RETRACTED ARTICLE: Effect of a single intra-articular high molecular weight hyaluronan in a naturally occurring canine osteoarthritis model: a randomized controlled trial

Background Osteoarthritis (OA) is a complex joint disease and chronic pain source, affecting a patient’s quality of life and posing a financial burden. As the dog is considered a nearly ideal species for translation research of human OA and the most used model for research, exploring spontaneous dog OA under the One Health/One Medicine concept can improve both humans and dogs’ health and well-being. Methods In a clinical treatment experiment, forty (N=40) joints were selected and randomly assigned to a control group (CG), which received 0.9% NaCl or a treatment (HG), which received Hylan G-F 20. Evaluations were performed on treatment day (T0), 8, 15, 30, 90, and 180 days post-treatment. They consisted of four different Clinical Metrology Instruments (CMI), evaluation of weight distribution, joint range of motion, thigh girth, radiographic and digital thermography imaging, synovial fluid interleukin-1 (IL-1), and C-reactive protein concentrations. Results were compared with repeated measures ANOVA, with a Huynh-Feldt correction, Paired samples T-test, or Wilcoxon signed-ranks test, with p<0.05. Results Patients had a mean age of 6.5±2.4 years and a bodyweight of 26.6±5.2kg, and joints graded as mild (n=28, 70%), moderate (n=6, 15%), and severe OA (n=6, 15%). No differences were found between groups at T0. Symmetry index and deviation showed significant improvements in HG from 30 days (p<0.01) up to 180 days (p=0.01). Several CMI scores, particularly pain scores, improved from 90 to 180 days. Radiographic signs progressed in both groups. In both groups, increasing body weight and age corresponded to worse clinical presentation. IA hyaluronan administration produced increased lameness in six cases, which resolved spontaneously. Conclusions This study characterizes the response to treatment with Hylan G-F 20, which can produce significant functional and pain level improvements in patients with OA, even those with factors related to worse response to treatment.


Background
Osteoarthritis (OA) is a highly prevalent disease worldwide, which affects all mammals and a leading cause of disability. It can negatively impact both the population's physical and mental well-being, with substantial healthcare resources and costs associated with managing the disease [1,2]. The dog is an ideal species to study human OA, with the advantages of being anatomically, biochemically, genomically, and molecularly similar to humans, with clinical progression and treatment similarities [3]. At the same time, they have a foreshortened lifespan but with human equivalent life and disease stages while sharing many environmental variations that influence human OA. The study of spontaneous canine OA and its treatment can add to the knowledge of the treatment of the human disease as well, under the One Medicine initiative [4,5].
OA is an incurable condition, and its management focuses on alleviating symptoms, particularly pain. An additional goal is to improve overall joint function while slowing down disease progression [5,6]. Hyaluronan, the high molecular glycosaminoglycan, is synthesized by chondrocytes and synovial fibroblasts [7]. It forms the backbone of proteoglycans aggregates interwoven with collagen to create hyaline cartilage's unique structure [8]. Information from animal models shows that endogenous hyaluronan is cleaved by free radicals in OA. Its quantity and quality are affected in OA joints, more severely in clinically affected patients, supporting its exogenous administration [9]. Even though its mechanism of action is not entirely understood and clinical trials have provided contradictory results, hyaluronan treatment aims to reduce pain and improve function by supplementing synovial fluid viscosity and elasticity [10]. Additional antiinflammatory, anti-nociceptive, and chondroprotective properties have been suggested, through the enhancement of cartilage synthesis, blunting response to IL-1, protection from the damage of oxygen free radicals, and protection of chondrocytes from apoptosis [7,11].
