Materials and Methods

Twelve skeletally mature horses were included in the study. These horses (8 Standardbreds, 2 Thoroughbreds, 1 Quarter Horse, and 1 mixed breed; 8 mares and 4 geldings) ranged from 3 to 10 years in age (mean ± SD, 5.3 ±2.1 years; median, 5 years) and from 382 to 558 kg in body weight (mean, 466 ± 59 kg; median, 475 kg). Horses were evaluated for evidence of musculoskeletal disease via detailed lameness examination by 2 examiners (EJR and AMC), radiographic evaluation of the metacarpophalangeal joints, and cytologic analysis of synovial fluid from the metacarpophalangeal joints. Horses were allocated randomly to study endpoints at 1 month (n = 3 horses), 4 months (4), 8 months (3), or 10 months (2). This study was approved by the University of Guelph Animal Care Committee, in accordance with Canadian Council of Animal Care guidelines.

Operative technique—In a pilot study¹⁵ that used cadaveric equine forelimbs severed at the distal end of the radius, the reported arthroscopic approach to the palmar aspect of the metacarpal condyle was modified for application of the impactor device. Under arthroscopic guidance, a sharp trochar placed through the instrument portal was used to make a mark on the cartilage at the proposed site of impact injury. Limbs were subsequently dissected to confirm appropriate placement of the cartilage injury.

During prior research at our laboratory in which fresh cadaveric distal metacarpal condyles were impacted with pressures of 30, 50, 70, 80, and 90 MPa via a 6.5-mm-diameter impactor tip mounted in a drop tower, we determined that an ideal impact pressure of 80 MPa was required to induce chondrocyte death extending deeper than the superficial zone without macroscopic cartilage damage. Impact pressures of 30 to 70 MPa induced chondrocyte death only in the superficial zone, whereas a pressure of 90 MPa induced structural damage to the cartilage as well as chondrocyte death extending to the deep zone. Evaluation of chondrocyte death was performed via paravital staining of cartilage.

In the present study, each horse received preoperative antimicrobial (penicillin G sodium [22,000 U/kg, IV]) and anti-inflammatory (phenylbutazone [4.4 mg/kg, IV]) treatment and was placed under general anesthesia as part of another study.¹⁶ The horse was positioned in lateral recumbency with the control limb placed uppermost, as determined by randomization. The treatment limb was always operated on first. Each metacarpophalangeal region was prepared and draped for aseptic surgery. An arthroscopic portal was created in the medial palmaroproximal aspect of the metacarpophalangeal joint. Arthroscopic evaluation of the palmar pouch was performed to confirm the absence of clinically important cartilage abnormalities. A needle was introduced into the joint distal to the abaxial border of the medial proximal sesamoid bone to determine the site for the instrument portal, which was placed in a routine manner.

A custom-designed aiming device (Figure 1) was clamped into the medial and lateral epicondylar fossae to guide the placement of the impactor tip. The aiming device consisted of a modified condylar clamp, which was used to secure the device to the limb via attachment to the epicondylar fossae, and a jig with adjustable rotation and medial to lateral location, which could be locked so that the impactor tip would be stabilized when placed through the hole in the jig. A handheld spring-driven impactor with a 6.5-mm-diameter nonporous tip¹⁴ was placed through the aiming device. Four impact injuries of 80 MPa each were created in a cloverleaf pattern at the midpoint of the palmarodistal aspect of the medial metacarpal condyle as close as possible to the site of naturally occurring disease, which required full extension of the metacarpophalangeal joint. Arthroscopic guidance was used to confirm contact with the articular surface, then the metacarpophalangeal joint was maximally extended as the impactor was slid along the articular surface, which caused movement of the impactor distally. The impactor tip was positioned perpendicular to the articular surface via fluoroscopic guidance to visualize the final impactor placement site (Figure 2). Bupivacaine (0.02 mg/kg) and morphine (0.1 mg/kg) were injected into the joint prior to closure of the skin with No. 0 polypropylene suture in a simple interrupted or cruciate pattern. A sham procedure was performed on the control limb in the same manner, including introduction of the impactor tip into the joint, but without creating an impact injury. The distal aspects of both forelimbs were bandaged prior to unassisted recovery from anesthesia.

