diagnosis was based on history, dogs weighing 40.0–80.0 kg. For carprofen, the registered general physical examination, orthopaedic examination, mean dosage of 4.4 mg/kg (range 3.5–5 mg/kg) once daily clinical signs and, if required, by additional analyses, includ- was administered, using Rimadyl Chewable Tablets 25, 75 ing radiographs. and 100 mg tablets (Pfizer Japan, Inc., Tokyo, Japan). Car- Exclusion criteria were a primary neurological disorder; profen was chosen as the positive control for this study, as lameness associated with neoplasia or known immunologi- it is the most widely used NSAID licensed for the relief of cal disorder; surgery conducted on any joint during the pre- osteoarthritis pain in Japan. vious 60 days; gross instability of the stifle joint; pregnant Assessment of efficacy: The clinician investigators, who or lactating female dogs or dogs intended for breeding; dogs were masked to the treatment groups, assessed the dog us- with severe concomitant disorders (heart, kidney or liver ing numerical rating scales (NRSs) for 5 indices: standing insufficiency, gastrointestinal disorder) that might have af- posture, lameness at walk, lameness at trot, willingness to ROBENACOXIB IN CANINE OSTEOARTHRITIS 1123

Table 2. Clinician investigator criteria for assessment of efficacy Score Criteria Standing posture [0] Normal stance [1] Slightly abnormal stance (partial weight-bearing of limb, but paw remains firmly in contact with floor) [2] Markedly abnormal stance (partial weight-bearing of limb, with minimal contact between paw and the floor) [3] Severely abnormal stance (no weight-bearing) Lameness at walk [0] No lameness; normal weight-bearing on all strides observed [1] Mild lameness with partial weight-bearing [2] Obvious lameness with partial weight-bearing [3] Marked lameness with no weight-bearing Lameness at trot [0] No lameness; normal weight-bearing on all strides observed [1] Mild lameness with partial weight-bearing [2] Obvious lameness with partial weight-bearing [3] Marked lameness with no weight-bearing Willingness to allow the clinician to lift the limb contralateral to the affected limb [0] Readily accepts contralateral limb elevation, bears full weight on affected limb for more than 30 sec [1] Offers mild resistance to contralateral limb elevation, bears full weight on affected limb for more than 30 sec [2] Offers moderate resistance to contralateral limb elevation and replaces it in less than 30 sec [3] Offers strong resistance to elevation of contralateral limb and replaces it in less than 10 sec [4] Refuses to raise contralateral limb Pain at palpation/mobilization [0] No pain elicited on palpation/mobilization of affected joint [1] Mild pain elicited, e.g. turns head in recognition [2] Moderate pain elicited, e.g. pulls limb away [3] Severe pain elicited, e.g. vocalizes or becomes aggressive or will not allow examiner to palpate/mobilize joint Overall response to treatment (compared to day 0) [0] Excellent: no detectable lameness, animal returned to normal activity [1] Good: marked reduction in lameness, but not completely resolved [2] Fair: only slight reduction in lameness [3] Poor: no improvement or condition worsened. The clinician investigator identified at baseline (day 0) the limb affected most markedly by osteoarthritis. Subsequent assessments were made of this limb on days 14 and 28. The primary endpoint of the study, the global functional disability, was the sum of the scores for standing posture, lameness at walk, lameness at trot, willingness to raise the contralateral limb and pain on palpation/mobilization. raise the contralateral limb and pain on palpation/mobiliza- Tolerability: The clinician investigator examined each tion (Table 2). These scores were recorded on days 0, 14 dog, recording appetite, vigor and fecal consistency on days (range 12–16) and 28 (range 24–32) (Table 1). The primary 0, 14 (range 12–16) and 28 (range 24–32). The clinician in- endpoint was the global functional disability score, which vestigator also collected and recorded data on adverse events was the sum of the individual 5 NRS values. Animal owners, by interviewing the owner on days 14 and 28. who were not blinded to the treatment group, also assessed Clinical chemistry and hematology: Blood samples for the activity, stiffness, lameness, and overall quality of life clinical chemistry and hematology were collected before of their dog on each dosing day using NRSs (Table 3). Both treatment on day 0 and then on days 14 (range 12–16) and clinicians and owners also assessed the overall response to 28 (range 24–32). Serum clinical chemistry variables com- treatment on days 14 and 28 (Tables 2 and 3). All individual prised activities of alanine aminotransferase, aspartate ami- clinician assessments and all owner assessments except for notransferase, alkaline phosphatase and gamma glutamyl primary endpoint were secondary endpoints. transferase, and concentrations of creatinine, total protein Concomitant treatments: Disallowed concomitant treat- and urea. Hematology variables comprised hematocrit, he- ments were all classes of analgesics including opioids and moglobin concentration, red blood cell count, platelet count, other NSAIDs, corticosteroids, sodium pentosan polysul- and total and differential white blood cell counts. phate, glucosamine, chondroitin sulphate and highly protein Statistical analyses: The software SAS, Version 9.1.3. bound drugs. Other concomitant treatments were allowed, (SAS®, Cary, NC, U.S.A.) was used for all calculations. as long as they did not interfere with the objectives of this Data are presented as mean ± standard deviation (S.D.). study. When concomitant treatments were used, details of To determine statistical significance, tests were 2-sided on treated conditions and dosing schedules were recorded. a 5% level (α=0.05). The primary analysis of both efficacy 1124 K. EDAMURA ET AL.

