In the present study, we have analyzed four key clinical trial outcome measures (client-owned questionnaire, activity monitoring, measurement of ground reaction forces, and use of rescue treatment) in a single randomized controlled trial evaluating analgesic treatment of naturally occurring OA in dogs. Previous trial studies have often compared a NSAID, such as carprofen, with a placebo arm using a more limited range of outcome measures. Limited data are available directly comparing treatment effects of several different types of oral analgesics within a single trial. Placebo effects were identified for all outcome measures studied. All of the trial medications had relatively small treatment effects on outcome measures in dogs with moderate to severe hip arthritis, suggesting that development of new analgesic medications with greater clinical efficacy is needed. However, a relatively short treatment period was used in the present study. Additionally, we did not use a crossover study design. This would have minimized background variance, but increased the risk of subject dropout from a clinical trial with four treatment arms. Over time, incremental refinement in trial design will likely aid robust examination of treatment effects in client-owned dogs with arthritis.
The signalment and clinical signs of the dogs enrolled in the trial were typical for client-owned dogs with hip OA
[14, 18]. Many canine OA treatment trials have studied patients affected with a range of different arthritis conditions. In the present study, we used more restrictive selection criteria to minimize variation in the study population. There is evidence to suggest that OA of joints other than the hip in dogs, such as the elbow, may be more refractory to medical treatment
. Additionally, stifle OA in dogs is often associated with joint instability from cruciate rupture, which our clinical experience suggests is also more refractory to medical treatment. OA commonly affects multiple joints in the dog
, and many dogs in the present study had multiple joints affected with OA, including degenerative changes to the lumbosacral spine. Therefore, some dogs may have had a neuropathic pain component to their clinical signs, although degenerative changes to the lumbosacral spine seen radiographically are not highly correlated with clinical signs
In general, owner compliance with the trial was good. After group assignment, there were no significant differences in dog mobility between groups. Withdrawal from the trial after enrollment only occurred because of the presence of a complicating major health problem that was not attributable to trial medication, or because of a dosing error in one dog. After two weeks of treatment with ABT-116, plasma concentrations were significantly increased, suggesting that a loading phenomenon exists with oral dosing, with increased bioavailability from chronic dosing. In contrast, plasma concentrations of tramadol were significantly decreased after two weeks of treatment. This likely reflects altered drug metabolism, which is a complex phenomenon
. Despite low plasma concentrations of tramadol and its active metabolite after oral dosing, analgesic efficacy is still detectable experimentally
. The antinociceptive action of tramadol in the dog is not fully understood and may involve effects on α2-adrenoceptors, as well as inhibition of norepinephrine reuptake and μ-receptor agonism
. We felt the inclusion of a tramadol treatment group in this trial was important in order to provide data for future non-NSAID drug candidate testing.
Dosing with ABT-116 was associated with an acute elevation in rectal temperature. This is an established phenomenon associated with administration of TRPV1 antagonists in animal models
[23–26], and is a biomarker of dosing efficacy. Attenuation of this side effect after continued treatment has been previously reported
, and was also apparent in the ABT-116 group at recheck on Day 22. The hyperthermia associated with TRPV1 antagonists is related to their potency for blockade of receptor activation by protons
. We did not detect any clinical signs associated with this temporary side effect in the dogs of the present study. In human beings, there is individual variation in the degree of hyperthermia that develops with treatment, and there was also some variation in the small cohort of dogs in the present study. We also detected a small but significant acute decrease in rectal temperature after carprofen administration, although the biological significance of this observation is unclear.
Goniometry measurement indicated that hip ROM in the ABT-116 and Tramadol groups was decreased, when compared with the Placebo group. In the ABT-116 and Tramadol groups, thigh muscle mass in the worst affected limb was also decreased relative to the contralateral pelvic limb. However, there were no significant differences in overall mobility between groups based on clinical severity scoring at trial entry and no specific treatment effects on either hip ROM and thigh muscle mass. Improvement in these parameters would likely only occur over a treatment period longer than 2 weeks. Clinical response to long-term OA treatment with oral analgesics, such as carprofen, can continue for at least 12 weeks
. In the Placebo group, effect sizes for placebo treatment on hip ROM and thigh muscle mass were small (≤ 0.2).