Photograph of a custom-designed aiming device consisting of an adjustable jig (black arrows) mounted on a modified condylar clamp (black arrowheads) and an impactor tip (white arrows) in place for making an impact injury on the palmaromedial aspect of the metacarpal condyle of a horse. The arthroscopic cannula (white arrowhead) is in place in the proximal palmar pouch of the metacarpophalangeal joint. Note that the body of the spring-driven impaction device must be attached to the impactor tip to create the impact injury. The horse is positioned in lateral recumbency with the medial side of the limb oriented up for surgery; the proximal portion of the limb is at the top of the image, and the foot is at the bottom of the image. The dorsomedial aspect of the limb is to the left of the image, and the palmarolateral aspect is to the right.

Citation: American Journal of Veterinary Research 73, 10; 10.2460/ajvr.73.10.1540

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Photograph of a custom-designed aiming device consisting of an adjustable jig (black arrows) mounted on a modified condylar clamp (black arrowheads) and an impactor tip (white arrows) in place for making an impact injury on the palmaromedial aspect of the metacarpal condyle of a horse. The arthroscopic cannula (white arrowhead) is in place in the proximal palmar pouch of the metacarpophalangeal joint. Note that the body of the spring-driven impaction device must be attached to the impactor tip to create the impact injury. The horse is positioned in lateral recumbency with the medial side of the limb oriented up for surgery; the proximal portion of the limb is at the top of the image, and the foot is at the bottom of the image. The dorsomedial aspect of the limb is to the left of the image, and the palmarolateral aspect is to the right.

Citation: American Journal of Veterinary Research 73, 10; 10.2460/ajvr.73.10.1540

• Download Figure
• Download figure as PowerPoint slide

Photograph of a custom-designed aiming device consisting of an adjustable jig (black arrows) mounted on a modified condylar clamp (black arrowheads) and an impactor tip (white arrows) in place for making an impact injury on the palmaromedial aspect of the metacarpal condyle of a horse. The arthroscopic cannula (white arrowhead) is in place in the proximal palmar pouch of the metacarpophalangeal joint. Note that the body of the spring-driven impaction device must be attached to the impactor tip to create the impact injury. The horse is positioned in lateral recumbency with the medial side of the limb oriented up for surgery; the proximal portion of the limb is at the top of the image, and the foot is at the bottom of the image. The dorsomedial aspect of the limb is to the left of the image, and the palmarolateral aspect is to the right.

Citation: American Journal of Veterinary Research 73, 10; 10.2460/ajvr.73.10.1540

• Download Figure
• Download figure as PowerPoint slide

Fluoroscopic image of the metacarpophalangeal joint of a horse during surgery, with the impactor tip (black arrow) in place within the custom-designed aiming device (black arrowhead) illustrated in Figure 1. The impactor tip is placed perpendicular to the articular surface of the palmar aspect of the metacarpal condyle, and the joint is moved into full extension prior to creation of an impact injury. The arthroscope (white arrow) is also in place in the palmar pouch of the joint.

Citation: American Journal of Veterinary Research 73, 10; 10.2460/ajvr.73.10.1540

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Fluoroscopic image of the metacarpophalangeal joint of a horse during surgery, with the impactor tip (black arrow) in place within the custom-designed aiming device (black arrowhead) illustrated in Figure 1. The impactor tip is placed perpendicular to the articular surface of the palmar aspect of the metacarpal condyle, and the joint is moved into full extension prior to creation of an impact injury. The arthroscope (white arrow) is also in place in the palmar pouch of the joint.

Citation: American Journal of Veterinary Research 73, 10; 10.2460/ajvr.73.10.1540

• Download Figure
• Download figure as PowerPoint slide

Fluoroscopic image of the metacarpophalangeal joint of a horse during surgery, with the impactor tip (black arrow) in place within the custom-designed aiming device (black arrowhead) illustrated in Figure 1. The impactor tip is placed perpendicular to the articular surface of the palmar aspect of the metacarpal condyle, and the joint is moved into full extension prior to creation of an impact injury. The arthroscope (white arrow) is also in place in the palmar pouch of the joint.

