Non-terminal animal model of post-traumatic osteoarthritis induced by acute joint injury

Abstract

Objective

Develop a non-terminal animal model of acute joint injury that demonstrates clinical and morphological evidence of early post-traumatic osteoarthritis (PTOA).

Methods

An osteochondral (OC) fragment was created arthroscopically in one metacarpophalangeal (MCP) joint of 11 horses and the contralateral joint was sham operated. Eleven additional horses served as unoperated controls. Every 2 weeks, force plate analysis, flexion response, joint circumference, and synovial effusion scores were recorded. At weeks 0 and 16, radiographs (all horses) and arthroscopic videos (OC injured and sham joints) were graded. At week 16, synovium and cartilage biopsies were taken arthroscopically from OC injured and sham joints for histologic evaluation and the OC fragment was removed.

Results

Osteochondral fragments were successfully created and horses were free of clinical lameness after fragment removal. Forelimb gait asymmetry was observed at week 2 (P=0.0012), while joint circumference (P<0.0001) and effusion scores (P<0.0001) were increased in injured limbs compared to baseline from weeks 2 to 16. Positive flexion response of injured limbs was noted at multiple time points. Capsular enthesophytes were seen radiographically in injured limbs. Articular cartilage damage was demonstrated arthroscopically as mild wear-lines and histologically as superficial zone chondrocyte death accompanied by mild proliferation. Synovial hyperemia and fibrosis were present at the site of OC injury.

Conclusion

Acute OC injury to the MCP joint resulted in clinical, imaging, and histologic changes in cartilage and synovium characteristic of early PTOA. This model will be useful for defining biomarkers of early osteoarthritis and for monitoring response to therapy and surgery.

Introduction

Joint trauma related to sports or other types of injury can cause acute inflammation, cartilage damage, and a significant increase in the long-term risk for the development of osteoarthritis (OA). In the United States, approximately 12% of the overall prevalence of symptomatic OA in humans is attributable to post-traumatic OA (PTOA), which corresponds to approximately 5.6 million adults¹. Traumatic joint injury often occurs in younger adults, predisposing them to pain and disability from PTOA by middle-age².

Early therapeutic intervention would ideally slow or halt the progression of OA. However, the etiopathogenesis of PTOA is not well understood and the relationship between the severity of cartilage injury and risk of joint degeneration has not been defined^3–5. Because the time between injury and onset of PTOA is often lengthy and tissue samples are difficult to acquire, the pathological characteristics of disease progression are difficult to study in humans. A variety of animal models have been developed to study different types of acute joint injury^6–11. All currently available models are terminal studies with highly variable translational benefit for clinical and morphological understanding of the onset and progression of OA¹².

Osteochondral (OC) fragmentation of the proximal phalanx in the metacarpophalangeal (MCP) joint is a common injury in racehorses that arises from hyperextension of the MCP joint. The initial clinical symptoms include several days of lameness (pain) associated with synovial effusion. If untreated, OA will often develop in these animals because the OC fragment creates partial to full-thickness cartilage erosion on the opposing cartilage surface of the third metacarpal bone^{13, 14}. As such, creation of an OC fragment in the MCP joint may serve as a useful model in the study of PTOA. Furthermore, because removal of OC fragments in clinical cases will often result in a high percentage of return to racing^{14, 15}, it is conceivable that the model can be non-terminal by removing the OC fragment at the end of the study period. Creation of an OC injury in the equine carpal joint has been used as an OA model to test various therapeutics^{16, 17}, showing translational benefit to humans¹⁸. However, like other OA models, horses are euthanized at the end of the study due to severity of injury. Compared to other animals, horses are ideal for translational human OA studies because of their athleticism, ability to carefully control exercise, and the size of equine MCP joints facilitates arthroscopic evaluation, serial arthrocentesis, and adequate arthroscopic tissue sampling. Thus, the equine MCP joint is amenable to creation of a traumatic injury model of OA that allows for ample data collection, while not requiring euthanasia of the animal at the end of the study.

The purpose of this study was to develop a non-terminal animal model of acute joint injury that demonstrated clinical and morphological evidence of the onset and progression of PTOA. We hypothesized that creation of an OC injury in the MCP joint of horses would result in clinical, radiographic, arthroscopic, and histologic changes characteristic of early PTOA. This non-terminal injury model would be useful for defining biomarker changes of early OA and for monitoring responses to therapy and surgery.

Methods

Animals

Twenty-two clinically normal Quarter Horses that were free of lameness and had radiographically normal MCP joints were used. Horses were divided into 2 groups: (1) horses that had an OC fragment created arthroscopically on the proximal dorsomedial aspect of the first phalanx in one randomly chosen MCP joint and a sham arthroscopic operation in the contralateral joint (n = 11) at week 0 (baseline); and (2) unoperated control horses (n = 11). In the injured group, an incongruent surface was created by the OC fragment that subsequently caused a chondral lesion on the opposing third metacarpal medial condyle (lesion of interest). In a pilot study of this model by some of the authors (TNT, MPB, KAM), arthroscopic evaluation showed that chondral lesions were minimal or absent at 8 weeks, but were well established by 24 weeks (details in Supplementary materials; S-Fig. 1). Based on this pilot data, week 16 was chosen as the end-point of the current study. The OC fragment was therefore removed at week 16 to prevent further chondral damage. For assessment of parameters, the OC injured and sham MCP joints were assessed separately in each horse, when applicable, and both MCP joints were combined for the control group. Therefore, the OC injured, sham, and control joints were analyzed as 3 separate groups. Longitudinal changes within each group over time were compared to week 0, and all 3 groups were compared to each other over time.

