1. Introduction
Post-traumatic osteoarthritis (PTOA) is a painful, protracted, degenerative disease of joints that affects an estimated 3 million horses and 250 million people worldwide [
1,
2]. It is reportedly the most common cause of lameness in horses, with significant economic impact due to costs associated with detection and treatment of PTOA [
3]. The disease process has not been completely elucidated; however, it is evident that the upregulation of catabolic pathways results in prolonged inflammation and, ultimately, degradation of the articular cartilage [
4]. There is no cure for PTOA, and current treatments are focused on relieving the pain via inhibiting inflammatory pathways (e.g., nonsteroidal anti-inflammatory drugs) or via viscosupplementation (e.g., synthetic hyaluronate). Treatment also focuses on modifying the disease state by reducing gross articular cartilage degeneration.
Pharmacological interventions that treat the underlying disease have been widely investigated and include small molecule, protein, and gene therapies, with many in pre-clinical large animal studies [
5,
6,
7,
8,
9]. An alternative strategy is one that uses a material to improve the joint environment. One such example is the use of an aqueous polymer solution injected intra-articularly to re-establish synovial fluid viscosity and lubricity, thereby providing chondroprotection [
10,
11,
12,
13,
14]. Inspiration for these biolubricants comes from mimicking the lubricating effects of endogenous bioproteins such as hyaluronic acid, mucin, and lubricin [
9,
15,
16,
17]. Examples include polyacrylates [polyacrylic acid, ref poly(acryloylamino-2-methyl-1-propanesulfone), poly(2-methacryloyloxyethyl phosphorylcholine and poly(2-methacryloyloxyethyl phosphorylcholine)]; polyolefins [poly(7-oxanorbornene-2-carboxylate)], polyamides (polylysine grafted HA), and polyacrylamide hydrogels (PAAG) [
9,
18,
19,
20,
21,
22,
23,
24,
25]. Enhanced lubrication in the joint is advantageous as it dissipates shear forces on articular cartilage, thereby reducing wear from repetitive joint loading [
9]. While the end goal of many synthetic biolubricants is the same, the chemical formulation, physical properties, manufacturing, and tissue interaction can be highly variable and therefore are not always directly comparable [
26].
One of these biolubricants, poly(2-methacryloyloxyethyl phosphorylcholine) (pMPC), is zwitterionic in nature-equal positive and negative electrical charges and when coated on a surface results in a hydrophilic, low-friction material [
9]. The pMPC functions to augment the extracellular matrix of cartilage. Application of pMPC, either in a linear or crosslinked format, to ex vivo bovine cartilage explants reduces the coefficient of friction and decreases tissue strain compared to saline and is superior to hyaluronic acid [
9]. In a model of repeated administration, the crosslinked pMPC outperforms the linear pMPC in its ability to reduce the coefficient of friction and cushion ex vivo cartilage surfaces. Furthermore, a sustained duration of effect was demonstrated, with a prolonged intra-articular residence time of over 30 days in a rat model for PTOA [
22]. Herein, we report the safety and efficacy of an intra-articular pMPC in an established model of PTOA in the horse. We hypothesized the pMPC would decrease the physical, gross, radiographic, histological and biochemical effects of PTOA in this highly translational model of PTOA.
2. Materials and Methods
2.1. Horses
Sixteen horses were included in the study. The horses consisted of Quarter Horses or mixed-breed Quarter Horses; were of a mixed population of mares and geldings; and ranged in age from 2 to 5 years old. Pre-study evaluations included general health, subjective lameness examination, evaluation of carpal effusion, and carpal radiographs. To be admitted to the study, horses required a lameness score of less than or equal to 1 out of 5 on a straight line [American Association of Equine Practitioners’ (AAEP) lameness scale [
27], 0 (normal gait) to 5 (non-weight bearing lameness)] and radiographically normal carpi. Horses were acclimatized to exercise on a high-speed treadmill over the course of 14 days prior to surgery. The study was approved by the institution’s animal care and use committee (Protocol 15-6239A).
