Introduction

Osteoarthritis (OA) affects 20% of the canine population and remains a debilitating and costly disease by quality of life impairment and owner expense. In dogs, OA is often secondary to injury or congenital abnormality and consists of overlapping stages of extracellular matrix degradation, chondrocyte proliferation, and chondrocyte and cartilage loss. Regenerative medical therapies are commonly used in humans to palliate OA pain and have also been reported in dogs. Regenerative medicine therapy, as defined by the United States National Institute of Health, is the “process of creating living, functional tissues to repair or replace tissue or organ function lost due to age, disease, damage, or congenital defects.”¹ Examples include stem cell therapy, platelet-rich plasma, prolotherapy and other modalities that seek to influence inflammation, tissue proliferation, and modulation of OA and other disease processes. In the veterinary literature, information supporting tissue regeneration in vivo is currently lacking.

Prolotherapy (PROLO) is a regenerative therapy with “proliferants”—classified as irritants, particulates, osmotics, chemotactics, or biologics—injected into diseased joints, tendons, ligaments, or a para-spinal area with the intent to provoke an inflammatory response and increased proliferation of tissues during repair.^2,3 Injections of concentrated dextrose, phenol, platelet-rich plasma, stem cells, autologous whole blood, sodium morrhuate, and dry needling, have been used for PROLO.⁴ Dextrose PROLO, the most commonly reported agent in human medicine, is hypothesized to cause localized tissue trauma due to osmotic shock that results in inflammation, subsequent production of numerous prostaglandins and growth factors, cellular proliferation, and a reduction in inflammatory interleukins.^5–10 In an in vitro model utilizing preosteoblasts and patellar ligament fibroblasts exposed to varying concentrations of a phenol-dextrose solution, increased collagen production was observed.⁵ In vivo experimental models have shown no changes in ligament maximum load to failure or energy absorption when dextrose was injected into rat Achilles tendons.¹¹ Other studies have found increased cross-sectional area with unchanged laxity in stretch-injured rat medial collateral ligaments and restorative effects on the cartilage matrix in an anterior cruciate ligament transection model in rabbits.^12,13

In human medicine, administration of dextrose PROLO has been reported for treatment of osteoarthritis, knee instability secondary to anterior cruciate ligament rupture, meniscal pathology, tendon injury, and various forms of spinal pain.^14–18 Systematic reviews in the human literature have found varied levels of efficacy for treatment of musculoskeletal conditions with dextrose PROLO and no reported significant adverse effects.^4,19,20 While numerous clinical trials utilizing PROLO for OA in humans exist, controlled, blinded, randomized trials are currently lacking.¹⁹ Two recent controlled studies of human knee OA did find that dextrose PROLO resulted in safe, substantial improvement in specific knee-osteoarthritis quality of life outcome measures such as pain, stiffness, function, and symptom severity in treated patients as compared to saline control injections when evaluated by the Western Ontario McMaster University Osteoarthritis Index.^2,21 Veterinary reports of dextrose PROLO are limited to isolated case reports, conference proceedings, and clinical reviews with no published methodological scientific research.^20,22–24

The purpose of this pilot study was to evaluate the effects of intra-articular 25% dextrose prolotherapy for treatment of naturally-occurring osteoarthritis of the elbow or stifle in dogs and to determine the ideal sample size required for a larger clinical study. Evaluation was by veterinary lameness exam, a previously validated owner pain survey, goniometry, and a pressure sensing walkway (PSW)^a utilized as outcome measures.^25,26 Our hypothesis was that dogs receiving PROLO injections would show improved veterinary lameness scores, better range of motion, and improved weight bearing of the PROLO-treated limb.

Materials and Methods

This was a randomized, double-blind, placebo-controlled clinical prospective trial designed to test the clinical effectiveness of dextrose PROLO in the relief of lameness and pain in dogs with naturally occurring osteoarthritis. This study was approved by the Kansas State University College of Veterinary Medicine Institutional Animal Care and Use Committee. The study population consisted of client-owned dogs presenting to the Veterinary Health Center at Kansas State University for evaluation and treatment of lameness due to osteoarthritis of the knee or elbow. Costs associated with lameness evaluation and treatment were paid by a grant from the American Kennel Club Canine Health Foundation. Fund, and owners received no other financial incentive to participate. Solicitation of patients consisted of electronic communication to referring veterinarians as well as faculty, staff, and students of the Kansas State University College of Veterinary Medicine.