Table 3. Owner criteria for assessment of efficacy Score Criteria Overall quality of life [0] Normal i.e. good [1] Slightly impaired [2] Moderately impaired [3] Severely impaired Level of activity during the day [0] Normal [1] Slightly impaired [2] Moderately impaired [3] Severely impaired Level of stiffness in the morning (after sleeping) [0] None [1] Mild [2] Moderate [3] Severe Degree of lameness during the day [0] None, the dog is normal [1] Mild: 1) does not whimper or cry upon voluntary joint movement, 2) rises from resting position with minimal difficulty, 3) climbs steps or jumps up near normally, 4) mild pain reaction when joint is moved (looking at joint, pulling away of limb) [2] Moderate: 1) occasional whimpering or yelp upon voluntary joint movement, 2) slow to rise from resting position, sitting preferred over standing, 3) reluctant to climb steps or jump up, 4) increased pain reaction when joint is moved (looking at joint, pulling away of limb) [3] Severe: 1) frequent whimpering or yelp upon voluntary joint movement, 2) increased difficulty in rising from resting position, 3) cannot climb steps or jump up, 4) will not allow person to handle joint (biting, growling, pulling away of limb) Overall response to treatment [0] Excellent [1] Good [2] Fair [3] Poor The clinician investigator interviewed the owner on days 14 and 28 to record the owners’ scores. The global owner score was the sum of the scores for activity, stiffness and lameness. and tolerability variables was performed on the per-protocol bution. For summary statistics and non-parametric analyses, (PPR) data set. A secondary analysis was conducted using but not (RM)AN(C)OVA analyses, missing values were the all-randomized animal (ARA) data set. There were no imputed using the last observation carried forward method. relevant differences in conclusions from the analyses of the For efficacy assessments, the initial RMANCOVA models PPR and ARA data sets, therefore only results from the PPR contained the following variables: baseline value, subject analysis are reported. (as a random effect), treatment group, time, treatment group Demographic and baseline data were compared between x time interaction. Covariables (except for treatment group) groups using the Mann-Whitney test for ordinal (e.g. body which gave P>0.10 were deleted from successive iterations weight) or binary data (e.g. sex), the Kruskal-Wallis test of the model. for non-binary nominal data (e.g. combination of sex and Noninferiority comparisons of the efficacy of robena- neutered status), and Fisher’s Exact test for frequency data. coxib compared to carprofen were performed by calculating For efficacy scores, clinical chemistry and hematology estimates and 95% confidence intervals [CI] for the ratio of data, analysis of variance (ANOVA) models was fitted. Co- the reciprocal of efficacy scores (mean of values at V2 and variance models (ANCOVA) were used, if there was at least VFinal, Table 1) in the robenacoxib group/carprofen group one covariate (e.g. baseline pain). Repeated measures mod- from the RMANOVA models. Reciprocals of clinical scores els (RMANOVA or RMANCOVA) were used, if there were were used so that estimates >1 indicate superior efficacy of repeated measurements after baseline, as for most efficacy robenacoxib. A delta (δ) value of 0.25 was selected, as was scores. Responses were in some cases log- or power-trans- used in a recent NSAID clinical trial in children [2], so that formed to produce the best approximation of a normal distri- noninferiority was concluded if the lower boundary of the ROBENACOXIB IN CANINE OSTEOARTHRITIS 1125

Table 4. Demographic variables at baseline Variable Robenacoxib Carprofen P value Number of dogs 21 11 Body weight (kg) 17.8 ± 14.1 19.3 ± 15.4 0.72 Age 10.0 ± 4.4 10.0 ± 4.1 0.92 Breeda) - - 0.18 Sex and neuter status 0.73 ▪ Male intact 6 (29%) 3 (27%) ▪ Female intact 2 (10%) 0 (0%) ▪ Male neutered 6 (29%) 3 (27%) ▪ Female neutered 7 (33%) 5 (45%) Joint most markedly affected by osteoarthritis 0.94 ▪ Elbow 5 (24%) 3 (27%) ▪ Hip 1 (5%) 0 (0%) ▪ Stifle 4 (19%) 3 (27%) ▪ Spine 4 (19%) 2 (18%) ▪ Metacarpal/tarsal 1 (5%) 1 (9%) ▪ Limb (joint not defined) 6 (29%) 2 (18%) Duration of signs (days) 193.4 ± 406.1 379.5 ± 1149.1 0.049 Baseline clinical scores assessed by the clinician investigator ▪ Global functional disability score 8.81 ± 3.60 8.64 ± 3.07 0.95 ▪ Standing posture 1.29 ± 0.85 1.64 ± 0.81 0.26 ▪ Lameness at walk 1.71 ± 0.64 2.0 ± 0.63 0.28 ▪ Lameness at trot 1.95 ± 0.86 1.82 ± 0.75 0.60 ▪ Willingness to raise contralateral limb 2.10 ± 1.0 1.64 ± 0.92 0.29 ▪ Pain at palpation/mobilization 1.76 ± 0.77 1.55 ± 0.82 0.40 Baseline clinical scores assessed by the owner ▪ Global owner score 5.71 ± 1.71 5.73 ± 1.49 0.73 ▪ Activity during the day 2.00 ± 0.71 2.09 ± 0.54 0.74 ▪ Stiffness in the morning 1.82 ± 0.68 1.73 ± 0.65 0.77 ▪ Lameness during the day 1.90 ± 0.70 1.91 ± 0.54 0.96 ▪ Overall quality of life 1.90 ± 0.62 2.00 ± 0.00 0.58 Data are mean ± S.D. or number of dogs (%). P values were calculated with Mann-Whitney, Kruskal- Wallis or Fisher’s Exact Probability tests. a) 16 breeds represented.