The OC injured group included 6 females and 5 castrated males, with a mean ± SD age of 5.6 ± 1.6 years; control horses included 5 females and 6 castrated males, with a mean age of 6.4 ± 2.1 years. All horses were exercise restricted in stalls (4 m × 4 m) for 2 weeks, beginning at week 0, and were then turned out in small paddocks (60 m × 40 m) for the duration of the 16 week study. All horses were exercised on a high-speed treadmill 5 days/week from week 2 to week 16 (details in Supplementary materials). All procedures were approved by the University of Minnesota Institutional Animal Care and Use Committee.

Surgical Procedures

At week 0, an OC fragment was created arthroscopically on the axial aspect of the proximal dorsomedial eminence of the first phalanx in one randomly chosen MCP joint of the 11 OC injured horses (Fig. 1A and B). Using a 6 mm bone gouge, 3 cuts, approximately 3–4 mm deep, were made through cartilage and subchondral bone (Fig. 1C). This formed approximately 3/4 of the circumference of a circle, leaving the joint capsular attachment on the dorsal aspect intact. A 4 mm osteotome was placed under the medial cut and tapped with a mallet (medial to lateral, angled about 30° down from the joint surface to a depth of 3–4 mm) to dislodge the fragment from the underlying subchondral bone (Fig. 1D). The resulting fragment was repositioned back into the fracture bed before closure of the joint so that the subchondral fracture bed was not directly exposed to the opposing chondral surface, thus resembling a naturally occurring OC fragment (Fig. 1E). The entire dorsal pouch of the OC injured joint (prior to and after creation of the fragment) and contralateral (sham) joint was explored and the condition of the articular cartilage and synovial membrane digitally documented (Smith & Nephew, Andover, MA). Samples (2–5 mm in length) of synovial membrane were collected arthroscopically from the medial and lateral dorsal joint capsule of all OC injured joints and from only the medial side in 5 of 11 sham operated joints. All operated joints were lavaged to remove debris and were routinely closed. Morphine (45 mg) was injected into OC injured joints. Horses were stall-rested until suture removal at 14 days (post-operative care detailed in Supplementary materials). Control horses were stall rested for 14 days to mimic the OC injured horse protocol.

Figure 1.

Anatomy of the equine metacarpophalangeal (MCP) joint and the specific site for creation of osteochondral (OC) fragment. (A) Left MCP joint in a live horse, with the third metacarpal bone (top) and first phalanx (bottom) outlined. (B) Cadaver specimen of the MCP joint in the same orientation as (A). The joint has been opened and flexed to show the distal end of the third metacarpal bone and the proximal end of the first phalanx. The site of surgical creation of the OC fragment on the proximodorsal first phalanx and site of the chondral lesion that develops secondary to the OC fragment (and site of histologic biopsy) on the dorsodistal third metacarpal bone are outlined. (C–E) Arthroscopic images of OC fragment creation on the first phalanx in the same orientation as (A) and (B) with the third metacarpal bone visible above the first phalanx. (C) Cuts made with a 6 mm bone gouge through articular cartilage and subchondral bone. (D) Elevation of the OC fragment from the subchondral bone bed with 4 mm osteotome. (E) Appearance of the OC fragment at the end of surgery. For all images, medial is to the left and lateral to the right.

At week 16, OC injured horses underwent a second arthroscopy to remove the OC fragment. Both OC injured and sham joints were explored and digitally documented (Fig. 2A). Full-thickness articular cartilage samples from OC injured joints (n = 11) were obtained from the cartilage surface of the dorsal distal third metacarpal medial condyle that opposed the OC fragment (kissing lesion wear-line). A 2 mm osteotome was used to make 2 linear, full-thickness cuts parallel to the kissing lesion (2 mm apart). The osteotome was then advanced in a parallel fashion through the articular cartilage into subchondral bone in a proximal to distal direction and an approximately 5–6 mm long section of cartilage and bone was elevated (Fig. 2B). The sample was carefully removed to avoid crush injury. For comparison, cartilage samples were also obtained from the same location in the contralateral sham joints of 6 randomly selected horses. Synovial membrane samples were collected from the medial and lateral dorsal joint capsule of all OC injured joints and from the medial side of all sham joints. After sample collection, the OC fragment was removed arthroscopically. Joints were lavaged and routinely closed. Morphine (45 mg) was injected into OC injured joints and the 6 sham joints from which cartilage samples were taken. Post-operative care was conducted as described for the first surgery. All horses were evaluated for lameness 3–4 weeks post-operatively and subsequently adopted out according to University protocol.

Figure 2.