The CBPI questionnaire is a validated reliable outcome measure for studies of OA in dogs
. In the present study, improvements in CBPI scores were seen in all treatment groups, including the Placebo group. The greatest improvement in CBPI score was seen in the Carprofen and Tramadol groups; significant improvements in both pain severity and pain interference scores were only seen in the Carprofen group. However, these improvements were not large enough to yield significant differences between groups. ES in the ABT-116 group was similar to the Placebo group. The magnitude of improvement seen with carprofen and placebo treatment was similar to other treatment studies using this outcome measure
. Interestingly, with tramadol treatment, improvement in pain interference was much greater than improvement in pain severity. Clinical improvements after tramadol were most evident in rising from lying down or ability to climb. This is interesting, particularly as plasma concentrations were decreased at Day 22, relative to Day 8. Additional studies are needed to assess in more detail the mode of action, metabolism, and clinical efficacy of tramadol in the dog during chronic oral dosing, as only sparse pharmacokinetic data are available to date
ES of placebo treatment on CBPI scores was modest and did not represent a significant change. This effect likely represents a combination of caregiver placebo effect and the regression to the mean phenomenon
. Regression to the mean is a statistical phenomenon that occurs in repeated-measures studies. Because outcome variables are observed with random error around a true mean, relatively high or low observations are likely to be followed by less extreme ones nearer to the true mean
. In our study, regression to the mean was reduced by use of two baseline measurements and random allocation of subjects to treatment groups. We did not provide training on recognition of clinical signs of OA pain in dogs for participating owners. Poor face validity may limit the value of the questionnaire instrument to some extent
. This problem could be overcome in future work by altering the study design and training the owners by administering and then withdrawing a positive control such as carprofen
. This concern was also emphasized by the pattern of responses identified during correspondence analysis of the CPBI data. The pattern of responses in the four treatment arms of the study was fairly similar, suggesting that none of the treatments used had a major effect on owner perception of patient mobility. In addition, when the pattern of responses to individual questions was examined, clustering of divergent response options was seen, suggesting that owners were not able to consistently pain score the subject dogs within the context of the present study design.
Activity monitoring using an accelerometer is a reliable outcome measure for OA treatment studies in client-owned dogs
. However, in the present study we found that dog mobility was highly variable between individual dogs, with approximately a 5-fold difference or more between the least active dog and the most active dog in each treatment group. Treatment effects did not yield significant between-group differences in overall daily activity. The ABT-116 group was the only group in which dogs exhibited a significant increase in activity after treatment. This increase was identified during the nighttime period. In contrast, dogs that received carprofen treatment were significantly less active at night, but had higher activity in the daytime. Since client-owned dogs usually mirror the habitual activity of their owners, it would be expected that most dogs are more active in the daytime and less active at nighttime. Whether nighttime increases in activity represent a beneficial treatment effect is unclear. It may be that higher nighttime activity is due to increased discomfort, such that the dog is not able to get comfortable and sleep soundly. However, ABT-116 dogs also had improved CBPI scores after treatment and no decrease in daytime activity, so this increase in nighttime activity may simply reflect improved overall mobility. Further work is needed to more comprehensively understand treatment-induced changes in habitual activity in dogs with OA, particularly with regard to sleeping at nighttime. Overall, ES of treatment on daytime and total activity were small, and the large variation in habitual activity between individual dogs is a major factor contributing to small treatment effects in this study, probably acting to increase the magnitude of regression to the mean.
In general, larger Fz and VI values at both the walk and trot are associated with improved mobility in dogs
[33–35]. Gait analysis at the trot provides a more sensitive indicator of lameness than analysis at the walk, particularly for mild lameness
. However, analysis at the walk may be better suited for dogs with more severe lameness, such as cranial cruciate ligament rupture
[33, 36]. In the present study, significant treatment effects on ground reaction forces were not identified, principally because of extensive variation between individual dogs. Our power analysis suggests that much larger sample sizes would be needed to detect significant drug treatment effects. We found that ES associated with treatment were greatest in the ABT-116 and Carprofen groups with improvement in kinetic gait analysis values, whereas ES associated with tramadol and placebo treatment were smaller and not generally associated with improvement in mobility. The placebo effect in force-plate gait analysis has been shown in other studies
[14, 34], and is again likely a consequence of the regression to the mean phenomenon.