95% CI for the ratio was greater than 0.75 (1-δ). the intensity of clinical signs was not significantly different Differences between groups at each time point were between groups at baseline, and scores at baseline were also compared with the Mann-Whitney test. Change from higher in the robenacoxib group for the primary endpoint, baseline in efficacy scores was analyzed using the paired- the global functional disability score (Table 5). The joints af- samples Wilcoxon test. fected included the elbow, hip, metacarpal/tarsal, stifle, spine, and other not defined. A total of 16 breeds were represented RESULTS with no significant difference between groups (P=0.18). The commonest breeds were cross-bred (5), Labrador Retriever Baseline and demographic data: A total of 40 dogs (ARA (4), Golden Retriever (3) and Minature Dachshund (3). All population) were enrolled in this study. A total of 8 dogs other breeds were represented by 1 or 2 dogs only. The mean violated pre-admission exclusion criteria (1 case received ± S.D. dosage of the test articles administered was 1.55 ± NSAIDs within 14 days prior to inclusion, and 7 cases did 0.34 mg/kg robenacoxib and 4.29 ± 0.53 mg/kg carprofen. not have any data post-treatment) and were removed from Clinician investigator efficacy assessment: The clinician the PPR data set. Therefore, a total of 32 dogs were included investigator’s efficacy assessments on days 0 (before treat- in the primary PPR population analysis (21 received robena- ment), 14 and 28 are summarized in Table 5. Values for all coxib, 11 received carprofen). There were no significant dif- assessed variables were similar at all three time points and ferences between groups in demographic and baseline data there were no significant differences between the groups for except for duration of clinical signs which was significantly any endpoint at any of the time points. At the time of the final longer (P=0.049) in the dogs that subsequently received assessment (on day 28), the percentage ± S.D. decrease in carprofen (Table 4). This was judged not to be relevant since mean values compared to baseline for the global functional 1126 K. EDAMURA ET AL.

Table 5. Clinician investigator efficacy scores Response Time Robenacoxib (n=21) Carprofen (n=11) P valuea) Global functional disability score Day 0 8.81 ± 3.60 8.64 ± 3.07 0.95 (Primary endpoint) Day 14 4.67 ± 3.50 5.36 ± 4.92 1.0 % cfbb) −50.5 ± 33.0 −40.0 ± 48.8 P valuec) <0.0001 0.0293 Day 28 2.86 ± 3.07 4.55 ± 5.34 0.57 % cfbb) −70.1 ± 29.9 −51.2 ± 52.9 P valuec) <0.0001 0.0234 Standing posture Day 0 1.29 ± 0.85 1.64 ± 0.81 0.26 Day 14 0.67 ± 0.73 1.00 ± 1.00 0.39 % cfbb) −42.1 ± 42.7 −40.9 ± 43.7 P valuec) 0.0005 0.0313 Day 28 0.52 ± 0.73 1.00 ± 1.00 0.53 % cfbb) −51.6 ± 46.2 −50.0 ± 50.0 P valuec) 0.0005 0.0313 Lameness at walk Day 0 1.71 ± 0.64 2.00 ± 0.63 0.28 Day 14 0.90 ± 0.62 1.18 ± 0.87 0.42 % cfbb) −44.4 ± 35.1 −39.4 ± 43.6 P valuec) 0.0001 0.0352 Day 28 0.62 ± 0.67 0.91 ± 1.04 0.56 % cfbb) −57.9 ± 42.7 −53.0 ± 52.6 P valuec) 0.0001 0.0391 Lameness at trot Day 0 1.95 ± 0.86 1.82 ± 0.75 0.60 Day 14 1.14 ± 0.91 1.18 ± 1.08 0.97 % cfbb) −39.7 ± 39.6 −36.4 ± 50.5 P valuec) 0.0002 0.1094 Day 28 0.71 ± 0.78 0.91 ± 1.14 0.83 % cfbb) −59.5 ± 40.0 −54.5 ± 52.2 P valuec) <0.0001 0.0313 Willingness to raise contralateral limb Day 0 2.10 ± 1.00 1.64 ± 0.92 0.29 Day 14 1.05 ± 0.92 1.09 ± 1.30 0.77 % cfbb) −56.3 ± 34.2 −36.4 ± 63.6 P valuec) <0.0001 0.2188 Day 28 0.57 ± 0.68 1.00 ± 1.34 0.66 % cfbb) −77.8 ± 27.9 −40.9 ± 66.4 P valuec) <0.0001 0.1719 Pain at palpation/mobilization Day 0 1.76 ± 0.77 1.55 ± 0.82 0.40 Day 14 0.90 ± 0.77 0.91 ± 1.04 0.83 % cfbb) −53.2 ± 39.3 −43.9 ± 65.1 P valuec) <0.0001 0.0625 Day 28 0.43 ± 0.68 0.91 ± 1.04 0.18 % cfbb) −77.0 ± 37.4 −43.9 ± 65.1 P valuec) <0.0001 0.1094 Overall response to treatment − clinician Day 14 1.24 ± 0.94 1.55 ± 1.13 0.46 Day 28 0.90 ± 0.94 1.27 ± 1.27 0.51 Data are mean ± S.D. a) Mann-Whitney test. b) % change from baseline (cfb). c) Paired-samples Wilcoxon test. ROBENACOXIB IN CANINE OSTEOARTHRITIS 1127 disability score was 70.1 ± 29.9 for robenacoxib and 51.2 DISCUSSION ± 52.9 for carprofen. There was a good to excellent level of efficacy of similar magnitude in both treatment groups. The principal findings of this study are that the new Differences between days 14 and 28 compared to day 0 were NSAID robenacoxib had noninferior efficacy and toler- highly significant (P<0.001, paired-samples Wilcoxon test) ability compared to carprofen for the treatment of the clini- for all 6 clinician global and NRS scores for robenacoxib cal signs of osteoarthritis in dogs. The relative efficacy of (Table 5). For carprofen, differences at days 14 and 28 com- robenacoxib was better than for carprofen for the primary pared to baseline were significant (P<0.05) for only 3 of 6 efficacy endpoint and all 12 secondary endpoints, although scores (global functional disability score, standing posture differences were not statistically significant. Statistical and lameness at walk). RMANCOVA analysis on the PPR noninferior efficacy of robenacoxib was shown for only 3 data set indicated no significant (P>0.1) effects of treatment secondary endpoints and not for the primary endpoint, due group or treatment group x time