Arthroscopic images of OC injured joints at week 16 with the third metacarpal bone on top and the first phalanx on the bottom (medial is to the left, lateral to the right). (A) OC fragment on proximal first phalanx (outlined by short arrows) and kissing lesion wear-lines in the articular cartilage of distal medial third metacarpal bone (long arrows). (B) Elevation of full606 thickness articular cartilage sample from the third metacarpal bone using a 2 mm osteotome. The sample is centered over the articular cartilage wear-lines.

Clinical Assessment of the Joint

MCP joint circumference measurements (details in Supplementary materials), synovial fluid viscosity scores, and joint effusion scores were evaluated in both forelimbs in all horses at weeks 0, 1, 2, then every 2 weeks until week 16. Joint effusion was subjectively graded as follows: 0 = none; 1 = mild; 2 = moderate; or 3 = severe, based on visual appearance and palpable swelling of the joint pouches. With the limb in flexion, synovial fluid was obtained at each time point using a 20 g × 1.5 inch needle from the palmar pouch of each MCP joint without lavage (for future biomarker analysis) and viscosity was measured (details in Supplementary materials). The presence or absence of hemarthrosis at sample collection was noted.

Gait Analysis

Kinetic gait analysis has been used to objectively distinguish pain (lameness) in horses¹⁹. Forelimb symmetry was assessed on all 22 horses using kinetic analysis (Advanced Medical Technology, Inc., Watertown, MA) at baseline and every 2 weeks for the study duration (details in Supplementary materials). In addition, subjective lameness scores (AAEP grading scheme²⁰) and pain response to flexion of the distal limb were analyzed throughout the study, as well as out to 4 weeks after OC fragment removal (details in Supplementary materials).

Radiography

Bilateral MCP joint radiographs were taken on all 22 horses at baseline and at week 16 using digital radiography (Sound-Eklin, Carlsbad, CA). Further details of views obtained are listed in Supplementary material. Radiographic images of baseline and week 16 were blinded and evaluated independently by two evaluators (TNT, MPB; scores for each joint were averaged) using a previously reported OC injury scoring system of 0–30²¹.

Arthroscopy

Arthroscopic exploration videos of OC injured and sham operated MCP joints were blinded and evaluated independently (TNT, MPB; scores for each joint were averaged) using a modified scoring system for evaluation of articular changes in response to OC injury²¹. Briefly, the total arthroscopic score (0–44) encompassed cartilage changes (0–23), synovial membrane changes (0–15), and the presence/size of the fragment(s) (0–6) (S-Table 1). Kissing lesion depth (CDK) was a sub-category of cartilage damage that assessed the articular cartilage change on the third metacarpal medial condyle in the region directly opposite the OC fragment where the cartilage sample was taken for histology (Fig. 2). Control horses were not subjected to arthroscopic evaluation.

Histology

Cartilage

Immediately after removal from the joint at week 16, full-thickness articular cartilage samples of the third metacarpal medial condyle from OC injured joints (11/11) and sham joints (6/11) were oriented longitudinally and the proximal end stained lightly with India ink to allow proper orientation for embedding (processing details in Supplementary material). Transverse sections (4 µm) were made near the middle (from a proximal to distal orientation) of each sample using a rotary microtome (Leica, Buffalo Grove, IL) and stained with hematoxylin and eosin (H&E) or Safranin O/fast green (all stained at the same time). Slides were blinded and examined independently via light microscopy (CSC, MKB; scores were averaged). Cartilage samples were graded for fibrillation/fissuring of the articular surface, chondrocyte cluster formation, chondrocyte death, and decreased staining for matrix proteoglycan using modifications of the microscopic grading system for articular cartilage in horses²². Fibrillation/fissuring was scored 0–4 as previously described²². Cluster (complex chondrone) density, where cluster was defined as multiple chondrocytes sharing pericellular matrix within a chondrone²³, was determined by counting the number of clusters in each sample and dividing by the total cartilage area. The total cartilage area was defined as the area between the superficial surface of articular cartilage and the most superficial tide mark. Percentage of the cartilage occupied by dead chondrocytes was quantitatively evaluated using histomorphometry, by measuring the area occupied by dead chondrocytes divided by the total cartilage area and multiplying that value times 100 (details in Supplementary materials; S-Fig. 3). To assess the loss of matrix proteoglycan, the percentage of the articular cartilage area that was composed of matrix exhibiting complete loss of Safranin O staining was determined by dividing the area of stain loss by the total cartilage area.

Synovial membrane

Week 0 and week 16 synovial membrane samples were placed in 10% neutral-buffered formalin immediately after collection and were processed routinely. Sections (4 µm) were prepared and stained with H&E. Slides were blinded and subjectively graded independently (CSC, MKB; scores were averaged) to evaluate cellular infiltrate (0–4), vascularity (0–4), intimal hyperplasia (0–4), subintimal edema (0–4), and subintimal fibrosis (0–4) as previously described²².

OC fragment

Details in Supplementary materials; S-Fig. 4.