2.2. Synthetic Biolubricant
The polymeric lubricant used was 5 w/v% poly(2-methacryloyloxyethyl phosphorylcholine) (pMPC) [
9]. The network polymer was synthesized via copolymerization with the crosslinker ethylene glycol dimethacrylate maintained at 1 mol% (mol/mol MPC). The polymers were purified via dialysis, lyophilized, and resuspended in deionized water at 5 w/v%.
2.3. Experimental Induction of Osteoarthritis
Following induction of anesthesia and routine preparation for bilateral carpal arthroscopy, one middle carpal joint was randomly selected for the surgical induction of OA while the other served as a sham-operated control joint (day 0). Surgeries were performed using a standard approach and a previously described model [
5]. Briefly, an 8 mm osteochondral fragment was created on the distal aspect of the radial carpal bone at the level of the synovial plica and left in situ attached to the joint capsule. The fragment gap was widened to 15 mm using an arthroscopic burr (Arthrex, Munich, Germany). The fragment and all debris from the burring were not removed from the joint [
5,
6]. In the sham-operated joint, the absence of significant lesions was confirmed.
All horses were preoperatively treated with Cefazolin (11 mg/kg IV) and Gentamicin (6.6 mg/kg IV). Phenylbutazone was administered at 4.4 mg/kg per os (PO) preoperatively and continued once daily for an additional 2 days.
2.4. Treatments
Twelve days after surgery, the horses were evaluated for lameness and designated to one of two groups (OA–treatment vs. OA–control). To equalize the lameness grades per group, they were ranked based on lameness scores and randomly designated to each group by alternating the designation. Eight horses were assigned to the biolubricant (pMPC) treatment group and eight to the untreated group (
Figure 1). On day 14, 1.25 mL of synovial fluid was collected from each horse in the treatment group via arthrocentesis from each middle carpal joint. Subsequently, the OA joint received an intra-articular injection of 6ml pMPC, while the contralateral sham-operated joint received 6 mL 0.9% saline intra-articularly. In the untreated group, similarly, each middle carpal joint had 1.25 mL of synovial fluid aspirated via arthrocentesis, followed by an intra-articular injection of 6 mL saline. The injector (LRG) was not blinded to treatment nor OA-induced joints. Horses were administered a 1.1 mg/kg dose of flunixin meglumine intravenously once daily for 3 days.
2.5. Exercise Protocol
Horses were housed individually in 3.65 × 3.65 m stalls. Days 1–12, horses were maintained on stall rest. Day 13 and again on Day 14 prior to treatment, the horses were exercised lightly at a trot (4.0–5.0 m/s) for 6 min on a high-speed treadmill (EquiGym, Lexington, KY, USA). The horses were allowed 4 days of rest following treatment (days 15–18).
2.6. Lameness Scores
Clinical examination and lameness evaluations were performed by a board-certified equine sports medicine specialist (KAS) unaware of the treatment groups. Baseline lameness was assessed following the initial treadmill acclimatization period and reported prior to surgery on day 0. Postoperative lameness exams were performed on days 10, 14, and 19 (prior to, day of, and following treatment) and then once weekly starting on day 21 until day 70. Subjective and objective lameness data were reported for each lameness evaluation. The subjective evaluation used the AAEP 1–5 graded lameness scale [
27]. Objective lameness data were collected using an inertial sensor system (Equinosis
® Lameness Locator, Columbia, MO, USA) [
28]. Specifically, the foresigned vector sum was documented, which indicates the direction (positive values for right forelimb lameness and negative values for left forelimb lameness) and magnitude of lameness calculated from millimeters of displacement. Clinical lameness is associated with a vector sum greater than 8.5 mm. At each lameness exam, horses were also evaluated for carpal effusion and response to flexion using a subjective ordinal grading scale of 0 to 4 (0 = normal, 1 = slight, 2 = mild, 3 = moderate, and 4 = marked/severe).