Human reports show that intra-articular hyaluronan, given once weekly for 3 weeks, increased mobility and reduces pain and the need for nonsteroidal antiinflammatory drugs to control pain [12]. A systematic review concluded that there is a lack of standardization regarding intra-articular hyaluronan administrations for hip OA, with no consensus on its efficacy [13]. Although it is not clear if any formulation has a superior diseasemodifying effect [14], high molecular weight products seem to produce better results, particularly in patients with mild radiographic disease [15]. A recent report showed that both single or 1-3 weekly injections of Hylan G-F 20 at 1 year following the first injection for knee OA are efficacious and generally well tolerated for long-term use [16]. Many studies performed in canine experimental OA models have failed to demonstrate clear benefits of hyaluronan supplementation [17]. In a canine surgical model, IA hyaluronan provided clinically significant improvement in pain, function, lameness, and kinetics compared to pre-treatment and saline control, without preventing OA's progression [18]. In a rabbit model, hyaluronan administration produced a more normal cartilage after immobilization [19]. In dogs with naturally occurring OA, treatment groups have significantly better results than a control group by the 6th week posttreatment [20].
Multiple agents influence OA catabolism, but interleukin 1 (IL-1) is commonly pointed out as the major proinflammatory cytokine [21]. C-reactive protein (CRP) is an acute-phase protein produced during inflammatory reactions or tissue injury from an early stage [22].
Radiographic examination is a staple in OA's assessment, and the ventrodorsal (VD) hip extended view is the most commonly used projection. An additional useful projection is the ventrodorsal flexed view, also called frog-legged view (FL), specifically in the evaluation of the circumferential femoral head osteophyte (CFHO) and caudolateral curvilinear osteophyte (CCO), early radiographic signs related to the development of the clinical symptoms [23]. Digital thermal imaging relies on the between physiologic functions generated heat and its relation with skin temperature control, being reliable in assessing inflammatory arthritis pain and osteoarthritic subjects [24]. Functional evaluation is also paramount in determining response to treatment in OA, and stance analysis has been reported as a sensitive evaluation for detecting lameness in dogs [25]. It evaluates weight distribution since patients commonly bear less weight on a painful limb [26]. An additional functional evaluation includes determining activity levels and mobility impairments since they are associated with musculoskeletal pain [27]. Pedometers are capable of measuring ambulatory activity with acceptable accuracy [28]. Clinical examination of patients commonly includes evaluating muscle masses, muscular atrophy being a consistent finding in OA patients, and determining the joint range of motion (ROM, flexion, and extension), which can present restrictions [29].
Pain is a hallmark of OA, and canine studies offer valuable data that may translate to humans [30,31]. For pain evaluation and its impact on patients' lives, several clinical metrology instruments (CMI) have been developed. The Liverpool Osteoarthritis in Dogs (LOAD) and the Canine Brief Pain Inventory (CBPI) are the most commonly used [27], with the CBPI being divided into a pain severity score (PSS) and a pain interference score (PIS) [32]. The Canine Orthopedic Index (COI, divided into four scores: stiffness, gait, function, and quality of life (QOL) and the Hudson Visual Analogue Scale (HVAS), developed to assess the degree of lameness in