Statistical Analysis

Data was assessed for normality using the Shapiro-Wilk test for normality and outliers were identified in the gait analysis data using extreme studentized deviate tests and removed from further analyses (details in Gait Analysis section of Supplementary material). F or MCP joint assessment data and forelimb symmetry, a repeated measure ANOVA with Tukey’s for multiple comparison was used for analysis. To verify that assumptions were fulfilled for the repeated measures ANOVA, normality was examined graphically for skewness or kurtosis as well as with the Shapiro-Wilk test, within group variances were examined to look at the homogeneity of variance, and sphericity was examined using a Maulchly’s test and adjusted for using the Greenhouse-Geisser adjustment. Radiographic, arthroscopic, and synovial membrane histological scores were analyzed using a Kruskal-Wallis test with a Dunn’s Multiple Comparison Test. A Mann Whitney U-test was performed on histologic cartilage scores. Correlations were assessed using the Spearman rank correlation. P<0.05 was considered significant.

Results

Osteochondral fragments were successfully created with minimal to no collateral iatrogenic damage in 6 right and 5 left MCP joints. At week 16, all OC fragments were in place at arthroscopic evaluation, with histologic evidence of fibrocartilagenous healing to bone in 10 of 11 fragments (S-Fig. 4). Removal of OC fragments at week 16 resulted in minimal subjective clinical signs of lameness by one month after surgery (mean, 95% CI; 0.3 out of 5 AAEP grading system²⁰, −0.16, 0.71). There were no complications related to the OC fragment.

Clinical Assessment of the Joint

OC injured joints had higher effusion scores compared to baseline, sham, a nd control joints, whereas sham and control joints were not different from baseline or each other (Fig. 3). OC injured joints also had greater joint circumference changes from baseline at all time points except week 1 and this change was greater than sham and control joints at multiple time points (S-Fig. 5). Control and sham joints also demonstrated joint circumference changes from baseline, but there were no differences between sham and control joints. Synovial fluid viscosity decreased in OC injured (weeks 1 and 2) and sham joints (week 1) compared to baseline and controls (S-Fig. 6). Hemarthrosis was only present in 1 of 440 MCP synovial fluid collections.

Figure 3.

Clinical assessment of MCP joint effusion and kinetic gait analysis. ( A) Mean (± 95% CIs) effusion scores (0–3) over a 16 week period for each MCP joint group (OC injured, sham, control), and differences between each group at each time point (Tukey’s post hoc). (B) Mean (± 95% CIs) forelimb percent symmetry scores of peak vertical force in OC injured and control horses over a 16 week period compared to baseline and compared to each other. A decrease in symmetry indicates pain (lameness). Repeated measures P-values are listed. (A) Differences from baseline are represented as: OC injured MCP joints to baseline: +++P < 0.001. Differences between groups are represented as: OC injured MCP joint to sham MCP joint: ***P < 0.001, *P < 0.05; OC injured MCP joint to control MCP joint: ^^^P < 0.001, ^^P < 0.01, ^P < 0.05. (B) Differences from baseline are represented as: OC injured horses to baseline: ++P = 0.0012.

Gait Analysis

The percent symmetry of peak vertical force decreased at week 2 compared to baseline in OC injured horses, but there were no differences between OC injured and control horses at any time point (Fig. 3). Distal limb flexion resulted in an increased pain response in OC injured limbs at weeks 2, 6, 10, and 12 compared to baseline or sham joints, and at weeks 2 and 10 compared to control joints (S-Fig. 7).

Radiography

Mean total radiographic and enthesophyte scores demonstrated more change in OC injured joints (S-Fig. 8) versus baseline, sham, and control joints (Table 1). There were no differences for subchondral bone sclerosis/lucency or joint space narrowing. Joint effusion at week 16 correlated with the enthesophyte score (R=0.580; P=0.005) and total radiographic score (R=0.814; P<0.0001) in OC injured limbs at week 16.

Table 1.

Mean radiographic scores (± 95% CIs) from OC injured, sham, and control MCP joints at weeks 0 and 16.

Radiographic scores	Week 0			Week 16
	OC injured	Sham	Control	OC injured	Sham	Control
	(n = 11)	(n = 11)	(n = 22)	(n = 11)	(n = 11)	(n = 22)
Osteophyte (0–6)	0.18	0.18	0.39	0.77	0.27	0.70
	(−0.09, 0.45)	(−0.22, 0.59)	(0.04, 0.73)	(0.18, 1.36)	(−0.04, 0.59)	(0.27, 1.14)
Enthesophyte (0–6)	0.18	0.09	0.50	1.82***^^	0.18	0.73
	(−0.09, 0.45)	(−0.11, 0.29)	(0.14, 0.86)	(0.83, 2.81)	(−0.09, 0.45)	(0.36, 1.09)
Total Radiographic score (0–30)	0.50	0.45	1.09	7.59***^^^	0.82	1.71
	(0.08, 0.92)	(−0.03, 0.94)	(0.47, 1.72)	(5.77, 9.42)	(0.31, 1.32)	(0.93, 2.48)

Bolded values represent a significant increase in OC injured joints at week 16 compared to week 0 (P < 0.05). Week 16 differences between groups are represented by: OC injured to sham:

^***P < 0.001; OC injured to control:

^{^^^}P < 0.001,

^{^^}P < 0.01. n = number of MCP joints evaluated in each group. The potential score range is listed parenthetically next to each radiographic category in the first column.