2.7. Diagnostic Imaging
Radiographic assessment of both carpi was performed prior to study inclusion (baseline; day 0), day 14, and day 70. Radiographic views included lateromedial, dorsopalmar, dorso 30° medial–palmarolateral oblique (DMPLO), dorso 45° lateral–palmaromedial oblique (DLPMO), and flexed lateromedial projections. A board-certified radiologist (MBF), blinded to treatment grouping, graded the radiographic examinations based on a previously established grading scale for five parameters: (1) osseous proliferation at the dorsal joint capsule (enthesopathy), (2) subchondral bone lysis of the radial carpal bone, (3) subchondral sclerosis of the radial carpal bone, (4) subchondral sclerosis of the third carpal bone, and (5) osteophyte formation. For each radiographic outcome parameter, a scale of 0 to 4 was used (0 = no detectable abnormality, 1 = slight change, 2 = mild change, 3 = moderate change, and 4 = severe change). A total radiographic score was also calculated for each limb based on the summation of scores from the 5 parameters.
2.8. Synovial Fluid Analysis
Synovial fluid was collected from each middle carpal joint on days 0, 14, 28, 42, and 70; approximately 1.25 mL was aspirated at each collection. Half of this volume was analyzed for total nucleated cell count and total protein concentrations (within 12 h of collection); the other half was centrifuged, and the supernatant was frozen at −80 °C in plastic microtubes until analysis for prostaglandin-E2 (PGE
2); interleukin 1 receptor antagonist protein (IL-1Ra) and glycosaminoglycan (GAG) concentrations were performed as previously described [
29,
30].
2.9. Serum Biomarkers
Whole blood was harvested from the jugular vein and processed to harvest serum on days 0 and 14 and then once every other week from days 28–70. An aliquot of serum was used to measure liver and kidney function enzymes, including Aspartate Aminotransferase (AST), Gamma-glutamyl Transferase (GGT), and Creatinine (within 12 h of collection). The remaining serum was frozen at −80 °C in plastic microtubes until PGE2 and GAG concentrations were analyzed.
2.10. Gross Pathology and Histology
Horses were euthanized with an overdose of intravenous sodium pentobarbital (Euthanasia Solution, VetOne, Boise, ID, USA). Immediately following euthanasia, their middle carpal joints were disarticulated, photographed, and evaluated for the presence/absence of fragments, fragment size, total cartilage erosion, total joint hemorrhage, full thickness and partial thickness cartilage erosion, kissing lesions, and synovial adhesions [
5]. All parameters were graded on a subjective ordinal scale (0 = normal to 4 = severe) apart from the kissing lesions and synovial adhesions, which were graded based on their presence (yes or no) [
5]. The expert grading the joints (KAS) was not blinded to the presence of osteochondral fragments but was blinded to treatment group assignments.
Following gross macroscopic evaluation, synovium, cartilage, and subchondral bone specimens were collected. Synovium was harvested from a villous area. Cartilage was harvested from the radial facet of the third carpal bone (C3), the fourth carpal bone (C4), and the distal radial carpal bone (RCB); subchondral bone samples were harvested from C3 and RCB. All samples were placed in neutral-buffered 10% formalin and processed for histologic evaluation. Cellular changes were assessed in synovial, cartilage, and subchondral bone samples using Hematoxylin and Eosin (H and E) staining. Synovium was evaluated for intimal hyperplasia, subintimal edema, subintimal fibrosis, and vascularity. Cartilage was evaluated for fibrillation, chondrone formation, chondrocyte necrosis, and focal cell loss. Subchondral bone was evaluated for osteochondral lesions, subchondral bone remodeling, and osteochondral splitting. Changes in GAG content were assessed in cartilage by staining with Safranin O and Fast Green (SOFG). For each of the previously mentioned samples, GAG content was measured as the presence of stain uptake in tangential, intermediate, radiate territorial, and radiate interterritorial zones. For all synovium, cartilage, and subchondral bone samples, individual scores were assigned as well as a summation score for all parameters. Histology was graded by a single evaluator (LRG) blinded to the treatment assignments using a modified Mankin scoring system [
31].