Citation: American Journal of Veterinary Research 73, 10; 10.2460/ajvr.73.10.1540

• Download Figure
• Download figure as PowerPoint slide

Postoperative care—Sodium penicillin (22,000 U/kg, IV, q 8 h) and phenylbutazone (4.4 mg/kg, IV or PO, q 24 h) were administered for 3 and 5 days, respectively. Additional analgesia (butorphanol [0.05 to 0.1 mg/kg, IV]) was available if required for moderate to severe lameness at a walk. The distal portions of the limbs were kept bandaged for 10 days, and sutures were removed 12 to 14 days after surgery. Horses were confined to stall rest for 14 days, prior to pasture turnout. Forced exercise (lunging in both directions on a dirt or grass surface) was initiated 2 weeks after surgery, starting at 5 minutes daily for 5 d/wk. During each subsequent week, daily exercise was increased by 5 minutes, to a maximum of 30 minutes/d. Forced exercise was discontinued 5.5 to 6 months after surgery due to constraints of weather conditions. Horses were randomly allocated to groups with endpoints at 1, 4, 8, and 10 months after impact injury, at which time they were euthanized (xylazine [300 mg, IV], followed by sodium pentobarbital [40.8 g, IV]).

Clinical assessment—Lameness examinations by 2 examiners (EJR and AMC or alternate), who were unaware of which limb was the treatment limb, were performed weekly for 5 months and then every other week for the remainder of the study. While the horse was at a trot in a straight line, lameness was graded as follows: 0 = normal, 1 = mild head nod at the trot, 2 = moderate head nod at the trot, 3 = obvious head nod with every step, 4 = partial weight bearing, and 5 = non-weight bearing. Half grades were assigned if indicated. Lameness was not graded at a walk. This grading scale was developed to provide more descriptive degrees of lameness at a trot in a straight line than the American Association of Equine Practitioners' lameness scale.¹⁷ For example, grades 1, 2, and 3 of the scale used in the present study all described lameness visible at a trot in a straight line, so they would all be considered a grade 3 on the American Association of Equine Practitioners' scale.

Radiographic examinations performed prior to entering the study and immediately prior to euthanasia consisted of dorsopalmar, lateromedial, flexed lateromedial, dorsomedial-palmarolateral oblique, dorsolateral-palmaromedial oblique, and 125° dorsodistal palmaroproximal (Hornof) projections of both metacarpophalangeal joints. Radiographs were evaluated by a board-certified radiologist, who was unaware of group assignments, and were scored for evidence of osteoarthritis (0 = normal, 1 = minimal, 2 = mild, 3 = moderate, and 4 = severe). This was a subjective assessment based on joint space, subchondral sclerosis, and size of osteophytes. Additional abnormalities, such as osteochondral fragments and proximal sesamoid bone sclerosis, were noted.

Synovial fluid analysis—Synovial fluid was collected from both metacarpophalangeal joints of each horse prior to entering the study and at 1, 2, 3, 4, 6, 8, and 10 months. Synovial fluid analysis, consisting of total protein concentration, WBC count, and differential leukocyte count, was performed immediately, and aliquots were frozen for later analysis of sGAG and COMP concentrations.

Synovial fluid sGAG—Synovial fluid samples were analyzed for sGAG concentration by the 1,9-dimethylmethylene blue dye assay as described.¹⁴ Results were expressed as a mean of 3 readings and reported as chondroitin sulfate C concentration (μg/μL of synovial fluid).

Synovial fluid COMP—A commercially available enzyme immunoassay kit^a designed for human synovial fluid and serum was used to measure COMP concentration in previously frozen synovial fluid aliquots. Several studies^18,19 have evaluated COMP concentrations in normal and osteoarthritic synovial fluid, and due to the highly conserved nature of this protein, the antibody in this kit is likely to cross-react with equine COMP. The assay was performed on all samples from injured joints at all time points; on all control samples at baseline (n = 12 horses), 6 months (5), and 8 months (5); and on 4 samples from control joints of randomly selected horses at all other time points.

Postmortem evaluations—Postmortem evaluation consisted of assessment of macroscopic appearance of the joint as well as assessment of the cartilage and bone after sectioning. The metacarpophalangeal joints were macroscopically scored (0 = normal, 1 = mild, 2 = moderate, and 3 = severe) for effusion, subcutaneous thickening, joint capsule fibrosis, enlargement of the entire joint, inflammation, cartilage color, cartilage translucency, cartilage surface roughness, osteophytes, synovial membrane hypertrophy, and synovial membrane darkening (0 = white, 1 = amber, 2 = orange, and 3 = red). Effusion was determined on the basis of ballottement of fluid between the fingers placed on the medial and lateral aspects of the palmar joint pouches, and degree was determined on the basis of clinical experience. Subcutaneous thickening was defined as increased soft tissue thickness directly under the skin, whereas joint capsule fibrosis referred only to increased soft tissue thickness at the site of the joint capsule. Enlargement of the entire joint indicated an increased outer diameter of the skin in the fetlock region, regardless of the site of thickening. Inflammation was defined as visible evidence of irritation of the synovial membrane (ie, hyperemia). An estimate of the percentage of abnormal cartilage area was performed in each region, including the medial metacarpal condyle, lateral metacarpal condyle, sagittal ridge, sagittal groove, medial aspect of the proximal phalanx, lateral aspect of the proximal phalanx, medial proximal sesamoid bone, and lateral proximal sesamoid bone. Abnormalities were drawn on a diagram of the joint, and digital photodocumentation was obtained for all joints prior to and following India ink staining.