Arthroscopy

Cartilage and total arthroscopy scores were higher at week 16 than baseline for the OC injured joints (Table 2). Synovial membrane, cartilage and total arthroscopy scores were higher in OC injured joints compared to sham joints at week 16 (Table 2). Arthroscopic scores correlated with multiple clinical and radiographic scores (Table 3).

Table 2.

Summary of mean arthroscopic and histologic data (± 95% CIs) obtained from OC injured and sham MCP joints at weeks 0 and 16.

	Week 0		Week 16
Arthroscopic scores	OC injured	Sham	OC injured	Sham
	(n = 11)	(n = 11)	(n = 11)	(n = 11)
Synovial membrane (0–15)	1.36 (0.84, 1.89)	0.95 (0.28, 1.63)	3.73 (2.16, 5.29)*	1.27 (0.59, 1.95)
Cartilage score (0–23)	1.68 (0.67, 2.69)	1.64 (0.36, 2.91)	10.23 (8.78, 11.67)***	1.77 (0.49, 3.05)
Total arthroscopy score (0–44)	3.05 (1.91, 4.18)	2.68 (1.15, 4.22)	18.14 (15.38, 20.89)***	3.14 (1.74, 4.53)
Histologic scores	OC injured	Sham	OC injured	Sham
Cartilage			(n = 11)	(n = 6)
Fibrillation/fissuring (0–4)			1.27 (0.96, 1.59)*	0.66 (0.12, 1.21)
Cluster formation (number/area)			8.14 (4.07, 12.21)**	1.33 (−0.89, 3.54)
Chondrocyte death (% area)			6.0 (4.1, 7.3)**	2.0 (0.8, 3.0)
Decreased matrix PG (% area)			15.0 (6.4, 23.0)	14.0 (0.5, 26.0)
Synovium (medial)	(n = 11)	(n = 5)	(n = 11)	(n = 5)
Cellular infiltrate (0–4)	0.95 (0.72, 1.19)	1.20 (0.64, 1.76)	0.95 (0.64, 1.27)	1.14 (0.98, 1.29)
Vascularity (0–4)	0.68 (0.41, 0.95)	0.50 (0.06, 0.94)	1.86 (1.28, 2.45)	1.00 (0.46, 1.54)
Intimal hyperplasia (0–4)	0.23 (−0.09, 0.54)	0.20 (−0.14, 0.54)	0.32 (0.15, 0.49)	0.14 (−0.08, 0.35)
Subintimal edema (0–4)	0.32 (0.09, 0.54)	0.60 (−0.42, 1.62)	0.77 (0.22, 1.32)	0.27 (−0.04, 0.59)
Subintimal fibrosis (0–4)	1.09 (0.80, 1.38)	0.80 (0.09, 1.51)	2.41 (1.87, 2.95)	1.36 (0.99, 1.73)

Bolded values represent a significant increase in OC injured joints at week 16 compared to week 0 (P < 0.05). Week 16 differences between OC injured and sham groups are represented by:

^***P < 0.001,

^**P < 0.01,

^*P < 0.05. n = number of MCP joints evaluated in each group. The potential score range or description is listed parenthetically next to each arthroscopic and histologic category in the first column.

Table 3.

Correlations [ρ - Rho values (P-values)] between arthroscopic scores and joint effusion, joint circumference percent change and radiographic scores in OC injured MCP joints at week 16.

	Arthroscopy
	Synovium	Cartilage	CDK	Total
Effusion	0.566 (0.006)	0.837 (<0.001)	0.750 (<0.001)	0.802 (<0.001)
JC % change	0.529 (0.011)	0.503 (0.017)	0.467 (0.028)	0.551 (0.008)
R-Enthes	0.297 (0.180)	0.563 (0.006)	0.621 (0.002)	0.459 (0.032)
R-Total	0.665 (0.001)	0.850 (<0.001)	0.815 (<0.001)	0.848 (<0.001)

Bolded ρ values represent statistically significant correlations. JC %=joint circumference percent change from baseline, R=radiograph, Enthes=enthesophyte, CDK=kissing lesion (region from which cartilage sample taken).

Histology

Cartilage

Fibrillation/fissuring, chondrocyte cluster formation, and chondrocyte death scores on the third metacarpal medial condyle were higher in OC injured versus sham joints at week 16 (Table 2; Fig. 4A–B). Loss of matrix proteoglycan staining (Fig. 4C–D) was not different between injured and sham cartilage (P=0.760). Chondrocyte death and cluster formation correlated with multiple clinical, radiographic and arthroscopic scores (Table 4). In addition, chondrocyte clusters correlated with chondrocyte death (R=0.776, P<0.0001), while there was no correlation between fibrillation/fissuring and chondrocyte death (R=0.352, P=0.165).

Figure 4.