2.11. Surface Topography
Surface topography was performed as previously described [
32]. In short, osteochondral plugs 4 mm × 4 mm × 8 mm were harvested from C3 and sent to the Department of Orthopedic Surgery, Rush University Medical Center for surface topography analysis. Using a scanning white light interferometry microscope, nine measurements were taken from the surface of each sample in a 3 × 3 square array. Mean parameters were computed for the following: maximum peak-to-valley depth (PV), root mean square roughness (Rq), arithmetic mean roughness (Ra), skewness (Rsk)—a measure of the symmetry of the deviations about the center plane—and the arithmetic average of the five highest peaks and five lowest valleys (SRz).
2.12. Statistical Analysis
Statistical analysis was performed using SAS 9.4. Residual diagnostic plots were used to evaluate assumptions of normality and equal variance. Some variables were transformed (using log or square root) in order to better satisfy model assumptions.
A mixed model was run separately for each response variable. All models included horse, within treatment, and phase as random effects. Seventeen variables were identified as primary responses, including lameness (subjective and objective), flexion, and effusion scores for clinical analysis; osteophytes and summation scores for radiographic analysis; IL-1Ra, PGE2, total protein, and total nucleated cell count for synovial fluid analysis; cartilage summation, fibrillation, full thickness, partial thickness, and total erosion scores; and synovial subintimal fibrosis and summation scores for gross and histologic analysis. For the additional response variables, a Bonferroni adjustment (corresponding to the other 66 responses) was used to control for multiple testing. If there was evidence of a treatment main effect or interaction based on F-tests, then pairwise comparisons were considered.
Clinical analysis, radiographic analysis, and synovial fluid analysis were measured for each limb at the aforementioned timepoints (
Figure 1). Treatment, OA status, day, and all interactions were included as fixed effects. At each timepoint, the four combinations of treatment and OA were compared using Tukey’s method. For each treatment/OA combination, comparisons versus day 14 (postoperative, prior to injection) were made using Dunnett’s method.
Serum biomarkers were measured for each animal at the aforementioned timepoints (
Figure 1). Treatment, day, and treatment*day interaction were included as fixed effects.
Gross pathology, histology, and surface topography were all measured on each limb at the final timepoint. Treatment, OA status, and treatment * OA interaction were included as fixed effects. The four combinations of treatment and OA were compared using Tukey’s method.
4. Discussion
We induce mild PTOA by creating an osteochondral fragment and subsequently treat the OA with a single administration of the biolubricant (pMPC) or saline (placebo control). The intra-articular administration of pMPC results in mild cartilage protective effects as indicated by GAG concentrations retained in cartilage; however, the pMPC induces a mild inflammatory state with increases in clinical, synovial, and radiographic scores. While the majority of the experimentally measured parameters are not statistically significant between treatment groups and all parameters are mild, with results being in the bottom quartile, there are some differences between the pMPC and saline-treated joints that should be highlighted. The hypothesis that pMPC decreases physical, radiographic, and biochemical effects in an equine PTOA model is rejected, while gross and histological effects may be interpreted as potentially chondroprotective.