A R T I C L E
dogs, are additional validated evaluation tools [33,34]. Digital thermal imaging is a technique that has recently gained attention. It relies on heat generated during physiologic functions and its relation with skin temperature control [35]. It has been used to assess inflammatory arthritis pain and differentiate normal from osteoarthritis subjects [24]. This study aimed to describe the effect of a high molecular weight hyaluronan product (Hylan G-F 20) in OA management in a naturally occurring canine model. We hypothesize that a single administration will reduce clinical signs of OA compared with a control group.

Methods
The study protocol was approved by the ethical review committee of the Universidade de Évora (ORBEA, approval n°GD/32055/2018/P1, September 25, 2018) and complies with the ARRIVE guidelines. Written informed consent was obtained from the Institution responsible for the animals. Twenty patients with naturally occurring bilateral hip OA, constituting a convenience sample, were signaled from a population of active police working dogs, comprising forty (N=40) hips joints. The diagnosis was made based on history, physical, orthopedic, neurological, and radiographic examinations. Additional inclusion criteria included a bodyweight of ≥ 20kg, age ≥2 years, and they should not have received any medication or nutritional supplements for at least 6 weeks. Patients with other suspected or documented orthopedic or concomitant disease and not tolerant of data collection were excluded.

Treatment administration
In a double-blinded study, patients were randomly assigned using the statistical analysis software to two groups, 10 dogs per group, and treated bilaterally: a control group (CG, n=20), which received an intra-articular (IA) administration of 2ml of 0.9%NaCl, and a treatment group (HG, n=20), which received a single IA administration of 2ml of Hylan G-F 20 (Synvisc®, Sanofi, Portugal). Radiographic examinations and IA administrations were conducted under light sedation, using a combination of medetomidine (0.01mg/kg) and buthorphanol (0.1mg/kg), both given intravenously simultaneously. The procedure for intra-articular administrations to the hip joint has been described before [36]. The patient was positioned in lateral recumbency, with the affected joint uppermost, to access the joint of interest. A window of 4 × 4cm surrounding the greater trochanter was clipped and aseptically prepared. An assistant then positioned the limb in a neutral and parallel to the table position. The joint space was accessed using a 21gauge with 2.5″ length needle, introduced just dorsal to the greater trochanter and perpendicular to the limb's long axis. Correct needle placement was confirmed by collecting synovial fluid (immediately collected and processed for future analysis), and the treatment or saline were administered. Patients were rested for three consecutive days following treatment, after which normal activity was resumed over 5 days. One and 3 days after the IA procedure, animals were examined by the assisting veterinarian for signs of exacerbated pain, persistent stiffness of gait, and changes in posture exhibited. Evaluations were conducted on days 0 (treatment day), 8,15,30,90, and 180 by the same researcher. An outline of procedures and evaluations conducted in each evaluation moment is presented in Table 1.

Evaluation of weight-bearing distribution
The weight distribution evaluation was performed with a weight distribution platform (Companion Stance Analyser; LiteCure LLC®, Newark, DE, USA). According to the manufacturer's guidelines, the equipment was placed in the center of a room, at least 1 m from the walls. It was calibrated at the beginning of each day and zeroed before each data collection. After this procedure, animals were placed with one foot in each quadrant of the platform, using gentle restraint when required. A left-right symmetry index (SI) was calculated with the following formula: SI=[(WBR-WBL)/((WBR+WBL) × 0.5)]×100 (WBR is the weight-bearing of the right limb, and WBL is the weight-bearing of the left limb). Negative values were made positive [37]. We additionally considered a deviation from normal 20% weight-bearing for a pelvic limb, calculated by subtracting WB to 20.

Digital thermography imaging
Digital thermography evaluation was conducted in a room with a controlled temperature, set at 21°C. Previous to collecting the images, animals were allowed to walk around the room for 30 min. They were then placed in an upright standing position, and a dorsoventral thermographic image was obtained, including the area from the last lumbar vertebra to the first coccygeal vertebra, at a distance of 60cm [38], FLIR ThermaCAM E25® model (FLIR Systems, Wilsonville, OR, USA). Images were analyzed using the free software Tools (FLIR Systems, Inc), with a rainbow color pallet. Boxes of equal size were placed on the hip joint's anatomical area on both views to determine mean and maximal temperatures.

Radiographic evaluation
In the VD radiographic projection [23], seven radiographic signs were assessed: irregular wear on the femoral head, making it misshapen and with a loss of its rounded appearance; a flattened or shallow acetabulum, with irregular outline; CCO; new bone formation on the acetabulum and femoral head and neck; a worn away

A R T I C L E
angle formed at the cranial effective acetabular rim; subchondral bone sclerosis along the cranial acetabular edge; and CFHO.

Clinical and laboratorial findings
Thigh girth was determined with a Gullick II measuring tape, at a distance of 70% thigh length, measured from the greater trochanter's tip, with an extended leg [39]. Hip joint ROM was obtained with a goniometer at extension and flexion with a flexed stifle [40]. Pedometers were worn around the patient's neck, attached to an adjustable lightweight collar [41]. Pedometers were worn for 1 week before the first evaluation moment to set a baseline value. For each of the following evaluations, animals worn the pedometer for a week before that evaluation moment. A mean daily count was calculated by dividing the registered number of steps by the number of considered days. In each evaluation moment, trainers completed a copy of HVAS, CBPI, COI, and LOAD after receiving the published instructions for each of them.

Results
This study sample comprised 40 joints of active police working dogs, with a mean age of 6.5±2.4 years, a mean bodyweight of 26.6±5.2kg, and of both sexes (13 males and 7 females). Dogs were of breeds commonly employed in police forces, similarly distributed between CG and TH: German Shepherd Dogs (n=6, 3 in CG and 3 in TH), Labrador Retriever (n=6, 3 in CG and 3 in TH), Belgian Malinois Shepherd Dogs (n=5, 3 in CG and 2 in TH), and Dutch Shepherd Dog (n=3, 2 in CG and 1 in TH). At the initial evaluation, joints were graded with the OFA hip grading scheme as mild (n=28, 70%), moderate (n=6, 15%), and severe (n=6, 15%). No differences were found between groups at the initial evaluation. Increased lameness was observed in 6 joints HG, which spontaneously resolved within 48-72h.