The joint surfaces were stained with India ink to delineate areas of disruption in the cartilage surface, and these areas were traced onto acetate sheets and then scanned into image analysis software,^b as described.²⁰ After manual calibration and thresholding, this software was used to measure the amount of damaged cartilage with India ink uptake, compared with the total amount of cartilage in each of the following regions: medial metacarpal condyle, lateral metacarpal condyle, medial aspect of the proximal phalanx, lateral aspect of the proximal phalanx, medial proximal sesamoid bone, and lateral proximal sesamoid bone. Results were expressed as the percentage of damaged area in each region.

Cartilage assessment—Cartilage samples, including the entire thickness of hyaline cartilage except for the calcified cartilage, were collected with a No. 15 scalpel blade from the lesion site or a corresponding location in control joints and from a site distant to the lesion on the palmar aspect of the medial metacarpal condyle. These cartilage samples were used for paravital staining, MTT cell viability assay, and sGAG analysis. Osteochondral sections for decalcified histologic evaluation were collected from the medial metacarpal condyle at the lesion site (adjacent to the cartilage samples collected for paravital staining, MTT cell viability assay, and sGAG analysis) and from the medial aspect of the first phalanx and the medial proximal sesamoid bone. Lactated Ringer's solution was applied, as necessary, to prevent drying of the cartilage surface during processing.

Paravital staining of cartilage—Fresh cartilage was sectioned perpendicular to the articular surface into 100-μm slices on a sectioning instrument^c with a vibrating razor blade and stained with green fluorescent nucleic acid stain^d (1:20 solution with lactated Ringer's solution) and ethidium bromide^e (1:20 solution with distilled water) in lactated Ringer's solution. The sections were viewed on a microscope with UV light via a wide band-pass filter, and digital images were captured. The superficial, middle, and deep cartilage zones were defined as 9%, 47%, and 44% of mean cartilage thickness, respectively, by use of image analysis software.^b Live (fluorescent green) and dead (fluorescent red) cells were counted via a semiautomated method after manual thresholding of the image. Percentage viability was calculated from the proportion of green fluorescent nucleic acid— and ethidium bromide–stained cells.

MTT assay for cartilage metabolism—Fresh cartilage for the MTT cell viability assay was weighed and incubated with MTT (50 μL of 3-[4,5-dimethyl-2-thiazolyl]-2,5-diphenyl-2H-tetrazolium bromide)^f and medium^g (500 μL) for 3 hours at room temperature (approx 22°C) on a mixer.^h The MTT was removed, and the cartilage was incubated with extraction buffer (500 μL of 96% isopropanol and 4% 1M hydrochloric acidⁱ) for an additional 30 minutes at room temperature on a mixer.^h The resulting solution was read on a plate reader in triplicate (100 μL/well) at an absorbance of 530 nm. The mean of 3 optical density readings was calculated and divided by the wet weight of the cartilage in milligrams to yield the mean MTT per milligram of cartilage.

Cartilage sGAG—Cartilage samples were analyzed for sGAG concentration by use of the 1,9-dimethylmethylene blue dye assay as described.¹⁴ Results were expressed as a mean of 3 readings and reported as chondroitin sulfate C concentration (μg/mg) of cartilage.