Lameness scores are mild (grade 1 ± 0.5, AAEP scale) for limbs with induced OA. While lameness scores are not significantly different between treatments, lameness scores for OA–pMPC limbs are consistently higher throughout the postoperative study period for both subjective and objective analyses. Vector sum values are above the threshold for detectable clinical lameness for OA–pMPC limbs for the majority of the postoperative period. It should be noted that sham–saline control limbs have elevated lameness scores on days 42 to 63, which could skew OA–saline lameness scores. Two horses in this study group demonstrated elevations in lameness scores on those limbs (evident on the subjective exam). However, the reasons for these lamenesses are unknown. Exclusion of these horses from lameness statistical analysis is considered; however, given the lack of statistical significance, remain in the study. Similar to lameness scores, middle carpal joint effusion and carpal flexion scores remain persistently elevated throughout the postoperative period for OA–pMPC joints. This is not to state that OA–pMPC-treated joints continue to worsen; rather, they do not improve as quickly as OA–saline joints. This is demonstrated by OA–saline joints returning to lameness that is not significantly different from the sham-operated limb by day 21 and day 28 for flexion scores and day 56 for effusion scores. Whereas, OA–pPMC limbs remain persistently elevated through day 28 for lameness scores and for nearly the entire duration of the study for effusion and flexion scores. In sham-operated joints (all received saline), lameness, effusion, and flexion scores for pMPC- and saline-treated horses are similar. Other intra-articular biolubricants [sodium hyaluronan (HA) and polysulfated glycosaminoglycan (PSGAG)], previously assessed in the carpal chip model, demonstrated a slight improvement in lameness scores, effusion scores, and flexion scores from their baseline at 14 days following OA induction when administered three times [
28]. Similarly to OA–pMPC-treated joints in the current study, past intra-articular biolubricant treatments have not demonstrated statistically significant improvements over saline in lameness and flexion scores [
28]. Multiple PSGAG-treated joints have, however, exhibited significantly improved effusion scores [
28]. Hyaluronan sodium chondroitin sulfate and N-acetyl-D-glucosamine combination (PG) in a similar model demonstrate significantly improved average lameness scores for PG-treated joints versus the placebo when administered four times during the study. However, lameness scores for PG versus placebo-treated joints were not significantly different at the end of the study [
29]. Effusion scores remain persistently elevated in clinically normal joints with the use of other biolubricants, such as polyacrylamide hydrogels [
30]. This result is attributable to the polymer being incorporated into the synovial membrane [
30,
31]; in this study, there was no evidence of incorporation of the pMPC into the synovial membrane on histological examination.
Synovial fluid analysis partially corroborates a mild inflammatory response. Statistically significant differences between groups are present on days 28 and 42 for synovial total protein, PGE
2, and IL-1Ra. All synovial parameters peak on day 28, with treatment groups returning to baseline levels by day 70. On day 28, the PGE
2 levels in OA–pMPC joints are statistically significantly higher than in OA–saline joints (193.22 pg/mL versus 62.72 pg/mL). PGE
2 levels are known to be significantly elevated in the osteochondral chip model days 7 to 49 postoperatively [
33]. Interestingly, these levels are closer to 400 pg/mL for OA-induced joints and 200 pg/mL for joints without chip fracture, with sustained elevations in OA-induced joints [
33]. When comparing this result with the total nucleated cell count, the OA–pMPC induces a similar high-normal inflammatory response on the cellular level (800 cells/uL for OA–pMPC joints and 400 cells/uL for OA–saline joints). This degree of inflammatory response is still within reported normal TNCC following elective carpal arthroscopy, with prior reports indicating a range of 500–1250 cells/uL 28 days postoperatively [
33,
34]. In sham-operated joints, synovial total protein, PGE
2, and IL-1Ra outcomes are similar to levels present in healthy animals.
IL-1Ra levels markedly increased for the OA–pMPC-treated joints relative to the other groups. Specifically, IL-1Ra concentration spikes on day 28 and then decreases; PGE
2 levels decrease thereafter, presumably in response to IL-1Ra. In contrast, this peak in PGE
2 and IL-1Ra is not observed in sham-operated limbs. The pathway for upregulation of endogenous IL-1Ra levels is not clear. However, chondrocytes in humans produce IL-1Ra in response to IL-1β and IL-6 [
35,
36]. Previous reports demonstrate significant elevations in naturally occurring arthritis, with the highest elevations noted in joints with septic arthritis [
37]. The effect of biolubricants on the production of IL-1Ra in horses is not well documented. One study investigated the effects of HA on synovial fluid following arthroscopy for routine osteochondral fragment removal, demonstrating an increase in IL-1Ra at 48 h postoperatively. However, cases were not followed beyond this point [
38]. Specifically in this model of carpal PTOA, IL-1Ra elevations following injection of autologous conditioned serum (ACS) result in sustained protein production to approximately 70 pg/mL [
39]. In the current study, IL-1Ra levels spike, approaching 25,000 pg/mL. It is unknown why IL-1Ra levels increase so substantially, but it may be in response to an inflammatory state induced by pMPC.