interaction. There was a sig- to wide confidence intervals explainable by a combination of nificant time effect (P<0.05), with a decrease in scores with relatively high variability in scores and low sample sizes. For time for all clinician endpoints. Noninferiority of robena- the primary endpoint, the global functional disability score, coxib compared to carprofen was demonstrated for only one the relative efficacy of robenacoxib versus carprofen was clinician endpoint, pain at palpation (lower limit of CI>75%, 1.244 (95% confidence interval 0.555–2.493). Therefore, Table 6). For the primary endpoint (global functional dis- we conclude that the efficacy of robenacoxib was at least as ability score), the estimated efficacy of robenacoxib was good as for carprofen. Noninferior efficacy of robenacoxib 1.244 (95% CI 0.555 − 2.493) relative to that of carprofen. compared to caprofen was demonstrated also in a previous Owner efficacy assessment: For all 5 owner assessed field study conducted in the EU [19]. The major differences global and NRS scores, robenacoxib demonstrated clear between the 2 studies are that the EU study involved treat- improvement at both the 14 and 28 day assessments and dif- ment for 3 months, and included more dogs (188), and these ferences from baseline were highly significant (P<0.0005, had a higher mean body weight (32.1 kg) compared to the Table 7). For carprofen, changes from baseline on both days mean body weight of 17.9 kg in our 40 dogs. The common- 14 and 28 were significant for 3 of the 5 scores (P<0.05). At est breeds in the EU study [19] were Labrador Retriever day 28, mean ± S.D. percentage decreases from baseline of (21%), crossbred (16%) and German Shepherd (12%), com- the global owner score were 60.8 ± 30.8 for robenacoxib and pared to crossbred (16%), Labrador Retriever (13%), Golden 49.8 ± 44.8 for carprofen. Differences between group means Retriever (9%) and Minature Dachshund (9%) in the present were not statistically significant. Using RMANCOVA, non- study. The finding of similar conclusions in the 2 indepen- inferior efficacy of robenacoxib compared to carprofen was dent studies provides higher confidence in the results. demonstrated for the global owner score and the owner’s In contrast to humans, animals cannot articulate their assessment of stiffness (Table 8). experience of pain, so that evaluating animal pain presents Tolerability: There were no reported adverse events in the significant challenges. Holton et al. [4, 5] concluded that the carprofen group and 3 benign adverse events in the robe- assessment of pain in dogs is best achieved by observing and nacoxib group, comprising single instances of diarrhoea (2 recording changes in behavior, locomotion and demeanor. dogs) and vomiting (one dog). Differences were not signifi- For these and other observations, NRSs were used in the cant (P=0.54). present study. They were preferred to simple descriptive Clinical chemistry and hematology: Summary data of scales which lack of sensitivity and visual analogue scales clinical chemistry and hematology variables with non- for which high inter-observer variability is common [5]. parametric statistical analyses are shown in Tables 9 and 10. Therefore, although the estimation of pain was subjective, There were no significant changes from baseline for any vari- the assessments used in the present study are considered to able using either ANCOVA or the Wilcoxon test. There were have provided good indications of efficacy, especially as no significant differences between groups for any variable several endpoints were used. using RMANOVA except for alkaline phosphatase during Carprofen was selected as the positive control drug for treatment which showed a significantly (P=0.016, data not this study for several reasons. First, carprofen is the most shown) higher increase from baseline with carprofen com- widely used NSAID in Japan for treating osteoarthritis in the pared to robenacoxib. There were no significant differences dog. Second, published data on carprofen are extensive; its between groups at days 14 and 28 using the Mann-Whitney efficacy has been well established in previous clinical trials test (Tables 9 and 10). Serum creatinine concentrations were in comparison with other NSAIDs, placebo or no alternative higher (P=0.013) at baseline in the dogs randomized to the treatment [6, 11, 13–15, 18, 22]. Superior efficacy of carpro- robenacoxib group, but there were no significant differences fen compared to a placebo has been demonstrated in dogs between groups at days 14 or 28 (Table 9). with the urate synovitis model of arthritis [1] and in dogs with Concomitant treatments: There were no significant dif- spontaneous osteoarthritis [6, 15, 22]. Thirdly, carprofen had ferences between groups for numbers or drug classes of the best gastrointestinal tolerability of 5 NSAIDs (carprofen, concomitant treatments. No adverse interactions between etodolac, flunixin, ketoprofen and meloxicam) in a compara- concomitant treatments and the test articles were observed. tive study when administered daily for 6 weeks [12]. Previ- ously, Forsyth et al. concluded that carprofen produced less gastroduodenal lesions than ketoprofen or meloxicam [3]. In 1128 K. EDAMURA ET AL.