Histologic evaluation, histologic scoring, and immunostaining—Osteochondral sections for histologic evaluation were fixed in neutral-buffered 10% formalin, decalcified in 20% citric acid and 40% formic acid mixed 1:1 before use, embedded in paraffin, and cut into 5-μm-thick sections, which were routinely stained with H&E, safranin O, and picrosirius red for collagen orientation. Sections were scored independently by 3 observers (EJR, MBH, and BJM) unaware of group assignments using the Osteoarthritis Research Society International scoring system.²¹ This score is calculated from the mathematical product of grade, which assesses the depth of cartilage damage indicating the severity of osteoarthritis, and stage, which assesses the area of cartilage affected. Stage was determined by manual measurements performed on digital images of the histologic slides and expressed as a percentage of affected cartilage, which was converted to a scoring system (0 to 4) as described.²¹

Immunohistochemical analysis was performed on the treatment group and control medial metacarpal condyle sections for COL2–3/4C_short and TUNEL staining for apoptosis, as described,¹⁴ except that the positive control for TUNEL staining was mammary gland tissue from a clinically normal female rat obtained 3 to 5 days after weaning rat pups and the negative control was normal equine articular cartilage. For control sections, TUNEL staining was performed on 2 randomly selected sections/endpoint (1, 4, and 8 to 10 months), for a total of 6 sections. Immunohistochemical slides were scored according to the degree of staining (mild, moderate, or marked) by 2 independent observers (EJR and an experienced technician).

Bone assessment—After evaluation of the intact third metacarpal bone via micro-CT, bone sections were collected from the medial metacarpal condyle for undecalcified histologic evaluation to assess fluorochrome uptake at the lesion site.

Micro-CT—The distal epiphysis of each third metacarpal bone was dissected free of soft tissues, and micro-CT images were acquired at 80 kV and 450 μA with a 45-μm isotropic voxel resolution.^j A calibration phantom present during scanning was used to convert CT values into Hounsfield units, as described.²² Images were reconstructed into a 3-D format and imported into analysis software.^k The spatial location of the impact site was measured on digital photographs obtained at postmortem examination, and these measurements were used to locate the impact site on the 3-D micro-CT image, with the mirror image coordinates used to identify a control site on the contralateral limb of the same horse. A 3 × 3 × 1-mm-deep region of interest was created at the bone surface of the condyle at the measured impact site, which was identified as the superficial region of the subchondral bone plate. An identical region of interest was placed 1 mm deep to the first region and was designated the deep region. For each region of interest, BMD (mg/mL), TMD (mg/mL), and bone volume fraction were calculated with the analysis software after automatic thresholding of the image, as described.²²

Evaluation of bone remodeling—A fluorochrome bone label was administered (oxytetracycline, 10 mg/kg, IV) to the remaining 9 horses at 3 months after impact injury. Calcein green (5 mg/kg, IV) was administered as a second bone label to the 4-month endpoint horses at 24 hours prior to euthanasia and to the 8- and 10-month endpoint horses at 6 months after impact injury. A second dose of oxytetracycline (10 mg/kg, IV) was also administered to the 8- and 10-month endpoint horses at 24 hours prior to euthanasia. Thus, the horses euthanized at 4 months received 2 doses of fluorochrome bone labels, whereas the 8- and 10-month endpoint horses received 3 doses. Fluorochrome bone labels were not administered to the horses euthanized at 1 month because there was insufficient time to administer 2 labels. At least 1 month between labels is considered ideal, and bone modeling would not be expected to begin on the day of surgery. Undecalcified sections for fluorochrome analysis were fixed in 100% ethanol for 48 hours and stored in 70% ethanol prior to infiltration with methyl methacrylate and embedding in benzoyl peroxide and methyl methacrylate. Fluorochrome sections were evaluated under UV light, and the number of remodeling osteons per hpf (200×) was counted sequentially for the full length of the articular surface. No other dynamic bone histomorphometry analyses were performed because on evaluation, oxytetracycline uptake was not discernible.