Radiographic studies also reveal persistent mild joint inflammation by evidence of more significant osteophytosis, the likely result of persistent synovitis. Both OA–saline and OA–pMPC joints show radiographic evidence of progressive osteoarthritic change. While the majority of the parameters are not significantly different between OA–saline and OA–PMPC joints, osteophytosis scores are significantly higher in OA–pMPC joints, indicating increased inflammation in these joints.
Interestingly, pMPC treatment of the OA joint reduces synovial GAG release, as evident by the lower concentrations of GAG in the synovial fluid, which may elucidate a protective effect. The differences between groups are largest and statistically significant on day 28, with the release of GAG being the lowest for the pMPC treatment group. At the end of the study, histological GAG scores are higher in cartilage of pMPC-treated joints confirming higher concentrations of GAG remained in the cartilage (
Figure 5B), though levels are not significantly different. On gross pathology, total erosion scores for OA–pMPC joints are graded as less than half of the OA–saline counterparts; however, there is no statistical significance. It is possible that a type II statistical error yielded statistical results that are not significant when a true effect is present. The surface topography scores are higher for pMPC-treated joints, indicating higher peaks (increased fibrillation) and lower troughs (cartilage defects/erosions) suggestive of more advanced osteoarthritis. However, scores are not significantly different between saline and pMPC-treated joints. Surface topography, like histology, evaluates a few small portions of cartilage and therefore may not be representative of the entire joint [
32].
Previous work on tribological measurements of ex vivo cartilage plugs has demonstrated cartilage protective effects. In the ex vivo study, the pMPC network reduces the coefficient of friction by 73% in cartilage explants compared to saline [
9]. It dissipates shear forces at the cartilage interface, thereby reducing damage. Further, due to its network polymer architecture and high hydration due to the presence of phosphorylcholine groups, it also exhibits a “cushioning effect” at the articular surface, which is thought to aid in preventing cartilage damage [
9]. In horses, polyacrylamide hydrogels (PAAG) are proposed to function by aggregating on articular cartilage to create a mechanical barrier, resulting in a decrease in the coefficient of friction (COF) by 30–40% relative to saline-treated cartilage explants [
23,
25].
Histologic grades for synovium, cartilage, and subchondral bone are low and not significantly different between pMPC-treated and saline-treated joints. Gross inspection of the joint and histologic evaluations of synovium and cartilage at the day 70 end point show no evidence of pMPC remaining. While pMPC is resistant to degradation by hyaluronidases and remains in the joint for over 30 days in a murine model, a longer duration of action is not represented here [
22]. Another proposed mechanism of action for PAAG in horses is synovial incorporation by macrophages, which improves joint capsule elasticity [
23]. The pMPC used in this study does not directly target synovial incorporation as its mechanism of action, and therefore, the effects on synovium are unknown. The information garnered in this study suggests that pMPC exhibits a minimal effect on synovium.
It is unknown why the pMPC initiates an inflammatory response in the joints. In in vitro assays, pMPC over a concentration range of 1 to 100 mg/mL is non-cytotoxic to fibroblasts and chondrocytes over a 72 h incubation period, while at the highest concentration is cytotoxic to synoviocytes. This concentration is significantly higher than that used in the current in vivo study. However, the sustained duration of exposure to the pMPC is greater in vivo.