Table 6. Noninferiority analysis using RMANCOVA for clinician investigator efficacy scores Response Quotient 95% Confidence Interval P value Global functional disability score (Primary endpoint) 1.244 0.555 − 2.493 0.25 Standing posture 1.102 0.463 − 2.248 0.78 Lameness at walk 1.270 0.561 − 2.564 0.49 Lameness at trot 1.171 0.594 − 2.143 0.60 Willingness to raise contralateral limb 1.542 0.645 − 3.521 0.25 Pain at palpation/mobilization 1.657 0.830 − 3.652 0.14 Overall response 1.234 0.527 − 2.514 0.55 Estimates are reported as reciprocals of scores (robenacoxib:carprofen) therefore values >1.00 indicate higher efficacy for robenacoxib compared to carprofen. Noninferiority demonstrated if lower limit of 95% confidence interval >0.75 (shown in bold).

Table 7. Owner efficacy scores Response Time Robenacoxib (n=21) Carprofen (n=11) P valuea) Global owner score Day 0 5.71 ± 1.71 5.73 ± 1.49 0.73 Day 14 3.19 ± 2.38 4.09 ± 2.77 0.31 % cfbb) −45.2 ± 33.5 −33.0 ± 39.5 P valuec) <0.0001 0.0313 Day 28 2.33 ± 2.20 3.18 ± 3.28 0.60 % cfbb) −60.8 ± 30.8 −49.8 ± 44.8 P valuec) <0.0001 0.0156 Activity during the day Day 0 2.00 ± 0.71 2.09 ± 0.54 0.74 Day 14 1.19 ± 0.98 1.55 ± 1.04 0.35 % cfbb −43.7 ± 41.7 −31.8 ± 40.5 P valuec) 0.0002 0.0625 Day 28 0.90 ± 0.94 1.18 ± 1.08 0.48 % cfbb) −56.3 ± 41.7 −50.0 ± 38.7 P valuec) <0.0001 0.0078 Stiffness in the morning Day 0 1.81 ± 0.68 1.73 ± 0.65 0.77 Day 14 1.00 ± 0.77 1.27 ± 1.01 0.42 % cfbb) −44.4 ± 38.5 −31.8 ± 46.2 P valuec) 0.0001 0.125 Day 28 0.71 ± 0.78 1.00 ± 1.10 0.57 % cfbb) −65.1 ± 35.7 −45.5 ± 52.2 P valuec) 0.0001 0.0547 Lameness during the day Day 0 1.90 ± 0.70 1.91 ± 0.54 0.96 Day 14 1.00 ± 0.84 1.27 ± 0.90 0.41 % cfbb) −46.0 ± 40.5 −36.4 ± 39.3 P valuec) 0.0001 0.0313 Day 28 0.71 ± 0.72 1.00 ± 1.18 0.72 % cfbb) −61.1 ± 39.2 −50.0 ± 54.8 P valuec) <0.0001 0.0391 Overall quality of life Day 0 1.90 ± 0.62 2.00 ± 0.00 0.58 Day 14 1.00 ± 0.77 1.09 ± 0.83 0.77 % cfbb) −50.0 ± 37.6 −45.5 ± 41.6 P valuec) <0.0001 0.0156 Day 28 0.57 ± 0.75 0.82 ± 0.87 0.44 % cfbb) −73.0 ± 35.1 −59.1 ± 43.7 P valuec) <0.0001 0.0078 Overall response to treatment - owner Day 14 1.24 ± 0.83 1.55 ± 1.13 0.49 Day 28 1.05 ± 0.92 1.27 ± 1.27 0.79 Data are mean ± S.D. a) Mann-Whitney test. b) % change from baseline (cfb). c) Paired-samples Wilcoxon test. ROBENACOXIB IN CANINE OSTEOARTHRITIS 1129

Table 8. Noninferiority analysis using RMANCOVA for owner efficacy scores Response Quotient 95% Confidence Interval P value Global owner score 1.308 0.784 – 2.146 0.27 Activity during the day 1.219 0.704 – 2.030 0.43 Stiffness in the morning 1.408 0.832 – 2.387 0.18 Lameness during the day 1.290 0.688 – 2.343 0.38 Overall quality of life 1.093 0.532 – 2.011 0.77 Overall response to treatment - owner 1.206 0.649 – 2.128 0.50 Estimates are reported as reciprocals of scores (robenacoxib vs carprofen) therefore values >1.00 indicate higher efficacy for robenacoxib compared to carprofen. Noninferiority demonstrated if lower limit of 95% confidence interval >0.75 (shown in bold).