For the purposes of the study, dogs were evaluated at wk 0, 6, and 12. Dogs were randomized to treatment or placebo group assignment prior to study enrollment. In order to be eligible for the study, dogs were required to have a body weight >20 kg, have a history of unilateral lameness as reported by the owner, and have a minimum of 5% decrease in peak vertical force (PVF) of the lame limb compared to the contralateral limb, measured as a percentage of body weight (kg) on a PSW^a. Prior to enrollment, dogs were permitted to be on nonsteroidal anti-inflammatories (NSAIDs), dietary supplements for arthritis, therapeutic diets, or other analgesics except for corticosteroids. Dogs receiving analgesic medications or supplements were enrolled if there had been no changes in these medications for 2 wk prior to or during the 12 wk study period. Exclusion criteria included any dog with a temperament not suited for PSW^a lameness or orthopedic examination, changes to analgesic medications within 2 wk of study enrollment or during the study, orthopedic surgery of any limb within 6 mo prior to initial evaluation, or failure to exhibit measureable lameness >5% as compared to the contralateral limb on the PSW^a. Dogs presenting with lameness due to cranial cruciate rupture were only included following full disclosure of the experimental nature of prolotherapy and if owners declined the recommended surgical therapy.

Initial evaluation included a brief owner questionnaire to define the limb affected and duration of lameness, history of any orthopedic surgery, and the type and duration of current pain medications or supplements. Owners were also given the Canine Brief Pain Inventory (CBPI), with the same individual required to complete the survey during each evaluation. The CBPI is a previously validated two-part owner questionnaire evaluating both the pain severity (PS, questions 1–4) and pain interference (PI, questions 5–10) associated with daily activities.^26,27 A complete orthopedic examination and visual veterinary lameness exam were performed with a lameness grade of 0–5 assigned by a single observer (JMS) as previously reported and described in Appendix 1.²⁵ The dog was walked across a PSW^a by one of two handlers to obtain five valid trials for evaluation of stance time, stride velocity, PVF, vertical impulse, and maximum peak pressure, using system-specific software^b. Walking velocity was controlled to achieve 1.0–1.9 m/s with a standard deviation of 0.1 m/s.^27,28 If a lameness of >5% PVF difference from the contralateral limb was detected, the dog was then sedated with hydromorphone^c 0.15 mg/kg and acepromazine^d 0.02mg/kg IV to obtain orthogonal computed radiographs of the elbow or stifle of the affected joint, as determined by orthopedic examination and palpation by a single observer (JMS). If suspicion of other joint involvement on the same limb existed, radiographs of the additional joints were obtained to rule out other causes of lameness. Enrollment was continued if radiographs confirmed osteoarthritis of the affected joint with no evidence of more than one source of arthritis or sources of pain on examination of that limb. Joint radiographs taken at 0 and 12 wk were scored for osteoarthritis by a blinded, board certified radiologist (LJA) at the time of study conclusion in a manner similar to previously reported criteria.²⁹ Radiographs were scored as follows: normal = 0, mild OA = 1, moderate OA = 2, severe OA = 3. Radiographic OA scoring was used to confirm presence of OA and to evaluate any changes in severity at final evaluation. Goniometry was performed by a single observer (JMS) using a two-arm plastic goniometer^e with one degree increments as previously reported.³⁰ The means of three values for flexion and extension of the affected and contralateral joint were recorded.³⁰