Statistical analysis—A generalized linear mixed model was used to analyze MTT, cartilage sGAG, micro-CT data, percentage of abnormal cartilage, effusion, subcutaneous thickening, scars, joint enlargement, inflammation, synovial darkening, synovial hypertrophy, India ink data, paravital staining, and fluorochrome data. Factors included in the model were treatment, lesion site versus distant site, and endpoint as well as their interactions. The random effects of horse and limb were taken into account. For data that were measured repeatedly over time (synovial fluid sGAG and COMP, synovial fluid analysis data, and lameness score), the Akaike information criterion^l was used to determine an error structure for the autoregression. For COMP, statistical software^m was used to control for unequal sample size. The assumptions of the ANOVA were assessed by comprehensive residual analyses. A Shapiro-Wilk test, Kolmogorov-Smirnov test, Cramer—von Mises test, and Anderson-Darling test were conducted to assess overall normality. Residuals were plotted against predicted values and explanatory variables (treatment, site, endpoint, and lesion) to look for patterns in the data that suggested outliers, unequal variance, or other problems. If residual analyses suggested a need for data transformation or data were presented as a ratio, analyses were performed on a log scale. If the overall F test was significant (P < 0.05), a Dunnett test for comparison of data at baseline within a treatment or a Tukey test between treatments and sites at each time was performed. To test for agreement between observers on scored data, such as lameness scores and histologic Osteoarthritis Research Society International scores, a weighted κ for > 2 categories and a simple κ for 2 categories were used with an exact P value. To test for differences between the treatment groups for binary scored data (translucency, color, surface roughness, and osteophytes), a McNemar test was performed. A Wilcoxon-Mann-Whitney test was used to compare the mean scores (1 to 4 for translucency, color, surface roughness, and osteophytes) between impact and control samples at each endpoint. Radiographic scores were analyzed with a Wilcoxon signed rank test for the difference before versus after within a treatment and for the difference between treatments. The data from the 8- and 10-month endpoints were pooled for all tests because time point was not a significant factor in the model for these 2 time points, and all analyses were blocked for horse. A 2-sided value of P < 0.05 was considered significant.

Results

Surgical outcome—The surgical technique was successfully completed in all 12 horses. Arthroscopic and fluoroscopic guidance were used to determine placement of the impact injuries, but the area of interest was not completely visible at full extension with either of these techniques. Despite this, the impactor tip appeared to be accurately placed. One horse developed a complication related to the surgical procedure in the control limb, when a No. 15 scalpel blade was broken during creation of the instrument portal and the blade fragment could not be retrieved from the tissues distal to the medial proximal sesamoid bone. This horse developed moderate (maximum grade, 3.5) lameness of the control limb after surgery, although the cause of lameness could not be definitively attributed to the blade fragment. This horse's control limb lameness was most severe 4 weeks after surgery but subsequently progressively improved to an intermittent lameness, with no visible lameness immediately prior to euthanasia at 4 months. No horses required rescue analgesia after surgery. Most horses had minimal to no visible lameness at a walk after surgery.

Clinical assessment—Lameness examinations revealed lameness in the impact-injured limb, the control limb, or both (Figure 3). The most severe lameness was grade 3 for an impact-injured limb and grade 3.5 for a control limb. Lameness was observed in the impact-injured limb for a mean of 6 ± 6.5 weeks (range, 0 to 21 weeks). However, there was no significant difference in lameness scores between impact-injured and control limbs. Agreement between observers was good, with a weighted κ of 0.72 (P = 0.01).

Mean ± SE lameness scores (evaluated at a trot) for impact-injured (black squares with solid error bars) and control (white triangles with dotted error bars) limbs obtained for up to 10 months after surgery in horses (1 month, 12 horses; > 1 to 4 months, 9 horses; > 4 to 8 months, 5 horses; and > 8 to 10 months, 2 horses).

Citation: American Journal of Veterinary Research 73, 10; 10.2460/ajvr.73.10.1540

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Mean ± SE lameness scores (evaluated at a trot) for impact-injured (black squares with solid error bars) and control (white triangles with dotted error bars) limbs obtained for up to 10 months after surgery in horses (1 month, 12 horses; > 1 to 4 months, 9 horses; > 4 to 8 months, 5 horses; and > 8 to 10 months, 2 horses).

Citation: American Journal of Veterinary Research 73, 10; 10.2460/ajvr.73.10.1540

• Download Figure
• Download figure as PowerPoint slide

Mean ± SE lameness scores (evaluated at a trot) for impact-injured (black squares with solid error bars) and control (white triangles with dotted error bars) limbs obtained for up to 10 months after surgery in horses (1 month, 12 horses; > 1 to 4 months, 9 horses; > 4 to 8 months, 5 horses; and > 8 to 10 months, 2 horses).