Table 9. Serum clinical chemistry variables Response Time Robenacoxib (n=21) Carprofen (n=11) P valuea) Alanine aminotransferase [IU/l] Day 0 105.2 ± 186.2 53.8 ± 74.4 0.10 Day 14 71.0 ± 76.5 43.2 ± 53.7 0.14 P valueb) 0.68 0.84 Day 28 87.8 ± 92.6 44.4 ± 58.3 0.065 P valueb) 0.20 0.84 Aspartate aminotransferase [IU/l] Day 0 34.0 ± 16.0 43.0 ± 34.3 0.97 Day 14 32.9 ± 14.3 28.4 ± 13.8 0.15 P valueb) 0.57 0.055 Day 28 34.1 ± 14.5 33.3 ± 13.8 0.80 P valueb) 0.93 1 Alkaline phosphatase [IU/l] Day 0 298.1 ± 623.1 129.2 ± 178.0 0.065 Day 14 291.1 ± 671.4 133.7 ± 223.2 0.19 P valueb) 0.082 0.29 Day 28 302.9 ± 704.4 153.4 ± 276.4 0.23 P valueb) 0.63 1 Gamma-glutamyl transferase [IU/l] Day 0 11.1 ± 13.9 11.1 ± 9.5 0.95 Day 14 9.4 ± 14.4 13.5 ± 20.4 0.87 P valueb) 0.67 0.72 Day 28 10.4 ± 16.4 10.9 ± 12.4 0.98 P valueb) 0.17 0.84 Total protein [g/dl] Day 0 6.11 ± 0.73 5.75 ± 0.53 0.25 Day 14 6.0 ± 0.54 5.64 ± 0.57 0.13 P valueb) 0.35 0.38 Day 28 5.86 ± 0.55 5.66 ± 0.61 0.47 P valueb) 0.90 1 Urea [mg/dl] Day 0 19.7 ± 9.2 15.0 ± 9.1 0.10 Day 14 20.3 ± 8.3 17.4 ± 12.6 0.099 P valueb) 0.23 0.37 Day 28 18.7 ± 6.9 19.7 ± 22.0 0.056 P valueb) 0.80 0.66 Creatinine [mg/dl] Day 0 0.88 ± 0.22 0.67 ± 0.20 0.013 Day 14 0.93 ± 0.34 0.82 ± 0.27 0.31 P valueb) 0.74 0.066 Day 28 0.90 ± 0.36 0.80 ± 0.33 0.36 P valueb) 0.59 0.34 Data are mean ± S.D. a) Mann-Whitney test. b) Paired-samples Wilcoxon test. 1130 K. EDAMURA ET AL.

Table 10. Hematology variables Response Time Robenacoxib (n=21) Carprofen (n=11) P valuea) Red blood cell count [1012/l] Day 0 6.89 ± 1.03 6.95 ± 1.19 0.71 Day 14 6.94 ± 1.16 7.00 ± 1.02 0.87 P valueb) 0.43 0.64 Day 28 6.80 ± 1.07 6.95 ± 1.00 0.78 P valueb) 0.71 0.88 Hemoglobin [g/dl] Day 0 16.0 ± 2.0 16.1 ± 2.7 0.66 Day 14 16.1 ± 2.5 16.2 ± 2.1 0.92 P valueb) 0.67 0.73 Day 28 15.9 ± 2.3 16.2 ± 2.0 0.78 P valueb) 0.87 0.95 Hematocrit [%] Day 0 50.2 ± 6.2 49.4 ± 7.7 0.98 Day 14 50.6 ± 7.8 50.0 ± 6.3 0.61 P valueb) 0.43 0.97 Day 28 49.7 ± 7.1 50.4 ± 6.4 0.98 P valueb) 0.86 0.47 White blood cell count Day 0 9.02 ± 2.85 8.87 ± 3.02 1.0 9 [10 /l] Day 14 9.01 ± 3.93 9.27 ± 4.28 0.77 P valueb) 0.53 0.88 Day 28 9.12 ± 3.93 11.35 ± 8.81 0.97 P valueb) 0.86 0.52 Data are mean ± S.D. a) Mann-Whitney test. b) Paired-samples Wilcoxon test. light of these published findings, carprofen may be regarded and up to 10 mg/kg daily for 1 month [10]. as the preferred NSAID against which to assess a new agent Strengths of the present study include the fact that it fol- such as robenacoxib for canine osteoarthritis therapy from lowed the gold-standard design of being multi-centered, both efficacy and safety perspectives. Carprofen is prefer- prospective and randomized. The principal limitations of the entially selective for COX-2 in dogs (COX2/COX1 potency study are related to non-blinding of owner assessments, the ratio of S+ enantiomer is 18), while robenacoxib is highly relatively small group sizes and inherent weaknesses of the selective (COX2/COX1 potency ratio is 129) [9]. Of clini- non-inferiority design [17]. Clinician investigator assess- cal relevance is the fact that COX-1 sparing properties have ments were masked via the use of the “blinding by func- been confirmed in dogs with chronic osteoarthritis with car- tion” technique. Inclusion of an additional placebo group, to profen [21] and in healthy Beagle dogs with robenacoxib [9]. provided internal validity, was judged unnecessary, since the The tolerability of both treatments was good as assessed efficacy of carprofen, the positive control, has been proven from adverse events, clinical signs, serum chemistry and he- to be superior to placebo both in the challenging urate crystal matology variables. The rationale behind the development of synovitis model [1] and in clinical dogs with osteoarthritis the coxib NSAIDs is that they should cause less damage to [6, 15, 22]. Furthermore, both Holtsinger et al. and Vasseur the gastrointestinal tract and less inhibition