While the dog was still sedated, the affected joint was clipped and prepped with chlorhexidine scrub and an alcohol wash. Aseptic injection of the PROLO agent (25% dextrose consisting of 4 mL sterile water^f, 1 mL 2% lidocaine^g, and 5 mL 50% dextrose^h) or the placebo (4 mL sterile water, 1 mL 2% lidocaine) was performed by a single blinded investigator (JMS). To maintain consistency, a volume of 5 mL was selected for injection in both the PROLO and placebo groups. Intra-articular injection was confirmed by detection of grossly observed joint fluid and joint distension. Volume of joint fluid removed was not recorded, and there was no attempt to remove any more than was necessary to subjectively confirm needle location based on the appearance of joint fluid. Dogs were monitored for any signs of postinjection pain for a minimum of 4 hr after treatment and then discharged with instructions to monitor for increased pain, lameness, or swelling. No dogs required postinjection pain medications based on clinician or owner assessment.

Dogs were re-evaluated at 6 and 12 wk. Evaluation at wk 6 involved the same historical questions, CBPI completion by the same owner, repeat orthopedic examination, visual lameness scoring, PSW^a evaluation, and repeat injection of treatment or placebo while utilizing the same sedation protocol as wk 0. The final evaluation at wk 12 consisted of all components of wk 6 except the intra-articular injection and with the addition of repeat radiographs of the affected joint.

Results

Seventeen dogs were evaluated. Ten dogs met inclusion and exclusion criteria and were enrolled. The most common cause for exclusion was failure to exhibit a measurable lameness of >5% as compared to the contralateral limb (n = 6), and one dog was eliminated based on suspected neurologic disease. Initial and wk 6 data from one placebo dog was included, but wk 12 data was lost as the dog died of an unknown cause 18 days after the 6 wk treatment. The injection site was not reported to be abnormal at the time of death, so it is believed to be unrelated to treatment. Of the 10 dogs enrolled, five were randomly allocated to receive PROLO, while the remaining five received the placebo. The mean age of all dogs was 5.7 yr (median 4.5 yr). There was no significant difference in mean age of the PROLO (5.7 yr) and placebo groups (7 yr, p = .601). The mean body weight at wk 0 for all dogs was 38.4 kg (median 35.8 kg). There was no significant difference for mean body weight of the PROLO (36.58 kg) and placebo groups (40.1 kg, p = .599). The mean duration of lameness was 19.5 and 9.4 mo for PROLO and placebo groups, respectively (p = .350) with no significant difference in initial median veterinary visual lameness scoring between treatment groups (p = 1.0). There were no significant differences between PROLO and placebo groups for the initial CBPI PS (p = .834) or PI scores (p = 1.0). Three stifles and two elbows were randomly assigned to the PROLO group, while three elbows and two stifles were assigned to the placebo group. Four of five dogs in the PROLO group and one of five placebo dogs had physical exam findings consistent with OA of the contralateral joint, with the difference between groups not statistically significant (p = .058). One PROLO and two placebo group dogs were concomitantly receiving concurrent NSAIDs and nutraceuticals. Age, body weight at each time period, duration of lameness, and ROM at each time period were not significantly different between PROLO and placebo groups.

Median lameness scores at wk 0 were 3/5 and 2/5 for PROLO and placebo groups, respectively. Median CBPI PS and PI scores are presented in Table 1. From 0 to 12 wk, median PROLO lameness scores improved by one point for PROLO and did not improve for the placebo group, with no significant difference between groups (p = .391). When the change in CBPI scores over time were evaluated, the CBPI PS score in the PROLO group from wk 6–12 (median = 0) was significantly less improved (p = .027) than the change of CBPI PS score from wk 6–12 (median = -6.5) in the placebo group. There were no other significant differences in change in CBPI PS or PI scores between groups. There were no other significant differences in the median CBPI PS and PI scores between treatment groups at any time period.

TABLE 1 Median Scores for CBI Pain Severity (PS) and CBI Pain Interference (PI) for Each Group at Each Time Period

View Fullscreen

There was no significant difference in mean change in ROM at wk 0–6 (p = .708), wk 6–12 (p = .424) or wk 0–12 (p = .393) between PROLO and placebo-treated joints (Table 2).

TABLE 2 Change in Range of Motion Over Time and Between Groups

View Fullscreen

Median OA scores for both PROLO and placebo groups were 2 at wk 0 and 12. There were no significant differences in OA scores between PROLO or placebo groups at wk 0 (p = .754) or at wk 12 (p = .806).