Histologic sections of articular cartilage and synovial membrane samples from sham and OC injured MCP joints. H&E staining of cartilage from a sham joint (A) and an OC injured joint (B) at week 16. Sham MCP joint cartilage (A) demonstrates a smooth articular surface and nuclear staining of chondrocytes throughout the section. The OC injured MCP joint cartilage (B) demonstrates superficial fibrillation (black arrows), chondrocyte clusters (red arrow), and areas of loss of nuclear staining (cell death; arrow heads) in the superficial zone of the injured joint cartilage. Safranin O stained sections of articular cartilage from a sham joint (C) and an OC injured joint (D) at week 16 demonstrate superficial loss of Safranin O staining in the sham joint (C) and a slightly more extensive loss of staining (not statistically significant), accompanied by superficial fibrillation, in the injured joint (D). H&E-stained sections of synovial membrane from a MCP joint prior to creation of the OC fragment at week 0 (E) and from an OC injured joint at week 16 (F), demonstrating the presence of subintimal fibrosis (black arrows) in the synovium of the injured joint. (A–F) Bar = 100 µm.

Table 4.

Correlations [ρ - Rho values (P values)] of histologic scores with joint effusion, radiographic, and arthroscopic scores in OC injured MCP joints at week 16.

	Histology Cartilage				Histology Synovium
	Fib/fis	Clusters	Chond.	PG loss	Cell	Vascularity	Hyperplasia	Edema	Fibrosis
			death		infiltrate
Effusion	0.167	0.747	0.707	0.240	−0.342	0.417	0.152	0.206	0.689
	(0.522)	(0.001)	(0.001)	(0.353)	(0.119)	(0.054)	(0.499)	(0.359)	(<0.001)
R-Enthes	0.084	0.428	0.489	−0.064	−0.224	0.125	0.073	0.146	0.409
	(0.750)	(0.087)	(0.046)	(0.807)	(0.316)	(0.581)	(0.747)	(0.517)	(0.059)
R-Total	0.363	0.495	0.669	−0.027	−0.306	0.332	0.431	0.147	0.641
	(0.152)	(0.043)	(0.003)	(0.920)	(0.167)	(0.131)	(0.045)	(0.514)	(0.001)
A-Synovium	0.348	0.357	0.621	0.047	−0.207	0.658	0.306	0.336	0.568
	(0.171)	(0.160)	(0.008)	(0.859)	(0.354)	(0.001)	(0.166)	(0.126)	(0.006)
A-CDK	0.420	0.717	0.942	0.118	−0.202	0.516	0.216	0.290	0.754
	(0.093)	(0.001)	(<0.001)	(0.652)	(0.367)	(0.014)	(0.335)	(0.190)	(<0.001)
A-Cartilage	0.351	0.828	0.897	0.190	−0.222	0.347	0.265	0.106	0.633
	(0.167)	(<0.001)	(<0.001)	(0.465)	(0.320)	(0.113)	(0.265)	(0.638)	(0.002)
A-Total	0.397	0.664	0.825	0.139	−0.240	0.458	0.346	0.160	0.627
	(0.115)	(0.004)	(<0.001)	(0.594)	(0.282)	(0.032)	(0.115)	(0.478)	(0.002)

Bolded ρ values represent statistically significant correlations. Fib/fis=Fibrillation/fissuring, Chond.=chondrocyte, PG=proteoglycan, R=radiograph, Enthes=enthesophyte, A=arthroscopy, CDK=kissing lesion (region from which cartilage sample taken).

Synovial membrane

Synovial vascularity and subintimal fibrosis scores were higher in the medial synovial membrane from OC injured joints at week 16 compared to baseline (Fig. 4E–F); however, there were no differences in synovial scores between OC injured and sham joints (Table 2). Medial and lateral synovial scores for OC injured joints were not different from each other at baseline or week 16. The subintimal fibrosis score correlated with clinical, r adiographic and arthroscopic scores, and vascularity correlated with arthroscopy scores (Table 4). In addition, synovial vascularity correlated with subintimal fibrosis (R=0.764, P<0.0001), and subintimal fibrosis correlated with chondrocyte death (R=0.770, P<0.0001).

OC fragment

Osteochondral fragments were composed primarily of an admixture of woven and lamellar bone (some areas of which were necrotic), fibrocartilage, and degenerate hyaline cartilage, and were accompanied by reactive/fibrotic synovium (S-Fig. 4).

Discussion

Osteochondral fragments were created arthroscopically in all MCP joints with minimal iatrogenic collateral damage. Use of a bone gouge facilitated creation of semi-circular fragments of equal size in the same location of each horse that resulted in an incongruent articular surface. This incongruence was maintained throughout the study even though most fragments (10/11) had histologic evidence of fibrocartilagenous healing to the subchondral bone, similar to a previous OC injury model in the horse where 9 of 12 fragments healed back to underlying bone²⁴. The articular incongruity caused cumulative increased contact stress at a focal site opposite the fragment (kissing lesion) that led to cartilage degeneration over time (wear-lines). Therefore, our model created a focal superficial cartilage lesion with minimal variability, which may be more advantageous than a joint instability model of PTOA (such as ACL rupture) where articular surface contact stress increases at variable sites creating variability in lesion morphology¹⁰.