In general, outcome measures, including hip ROM, thigh muscle mass, total CBPI, total daily activity, and kinetic gait analysis parameters at the end of the trial were significantly correlated with baseline measurements. This indicates reasonable measurement precision error over time. Although a two-week treatment period has often been used for this type of trial
[14, 19, 29, 32], longer treatment periods
[28, 37] may have led to larger improvements in dog mobility.
Analgesic treatment had significant effects on use of rescue medication by owners during the treatment period. This effect was most evident for ABT-116 and least evident for tramadol. One possible explanation for this finding in the tramadol group is that owners may have interpreted sedative effects from tramadol treatment as a sign of pain. This reinforces the idea that face validity of CBPI scoring may be improved by development of an a priori training protocol for owners as part of trial study design.
Thigh muscle mass, but not hip ROM, was also inversely correlated with dog age and clinical severity at trial entry, suggesting that aging and more severe impairment of mobility are associated with greater disuse atrophy of muscle in affected limbs. Similarly, total CBPI score at the end of the trial was also correlated with clinical severity at trial entry. This observation suggests the CBPI questionnaire may have reasonable face validity
. Dog activity at the end of the trial was also significantly and inversely correlated with clinical severity and dog age. Similar correlations were less evident for kinetic gait analysis parameters. Interestingly, use of rescue treatment in Week 3 was not correlated with clinical severity or use of rescue treatment in the baseline week. Whilst the ABT-116 ES for reduction in use of rescue treatment was larger than for tramadol, this difference was not reflected in CBPI questionnaire scoring. These observations suggest that the clinical criteria that owners used to decide whether or not use of rescue treatment was indicated were different from the criteria owners used for CBPI scoring of their dog.
There were several limitations with this study. A strength of this report is the use of client-owned dogs as the study population. However, this is also a limitation, as studies of naturally occurring OA in dogs are associated with factors that may be confounding, such as the presence of concurrent medical conditions, variations in the home environment of each dog, and the potential for owner errors in trial compliance. Randomization of trial participants is an important strategy to reduce bias and minimize the magnitude of regression to the mean
. In order to ensure equivalence of clinical severity in each arm of the study, we adjusted group assignment during recruitment of the last 12 study dogs to ensure equivalent clinical severity in each group. Alternate strategies during recruitment and treatment assignment, such as randomization blocked on clinical severity, could have been used to address this problem.
Measurement of plasma drug concentrations suggested that dosing or plasma sample labeling errors had occurred in a small number of samples. In addition, one dog was withdrawn after trial completion because of a dosing error. Owners were asked to maintain their habitual routine during the clinical trial period. However, some owners commented that they would allow their dog to become more active if they thought there was a clinical response to analgesic treatment. Therefore, owner bias may still influence outcome measures, even in a controlled trial. Our study was also restricted to use of medium to large breed dogs to enable kinetic gait analysis with the OR6-6-1000 force platform. Use of a pressure mat walkway would have allowed multiple footfalls to be collected in one pass, as well as reducing restrictions on the size of participating dogs
. Collection of 5 valid trials is often used during kinetic gait analysis in dogs
[34, 37]. Although trial repetition is an important factor contributing to variance, vertical forces are not significantly affected by trial repetition
. In some dogs in the present trial it was not possible to collect 5 valid trials for each limb, because of the severity of lameness. We analyzed kinetic gait data from the most severely affected pelvic limb. Other approaches could have been considered, such as calculation of a symmetry index
The overall goal of this study was to determine ES of different outcome measures in a client-owned dog model. Many of the outcome measures we assessed had low statistical power, and a larger number of dogs/group would be required for significant results in an efficacy clinical trial. Use of a crossover study design would help to address this concern by better accounting for intra-individual variation between days and weeks of the trial. However, with a study that includes four treatment arms, use of crossover trial with four treatment and three washout periods would likely have reduced client compliance and increased the risk of dropout from the trial. The dosage of ABT-116 used was based on prior research. It is possible that altered dosage and duration of treatment may improve efficacy in treatment of OA pain. Dosage at 3 mg/kg three times daily resulted in effective TRPV1 inhibition based on the development of acute hyperthermia in treated dogs.