Kinetic gait data for PROLO and placebo-injected limbs are presented in Table 3. The stride velocity of the contralateral limb in the PROLO group (-12.15%) decreased significantly compared to that of the contralateral limb in the placebo group (19.94%) from wk 6–12 (p = .005). For measured stance time, stride velocity, PVF, vertical impulse, and maximum peak pressure, there were no significant percentage changes over time in either the treated or contralateral limbs between treatment groups, despite the apparent improvement of PROLO dogs compared to placebo (Figures 1, 2). Individual changes for all dogs with an increase in PVF >4% are presented in Table 4. Likewise, all dogs with a decrease of >4% in PVF are reported in Table 5. Two dogs with PROLO injections in cruciate deficient stifles (Table 5) experienced large increases in weight-bearing forces. It was also noted that in the group that had decreased PVF (Table 5), dogs tended to have much longer duration of lameness (range 1.5–48 mo), had more elbow osteoarthritis as compared to stifle, and also had higher weight (range 33–49.5 kg) than the dogs that improved (range 24–33). Post hoc power analysis indicated that a sample size of 29 to 106 animals per group would have been needed to see a significant difference in peak vertical force between treatment groups at various time intervals.

TABLE 3 Percent Change and Standard Deviation for Ground Reaction Forces Measured by Pressure Sensing Walkway (Injected Limbs Only) Between Groups and Over 0, 6, and 12 Wk

View Fullscreen

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

TABLE 4 Dogs with Increase of ≥4% in Peak Vertical Force from 0–12 Wk

View Fullscreen

TABLE 5 Dogs with Decrease of ≥4% in Peak Vertical Force from 0–12 Wk

View Fullscreen

Discussion

Overall, the results of this study did not find a statistically significant difference in most subjective and objective parameters of lameness evaluated after injection of osteoarthritic joints with dextrose prolotherapy as compared to placebo. The two findings that were significantly different between treatment groups (CBPI PS score and stride velocity at wk 6–12) do not support benefit of dextrose prolotherapy as compared to placebo, but such sporadic findings are likely a result of normal variable distribution. Dogs in both treatment groups did not exhibit significant differences for criteria of age, body weight, affected joint, use of analgesics, and pre-treatment subjective veterinary lameness scoring. Mean duration of lameness was 19.5 and 9.4 mo for PROLO and placebo groups, respectively. While this difference was not found to be significantly different, the longer duration of lameness could potentially have contributed to a decreased response to PROLO as measured by CBPI results, range of motion, and pressure sensing walkway data. While we could not demonstrate that increased duration of lameness was related to increased severity of OA and morbidity in this study, it is possible that the PROLO dogs may have been more likely to have chronic pain. Likewise, the number of dogs with contralateral limb abnormalities was not significantly greater in the PROLO group (four) as compared to placebo (one) but may have affected veterinary lameness scoring and CBPI results.

There was no significant difference in subjective veterinary lameness scoring between groups at any recorded interval. This would suggest that the longer duration of lameness noted in the PROLO group was inconsequential to outcome. Past studies have found that subjective lameness evaluation is inferior to force plate analysis unless severe lameness exists.³¹ Dogs in the PROLO group showed a greater improvement in lameness score from wk 0–12 as compared to placebo, but this result was not significantly different. CBPI PS and PI scores found that the placebo group showed significant improvement as compared to treatment when evaluated from time 6–12, but there was no significant difference for all other times between groups. Possible explanations for this finding include placebo effect influencing owner survey results, lack of significant treatment effect in the dextrose prolotherapy group, or the small number of dogs evaluated. Recent evaluation of the ability of the CBPI to detect significant improvement in osteoarthritic dogs treated with carprofen found that a minimum pretreatment inclusion criteria of >2 for both mean PS and PI scores was necessary to detect improvement with treatment.²⁷ This same study found that criteria of a decrease in PS of >1 and PI of >2 resulted in the most statistical power to predict if a treatment would lead to response in an individual dog. In the present study, both groups had equivalent total PS and PI scores well above these minimum criteria with only one dog in each group having values <2 for either mean PS or PI, or both. Based on published CBPI criteria and our data, our dogs had sufficient lameness to allow for detection of significant treatment effect. Both PROLO and placebo had mean CBPI PS/PI improvements, but the changes were not significant.