Clinical and CMI results
Values recorded for different evaluations in each group at T0 are presented in Table 2. Comparing results between groups with repeated measures ANOVA with a Huynh-Feldt correction, significant differences between groups were found concerning deviation (F(5, 160)=3. 7 Fig. 1. Results of the Kaplan-Meier test are presented in Table 3.

Functional assessment
Stance analysis X X X X X X Pedometer X X X X X X Goniometry X X X X X X Thigh girth measurement X X X X X X

Imaging
Digital Thermography X X X X X X Digital radiography X X X X

A R T I C L E
Kaplan-Meier curves for stiffness score and PIS are presented in Figs. 2 and 3, respectively.

Radiographic evaluations
The frequency of different radiographic findings at the initial and final evaluations is presented in Table 4

Discussion
Osteoarthritis is the most commonly diagnosed joint disease in human and veterinary medicine, with limited treatment options. In addition to the anatomical and biochemical similarities between dogs and humans, they also share an environment and lifestyle. For those reasons, the study of animal OA could be beneficial for both species [2,5]. To our knowledge, this is the first study to describe the effect of a single injection of high-density hyaluronan (G-F 20) on several clinical, imaging, and laboratorial signs in a naturally occurring canine model, with a long follow-up period. Dog OA, particularly naturally occurring OA, resembles closely human OA regarding anatomy, disease heterogeneity, and progression [42]. Many studies performed in canine experimental OA models have failed to demonstrate clear benefits of hyaluronan supplementation [17]. IA hyaluronan provided clinically significant improvement in animals with stifle OA in pain, function, lameness, and kinetics compared to pre-treatment and saline control in a canine surgical model. Maximum benefits were noted at 4-8 weeks and gradually tapered down by a 6-month evaluation time point [18]. In dogs with naturally occurring OA, treatment groups have significantly better results than a control group by the 6th week post-treatment but accompanied by exercise restrictions, leading to improvements in the control group [20]. In this study, we have observed significant improvements in the HG with several evaluation modalities, which, in some cases, lasted up to the last evaluation moment, at 180 days post-treatment. These include functional improvements measured by the evaluation of weight-bearing, to improvements in other dimensions of OA, as measured with the CMIs, but particularly with the two scores of the CBPI. In addition to group improvements in HG, individual CMI scores also improved in most animals from the first evaluation post-treatment, but particularly after 15 days. This improvement is observable with the Kaplan-Meier test results for SI, with results in HG taking significantly longer to return to baseline values. It was also noticeable with different CMI scores and dimensions. Although clear anatomical similarities exist, some care must be taken when extrapolating dogs to humans. The dog, being a quadruped, supports 60% of