Arthroscopic surgery at week 16 allowed collection of small, but adequate cartilage and synovial biopsies as well as removal of the OC fragment to provide relief from the incongruity. Collection of clinical and histologic data and removal of the inciting cause of injury through minimally invasive techniques allowed the horses to be free of clinical signs of lameness (0.3 out of 5 subjective lameness grade) by one month after OC fragment removal. To our knowledge, this is the first non-terminal joint injury model that provides the possibility for evaluating the joint environment immediately after injury as well as after surgical removal of the inciting cause. Arthroscopically, wear-lines on cartilage were identified on the articular surface of the third metacarpal bone opposing the OC fragment and were positively correlated with chondrocyte death and cluster formation. These superficial articular cartilage lesions appeared to be similar to those occurring in the human patella²⁵, in which wear-lines may represent structural failure of the cartilage surface. This affects the biomechanical properties of cartilage and may represent an early stage of degeneration²⁵. In our pilot study (S-Fig. 1), minimal to no wear-lines could be identified arthroscopically in the 3 horses examined at 8 weeks, whereas multiple wear-lines were present in all horses by 24 weeks. In the current study, the wear-lines identified arthroscopically demonstrated multiple histologic changes, including mild fibrillation, chondrocyte death and the presence of chondrocyte clusters in the superficial zone. When combining the arthroscopic findings of our pilot study with the arthroscopic and histologic findings of our larger current study, it is suggestive that the cartilage lesions identified at week 16 in our model represent an early stage of PTOA since there is progressive change in the development of the wear-lines over time. This is further supported by the presence of chondrocyte death and cluster formation around the wear-lines at week 16, which represent some of the earliest signs of cartilage damage. A study of mechanical trauma on cartilage showed that chondrocyte apoptosis increased with increasing load⁷, while another study suggested that chondrocyte cluster formation may represent early cartilage repair since the number of proliferating chondrocytes increased with progression of OA²⁶. In addition, i n other models of OA^{10, 27, 28}, chondrocyte clusters tend to localize around areas in which cell death is most pronounced. Our histological evaluation supported this finding (Fig. 4) and showed a positive correlation between cell death and the presence of chondrocyte clusters in OC injured cartilage. In most OA studies, glycosaminoglycan (GAG) release increases from the superficial zone in response to mechanical trauma⁷. However, we did not see any difference in Safranin O staining between the OC injured and sham cartilage sections at week 16. Our findings are similar to the canine groove model, another OA model of subtle chondral injury, where there was no difference in Safranin O staining even though there was histologic evidence of structural and cellular differences²⁹. They demonstrated that subtle changes in PG content could not be detected by histochemistry even though there was biochemical evidence of PG release. In general, it appears that evaluating PG loss using Safranin O/fast green uptake is subjective and unreliable unless there is complete PG loss, suggesting that this stain may not be optimal for assessing PG loss with superficial cartilage lesions²². Safranin O staining may be best when there are substantial abnormal mechanical forces generated by the injury where chondrocyte death/proliferation reaches the middle to deeper cartilage zones resulting in greater PG loss than was identified in our model.

Unlike dogs and smaller animals that may ambulate on three limbs after injury, horses inherently bear weight on all 4 limbs unless the pain is severe. This makes the horse useful for objective measurements of low grade musculoskeletal pain because it ensures that the injured limb will remain in use throughout the duration of the study. In horses, any decrease in peak vertical force likely results from increased pain in that limb¹⁹. OC injury in our model caused transient pain that could be identified using ground reaction forces and subjective response to joint flexion. These transient signs of pain are consistent with that often seen after acute joint trauma in humans^{30, 31}, where there is initial pain from intra-articular hypertension, subchondral bone trauma, or soft tissue injury that subsequently subsides until PTOA becomes advanced³. We could not determine the specific source of transient joint pain in our horses after OC injury, but possibilities include synovial/capsular inflammation, subchondral bone trauma around the fragment, and/or joint effusion. There was evidence of both joint effusion and joint circumference changes in OC injured joints throughout the study, which may have contributed to transient increased flexion responses from baseline between weeks 2 to 16 (S-Fig. 7), even though lameness decreased over time (Fig. 3). With PTOA, enthesophyte formation and capsular fibrosis can be potential sources of pain, especially with forced flexion. These were both identified at week 16 in our study, but by the design of the study, we were unable to identify whether they developed earlier such that they could explain the pain response to flexion at weeks 6, 10, and 12.

Joint effusion was one of our best clinical correlates to the presence and severity of disease. The exact cause for the effusion in our model is unknown, but we know that arthroscopic exploration itself did not cause effusion because there was no increase in effusion scores in sham operated joints. In the acute period after injury, joint effusion may be caused by intra-articular bleeding, increased synovial fluid production, or inflammatory cell infiltration secondary to trauma^{3, 32–34}. In our study, hemorrhage was present at the time of OC fragment creation because we fractured healthy subchondral bone. However, there was no evidence of blood within the joint during synovial fluid collections at week 1 in any horse. In fact, hemarthrosis was noted only once thereafter, indicating that hemorrhage was not a continued source of joint effusion or pain.