PROLO dogs gained an average of 3.4 degrees and placebo dogs lost 4.5 degrees ROM from enrollment to study conclusion. Despite this apparent difference, the overall mean change in range of motion was not significantly different between treatment groups.

We did not find significant differences between ground reaction force change over time but did note significant difference for one kinetic variable (stride velocity) between treatment groups. As shown in Table 4, some dogs diagnosed with bilateral cranial cruciate ligament ruptures had increased ground reaction forces in the treated limb, but this increase was not statistically significant within our small sample size. The fact that the stride velocity of the contralateral limb in the treated group decreased significantly compared to that of the contralateral limb in the untreated group from wk 6–12 is most likely due to transient increased contralateral lameness in the PROLO dogs with known contralateral disease (n = 4). This finding may also be consistent with the lack of significant positive treatment effects observed in PROLO treated limbs. Further investigation is warranted.

Another potential confounding variable is that placebo injection could have acted as a positive control. The placebo injection could have lead to some transient improvement in pain due to the presence of the local anesthetic as well as caused a localized inflammation, thereby mimicking the effects of PROLO. In retrospect, synovial fluid analysis at all time points could have further characterized the response to injections. It is also possible that different volumes of injections might yield differing effects due to inherent anatomic variations among different joints or dogs of different size. Further evaluation of these parameters could be clinically useful.

The primary limitation of this study was the small number of dogs and, thus, low power achieved. Post hoc power analysis indicated that a sample size of 29 to 106 animals would have been needed to see a significant difference in peak vertical force between treatment groups at these time intervals. Other limitations are those inherent to prospective studies involving dogs with multiple joints affected by OA. While it was our intent to identify and enroll dogs with reported single-limb lameness, many of our patients, in fact, had contralateral OA and OA affecting both rear and forelimbs. The fact that 4/5 dogs in the PROLO group had contralateral disease may have contributed to the lack of improvement noted. Another potential limitation would be the inclusion of 5/10 dogs with lameness due to cranial cruciate ligament rupture in the study. The failure to show improvement in all dogs in this subset may have been due to continued instability due to ligament rupture or a type II error. Based on several studies in human literature showing benefits of dextrose PROLO in anterior cruciate ligament associated instability, we felt it worthwhile to include dogs with cranial cruciate ligament disease.^16,18,21 One study in humans found that dextrose prolotherapy-treated knees had statistically improved anterior displacement and patient-reported knee buckling.¹⁶ In addition, PROLO has been reported in several case reports as well as conference proceedings as having potential benefits as a nonsurgical treatment for cranial cruciate ligament disease in dogs, and further investigation is warranted.^22–24 It is possible that subsequent studies limiting enrollment to dogs with OA and no cruciate deficiency or other cause of joint instability might find differing results for prolotherapy. However, the population was representative of dogs presenting for management of OA in a practice setting and thus appropriate for this type of study.

One other limitation includes NSAID use during the study. While NSAID use was only present in a small portion of dogs in our study, it's possible that use could have biased results. Ideally, dogs would have all been taken off analgesic medications and undergone a washout period prior to study enrollment. Although controversial, many practitioners of prolotherapy in humans recommend discontinuing NSAIDs after treatment to prevent inhibition of the desired inflammatory response.³ We elected to allow dogs to be continued on any previous medical management to avoid increasing their pain. Future studies might include more stringent exclusion guidelines.

Conclusion

This study failed to demonstrate significant benefits of intra-articular dextrose prolotherapy as compared to placebo for treatment of osteoarthritis of the elbow and stifle in dogs. The treatment was well tolerated and inexpensive. Based on the results of this study, the authors recommend further studies with greater numbers of dogs to definitively investigate prolotherapy in a larger population of dogs.