of blood clotting et al. [6, 22] established the efficacy of carprofen compared compared to non-selective NSAIDs [23]. Lumiracoxib, an to placebo using NRSs similar to those used in the pres- analogue of robenacoxib, was proven to produce less fre- ent study. In optimally designed noninferiority studies, the quent serious gastrointestinal tract adverse events (bleeding, methods and outcome measures should be similar to those perforation or obstruction) compared to the non-selective used in the original studies of the active control [17]. Fur- NSAIDs ibuprofen and naproxen in humans [20]. The clini- thermore, inclusion of a placebo group would have caused cal trial in dogs described in this study was not powered to ethical concerns, since several NSAIDs are registered and test for superior tolerability or safety; notably the number of widely used to treat canine osteoarthritis in Japan. dogs was too small, the treatment time was short (1 month) In conclusion, once daily administration of robenacoxib and the safety variables measured were not sensitive. A tablets had noninferior efficacy and tolerability compared to dedicated study with larger numbers of dogs and/or specific carprofen for the treatment of the clinical signs of osteoar- assessments of the gastrointestinal tract are needed to test thritis in dogs. for potential superior safety. Robenacoxib has a good safety index in healthy dogs, producing no biologically relevant ACKNOWLEDGMENTS. We thank Drs Shinji Abe, Jun toxicity at dosages as high as 40 mg/kg daily for 1 month Ando, Masaya Hikosaka, Koji Hosomizo, Tatsuo Kaneshige, ROBENACOXIB IN CANINE OSTEOARTHRITIS 1131

Masako Kawamata, Hirokazu Kimura, Toru Kurita, Ya- Moutinho, F. Q., Takahira, R. M. and Brandao, C. V. S. 2007. sushi Matsumura, Noriyuki Minegishi, Masaaki Mochizuki, Evaluation of adverse effects of long-term oral administration Toshiaki Mogi, Hideyuki Nagae, Shigeru Nakamura, Shin- of carprofen, etodolac, flunixin, meglumine, ketoprofen and ichi Namba, Yuzuru Sakai and Masahiro Takahashi for meloxicam in dogs. Am. J. Vet. Res. 68: 258–264. [Medline] [CrossRef] seeing clinical cases and Professor Peter Lees for assistance 13. Mansa, S., Palmer, E., Grondahl, C., Lonaas, L. and Nyman, with the manuscript. G. 2007. Long-term treatment with carprofen of 805 dogs with osteoarthritis. Vet. Rec. 160: 427–430. [Medline] [CrossRef] REFERENCES 14. McCarthy, G., O’Donovan, J., Jones, B., McAllister, H., Seed, M. and Mooney, C. 2007. Randomised double-blind, positive- 1. Borer, L. R., Peel, J. E., Seewald, W., Schwalder, P. and Spreng, controlled trial to assess the efficacy of glucosamine/chondroitin D. E. 2003. Effect of carprofen, etodolac, meloxicam, or butor- sulphate for the treatment of dogs with osteoarthritis. Vet. J. 174: phanol in dogs with induced acute synovitis. Am. J. Vet. Res. 64: 54–61. [Medline] [CrossRef] 1429–1437. [Medline] [CrossRef] 15. Moreau, M., Dupuis, J., Bonneau, N. H. and Desnoyers, M. 2. Foeldvari, I., Szer, I. S., Zemel, L. S., Lovell, D. J., Giannini, 2003. Clinical evaluation of a nutraceutical, carprofen and E. H., Robbins, J. L., West, C. R., Steidle, G., Krishnaswami, S. meloxicam for the treatment of dogs with osteoarthritis. Vet. Rec. and Bloom, B. J. 2009. A prospective study comparing celecoxib 152: 323–329. [Medline] [CrossRef] with naproxen in children with juvenile rheumatoid arthritis. J. 16. Paster, E. R., LaFond, E., Biery, D. N., Iriye, A., Gregor, T. P., Rheumatol. 36: 174–182. [Medline] Shofer, F. S. and Smith, G. K. 2005. Estimates of prevalence 3. Forsyth, S. F., Guilford, W. G. and Lawoko, C. R. O. 1996. of hip dysplasia in Golden Retrievers and Rottweilers and the Endoscopic evaluation of the gastroduodenal mucosa following influence of bias on published prevalence figures. J. Am. Vet. non-steroidal anti-inflammatory drug administration in the dog. Med. Assoc. 