In humans with knee OA, joint effusion can be a common finding even with minimal to no clinical signs of inflammation³⁵. In most animal models of OA, or PTOA in humans, inflammation and effusion are often present due to microscopic particles (joint debris) that incite acute synovitis via macrophagic phagocytosis^{36, 37}. In our study, most debris resulting from OC fragment creation was lavaged from the injured joints at the end of the procedure. However, it is possible that intermittent signs of effusion and pain may have corresponded with the release of cartilage debris from the score lines identified via arthroscopy and histology at week 16, but we have no direct evidence to support this possibility since there was minimal histologic evidence of debris within the synovium at week 16. Increased effusion may have been present due to daily treadmill exercise after acute injury³⁸, which would suggest that there is at least transient inflammation in the joint that may wax and wane after injury. Even though our study cannot document histologic evidence of synovitis at week 16, the clinical signs of effusion and pain (acute lameness by week 2 and flexion responses at weeks 2, 6, 10 and 12) supports that inflammation may come and go in the early stages of PTOA. In fact, our arthroscopic examination of horses in our pilot study (S-Fig. 1) demonstrated that there was slight hyperemia by 8 weeks after injury that progressed to hypertrophy and hyperplasia of the synovial membrane by week 24. Ultimately, analysis of biomarkers of inflammation in the synovial fluid is required to characterize the inflammatory process after injury. Regardless, similar to previous studies^{39, 40}, we found strong correlations between effusion and arthroscopic synovium and cartilage scores as well as effusion and histologic cartilage scores at week 16, suggesting that synoviopathy and cartilage degradation play a role in excessive fluid production and retention. It is also possible that intimal cells of fibrotic villi have decreased ability to filter synovial fluid, resulting in increased effusion. Our study did demonstrate a non-inflammatory fibrotic reaction in the synovium that was confined to the area of the OC fragment at week 16 and was correlated to histologic findings of increased vascularity and fibrosis in the injured joints. This non369 inflammatory synovial response would be similar to other surgically induced OA models as well as early knee OA in humans,where arthroscopy demonstrates synovial hypertrophy and hyperemia^39–44.

Histological evaluation of cartilage from control joints may have illustrated further differences from injured joints, beyond those seen in sham joints, and thus was a limitation of the study. However, histologic changes in sham joints were minimal and no significant differences were seen in clinical or radiographic scores between sham and control joints, other than viscosity scores at week 1. Therefore, the sham joint may serve as a control for the OC injured joint, but further biomarker analysis of the fluid from these horses will help support or refute this premise. Although minimal radiographic changes were present, additional advanced imaging such as MRI may have been useful to identify additional changes in soft tissues or subchondral bone not typically seen with radiography^{45, 46}. Although MRI was not used in our study, both standing and recumbent magnets can be used for longitudinal examination of this region in the horse and should be considered for future studies. However, we were able to use arthroscopic imaging which is often used as the gold standard for identification and staging of articular cartilage disease^{47, 48} since it allows better assessment of the cartilage surface and synovial membrane than MRI, including characterization of the location and severity of lesions^{49, 50}. The large size of the animals used in this study was advantageous for arthroscopic and histological data collection, as well as for synovial fluid collection (average 4.3 mL/MCP joint/sample). However, appropriate facilities and personnel to handle horses are required, and the cost, although similar to canine models, is greater than for laboratory animals. Although we were not able to follow these horses out to 12 – 24 months after fragment removal to determine their long-term health, previous clinical studies where similar naturally-occurring OC fragments have been removed in equine athletes have shown a high percentage of return to use^13–15.

The lesions identified in the current study as well as those identified in our pilot study do not fit the historical definition of OA⁵¹. However, the historical definition arguably fits the more advanced disease state where there is substantial structural damage to multiple tissues leading to persistent pain. Although early OA is much more difficult to identify, it is critical to do so because there may be potential in the early stages of OA to halt or slow the progression of the disease⁵². In a recent study to classify patients with early OA of the knee⁵³, it was suggested that patients could be defined as having early OA if they had episodes of pain, minimal to no radiographic changes (K–L grade 0 to II), and had either cartilage lesions identified by arthroscopy or cartilage/bone/soft tissue lesions identified on MRI. Based on this description, our horses would be classified as having early OA because they demonstrated variable pain throughout the 16 weeks based on force plate and subjective flexion responses, had minimal radiographic changes and demonstrated cartilage lesions arthroscopically. In addition, based on arthroscopic changes identified at weeks 8 and 24 in our pilot study (S-Fig. 1), we have evidence that the synovial and chondral changes progress over time. It is unlikely that these lesions would be self-limiting and heal spontaneously in the presence of the OC fragment in the joint. In fact, the progressive changes identified in the articular cartilage from our pilot study are similar to those described in other models of osteoarthritis, such as the canine groove model²⁹. Therefore, we feel that creation of an OC fragment in the equine MCP joint consistently resulted in acute traumatic injury with subsequent development of superficial articular changes that are representative of early PTOA.

In summary, creation of an OC fragment in the equine MCP joint resulted in consistent clinical, morphological, and histological features of acute traumatic injury with subsequent superficial articular changes. This model provides a suitable non-terminal model of PTOA that can be used to evaluate the clinical, morphological, and biochemical changes associated with early PTOA. This model will be useful for defining biomarkers of early OA and for monitoring response to therapy and surgery.