A R T I C L E
body weight in the thoracic limbs and 40% in the pelvic limbs, which differs from the biped posture of humans, which can affect OA's progression [43,44]. A proposed direct analgesic effect for hyaluronan has been suggested in animal models by action over the opioid receptor [45]. An additional proposed mechanism of action for hyaluronan is producing endogenous hyaluronan production by the exogenous administration, based on in vitro and in vivo studies [46]. This last mechanism may be supported because the product is rapidly cleared from the joint, and maximal clinical improvement does not occur for several weeks, between 60 and 90 days, while persisting for much longer [47]. Our results partly support these findings, with the difference that significant improvements were reached sooner and lasted longer. Although we did not measure the amount and the persistence of the exogenous hyaluronan within the joint, the visual examination of SF in HG at the 8-day evaluation showed a clear SF, with increased viscosity that of Hylan G-F 20.
OA is a low-grade inflammatory disease, and IL-1 is the most important pro-inflammatory cytokine responsible for the catabolism in OA, affecting the disease's progression [48], and the histopathology and pathogenesis of dog OA closely resembles that of human OA [5]. IA hyaluronan inhibits degenerative cartilage changes in animal models due mainly to its pro-inflammatory cytokines and degradative enzymes [49]. Low molecular weight hyaluronan seems to be most effective in reducing the release of cytokines [50]. Although a decrease in IL-1 levels was recorded in both groups, at 8 days, its concentration in CG was significantly lower. At this moment, this is probably due to the removal of synovial fluid at treatment day, followed by the injection of 0.9% NaCl, similar to the effect of a joint lavage, which may be more effective than the administration of hyaluronan in reducing IL-1 levels. Still, IL-1 concentration levels remained lower than those at the initial evaluation in both groups. As this study was a clinical treatment experiment, no joint histological samples were collected, which would help evaluate differences between Hylan G-F 20 and 0.9% NaCl injection. The reduction of IL-1 may reduce inflammatory levels, which are reflected in the temperature values recorded during the thermographic evaluations. Measurements made on the Lt view, in particular, recorded variations throughout the entire follow-up period, with lower levels being recorded in CG.
Pain is the most relevant clinical sign of OA, and its evaluation is paramount to determine OA treatment efficacy so that data may be translated to human medicine [30]. There is strong evidence that humans and animals' type of pain is analogous, as they share neurophysiologic similarities [51]. However, painful experiences in OA are complex, involving several dimensions [52]. While extremely useful in a clinical setting, CMIs can be susceptible to the caregiver placebo effect, associated with the variability in emotional and cognitive components of pain perception. On the other hand, the animal itself will not show a significant placebo effect, and the ability to perform daily activities will likely reflect a lower level of pain [53,54]. We used several CMIs, to try to capture multiple dimensions of OA. As a whole, individual CMI scores in CG tended to worsen through time, while HG scores tended to improve. Still, some animals in CG showed improvements. While some patients with OA may spontaneously improve, a more plausible explanation is related to removing cytokine-loaded SF at treatment day, followed by the injection of 0.9% NaCl, similar to the effect of a joint lavage. Placebo saline injections have shown functional improvements that can last up to a 6-month follow-up [55].
Radiographic evaluation is a staple of OA monitoring. CCO and CFHO represent early radiographic signs that predict the development of hip OA clinical signs [23]. Previous reports have described that hyaluronan could not prevent OA progression based on radiographic assessment [18]. However, it decreased signs of pain and improved joint function after the onset of OA [56]. Our results support these findings. In CG, several radiographic findings progressed throughout the follow-up period, as expected in the disease's natural evolution. This was also observed in HG, even though some radiographic findings did only change at 180 days. Still, despite the evolution of radiographic findings, patients in HG showed better clinical, functional, and pain findings than CG. Also, in the 8-30 days' evaluation period, no significant differences were observed in HG between animals with and without CCO and CFHO at the initial evaluation.
OA risk factors are well characterized and include having a higher bodyweight or being of older age [2]. To assess these factors' influence in response to treatment, we applied different cut-off values for weight. In both groups, increasing body weight generally corresponded to worse CMI. In HG, heavier patients had SI evaluation and deviation, even though the group still had better results than CG throughout the study. Previous reports indicated that larger dogs achieved improvements of 30% or more at 12 weeks [47]. We described improvements earlier, even in heavier patients. Male dogs also scored worse in considered CMI, but this may be related to the fact that male dogs were significantly heavier than females in all considered moments.
Regarding age, similar results were observed. Considering animals above the sample's mean age, these patients scoring worse on almost all CMI scores had lower pedometer counts and higher thermography values. Since OA is a chronic, progressive disease, it was not unexpected to see older patients record worse

A R T I C L E
evaluations, which may be linked with the disease's progression at its clinical signs. While expected, the difference in treatment results is quite pronounced, more than the effect of increased body weight. IA hyaluronan administration has been described as producing mild heat, swelling, and/or erythema postinjection, which resolved spontaneously within a week [18]. These adverse effects are well tolerated and usually restricted to the injected joint [57,58]. Similarly, we observed increased lameness in six cases, reflecting on the 8-day SI and deviation evaluations, when significantly worse scores were kept at HG. This spontaneously resolved by the 15-day evaluation. No additional medication was administered to the animals during the followup period. Considering the obtained results, Hylan G-F 20 may be a good therapeutic option for managing canine hip OA. Its administration was able to reduce pain levels and improve joint function compared to a control group. Due to the close resemblance of canine and human OA, it is possible that the same recommendation can be made for human hip OA. Still, as some differences in weight bearing exist between the two species, futures studies should enroll a greater number of animals and assess if similar results are observed in humans.

Conclusions
This study describes the effect of a single injection high molecular weight hyaluronan product in a naturally occurring canine model, with a long follow-up period. It provides important information for the characterization of the response to treatment, showing that Hylan G-F 20 can produce significant functional and pain level improvements in patients with OA, even those with factors related to worse response to treatment. For that reason, Hylan G-F 20 can be considered as a good therapeutic option for OA management, even in more advanced cases.