226: 387–392. [Medline] [CrossRef] N. Z. Vet. J. 44: 179–181. [Medline] [CrossRef] 17. Piaggio, G., Elbourne, D. R., Altman, D. G., Pocock, S. J., Evans 4. Holton, L. L., Scott, E. M., Nolan, A. M., Reid, J., Welsh, E. S. J. W., CONSORT group 2006. Reporting of noninferiority and and Flaherty, D. 1998. Comparison of three methods used for equivalence randomized trials: an extension of the CONSORT assessment of pain in dogs. J. Am. Vet. Med. Assoc. 212: 61–66. statement. JAMA 295: 1152–1160. [Medline] [CrossRef] [Medline] 18. Pollmeier, M., Toulemonde, C., Fleishman, C. and Hanson, P. 5. Holton, L., Reid, J., Scott, E. M., Pawson, P. and Nolan, A. 2001. D. 2006. Clinical evaluation of firocoxib and carprofen for the Development of a behaviour-based scale to measure acute pain treatment of dogs with osteoarthritis. Vet. Rec. 159: 547–551. in dogs. Vet. Rec. 148: 525–531. [Medline] [CrossRef] [Medline] [CrossRef] 6. Holtsinger, R. H., Parker, R. B., Beale, B. S. and Friedman, R.L. 19. Reymond, N., Speranza, C., Gruet, P., Seewald, W. and King J. TM 1992. The therapeutic efficacy of Carprofen (Rimadyl-V ) N. 2011. Robenacoxib vs. carprofen for the treatment of canine in 209 clinical cases of canine degenerative joint disease. Vet. osteoarthritis; a randomized, noninferiority clinical trial. J. Vet. Comp. Orthop. Traumatol. 5: 140–144. Pharmacol. Ther. published on line 12 Apr., 2011. 7. Johnston, S. A., McLaughlin, R. M. and Budsberg, S. C. 2008. 20. Schnitzer, T. J., Burmester, G. R., Mysler, E., Hochberg, M. C., Nonsurgical management of osteoarthritis in dogs. Vet. Clin. Doherty, M., Ehrsam, E., Gitton, X., Krammer, G., Mellein, B., North Am. Small Anim. Pract. 38: 1449–1470. [Medline] Matchaba, P., Gimona, A. and Hawkey, C. J. 2004. Comparison [CrossRef] of lumiracoxib with naproxen and ibuprofen in the Therapeutic 8. King, J. N., Dawson, J., Esser, R. E., Fujimoto, R., Kimble, Arthritis Research and Gastrointestinal Event Trial (TARGET), E. F., Maniara, W., Marshall, P. J., O’Byrne, L., Quadros, E., reduction in ulcer complications: randomised controlled trial. Toutain, P. L. and Lees, P. 2009. Preclinical pharmacology of Lancet 364: 665–674. [Medline] [CrossRef] robenacoxib: a novel selective inhibitor of cyclooxygenase-2. J. 21. Sessions, J. K., Reynolds, L. R. and Budsberg, S. C. 2005. In Vet. Pharmacol. Ther. 32: 1–17. [Medline] [CrossRef] vivo effects of carprofen, deracoxib and etodolac on prostanoid 9. King, J. N., Rudaz, C., Borer, L., Jung, M., Seewald, W. and production in blood, gastric mucosa and synovial fluid in dogs Lees, P. 2010. In vitro and ex vivo inhibition of canine cyclooxy- with chronic osteoarthritis. Am. J. Vet. Res. 66: 812–817. [Med- genase isoforms by robenacoxib: a comparative study. Res. Vet. line] [CrossRef] Sci. 88: 497–506. [Medline] [CrossRef] 22. Vasseur, P. B., Johnson, A. L., Budsberg, S. C., Lincoln, J. D., 10. King, J. N., Arnaud, J. P., Goldenthal, E.I., Gruet, P., Jung, M., Toombs, J. P., Whitehair, J. G. and Lentz, E. L. 1995. Random- Seewald, W. and Lees, P. 2011. Robenacoxib in the dog: target ized, controlled trial of the efficacy of carprofen, a non-steroidal species safety in relation to extent and duration of inhibition anti-inflammatory drug, in the treatment of osteoarthritis in of COX-1 and COX-2. J. Vet. Pharmacol. Ther. 34: 298–311. dogs. J. Am. Vet. Med. Assoc. 206: 807–811. [Medline] [CrossRef] 23. Warner, T. D., Giuliano, F., Vojnovic, I., Bukasa, A., Mitchell, J. 11. Lipscomb, V. J., AliAbadi, F. S., Lees, P., Pead, M. J. and Muir, A. and Vane, J. R. 1999. Nonsteroid drug selectivities for cyclo- P. 2002. Clinical efficacy and pharmacokinetics of carprofen in oxygenase-1 rather than cyclo-oxygenase-2 are associated with the treatment of dogs with osteoarthritis. Vet. Rec. 150: 684–689. human gastrointestinal toxicity. A full in vitro analysis. Proc. [Medline] [CrossRef] Natl. Acad. Sci. U.S.A. 96: 7563–7568. [Medline] [CrossRef] 12. Luna, S. P. L., Basilio, A. C., Steagall, P. V. M., Machado, L. P.,