Citation: American Journal of Veterinary Research 73, 10; 10.2460/ajvr.73.10.1540

• Download Figure
• Download figure as PowerPoint slide

Radiography did not consistently reveal progression of the injury during the study period. Six impact-injured limbs and 2 control limbs had increased radiographic scores at endpoint. Radiographic scores for impact-injured limbs were significantly higher at endpoint than baseline, and no significant (P = 0.50) difference was noted for control limbs. However, pairwise comparison of impact-injured and control limbs in the same horse revealed no significant (P = 0.13) difference. The low α error warranted a post hoc power calculation, which was 57%. Although initial assessment by a board-certified radiologist found all baseline radiographs to be free of evidence of osteoarthritis, masked scoring by the same radiologist at the end of the study found mild osteoarthritis in 1 limb and minimal osteoarthritis in 4 limbs.

Cytologic evaluation—Results of synovial fluid analysis were within reference ranges for all horses at all time points. There was no difference in synovial fluid variables between impact-injured and control joints.

Synovial fluid sGAG and COMP—Biomarkers of osteoarthritis in synovial fluid only indicated a mild effect from impact trauma on the joint. Synovial fluid sGAG concentrations were not significant (P = 0.48) for treatment. Synovial fluid COMP concentrations had a significant effect for the combination of treatment and synovial fluid collection time, with higher concentrations in impact-injured joints. At 8 months, there was a significant (P = 0.01) difference in COMP concentrations between impact-injured joints and control joints (Figure 4).

Mean ± SE COMP concentrations in synovial fluid from impact-injured (gray bars) and control (white bars) joints obtained for up to 8 months after surgery in horses. Data were obtained from all impact-injured joints at all time points. Data from control joints were obtained from 12 horses at baseline; 4 horses at 1, 2, 3, and 4 months; and 5 horses at 6 and 8 months. ^a,bGroups with the same letter are significantly (P < 0.05) different.

Citation: American Journal of Veterinary Research 73, 10; 10.2460/ajvr.73.10.1540

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Mean ± SE COMP concentrations in synovial fluid from impact-injured (gray bars) and control (white bars) joints obtained for up to 8 months after surgery in horses. Data were obtained from all impact-injured joints at all time points. Data from control joints were obtained from 12 horses at baseline; 4 horses at 1, 2, 3, and 4 months; and 5 horses at 6 and 8 months. ^a,bGroups with the same letter are significantly (P < 0.05) different.

Citation: American Journal of Veterinary Research 73, 10; 10.2460/ajvr.73.10.1540

• Download Figure
• Download figure as PowerPoint slide

Mean ± SE COMP concentrations in synovial fluid from impact-injured (gray bars) and control (white bars) joints obtained for up to 8 months after surgery in horses. Data were obtained from all impact-injured joints at all time points. Data from control joints were obtained from 12 horses at baseline; 4 horses at 1, 2, 3, and 4 months; and 5 horses at 6 and 8 months. ^a,bGroups with the same letter are significantly (P < 0.05) different.

Citation: American Journal of Veterinary Research 73, 10; 10.2460/ajvr.73.10.1540

• Download Figure
• Download figure as PowerPoint slide

Postmortem evaluations—At postmortem examination, impact lesions were located in a variety of areas throughout the palmar aspect of the medial metacarpal condyle. There was a wide variation in both the extent and depth of the lesions, with some being focal and others more diffuse. In some joints with small lesions, it appeared that the 4 impact injuries had been placed in overlapping sites, rather than in a cloverleaf pattern. With a small lesion, it was also more difficult to obtain all of the required samples from the area with macroscopic evidence of damage, resulting in sampling of tissue from the less affected periphery of the impact site. It appeared that in some joints, the impactor had glanced off the cartilage, creating a shear injury, whereas in others, the cartilage defect suggested a more ideal perpendicular impact. Independent evaluation of the postmortem images (Figure 5) by the authors (EJR, AMC, MBH, and BJM) categorized the lesions in terms of severity as mild in 3 horses, moderate in 4 horses, and severe in 5 horses. When these categories were evaluated by endpoint, there were 1 moderate and 2 severe lesions at 1 month, 1 mild and 3 moderate lesions at 4 months, 1 mild and 2 severe lesions at 8 months, and 1 mild and 1 severe lesion at 10 months.

Photographs of impact-injured metacarpal condyles (palmar aspect) obtained after euthanasia in 4 horses. Asterisks indicate the medial metacarpal condyles. A—Four months after impact injury, the lesion is of moderate severity and ideally located at the midpoint of the condyle, just palmar to the transverse ridge. B—Four months after impact injury, the lesion is mild and superficial. C—Ten months after impact injury, the lesion is severe and diffuse. D—One month after impact injury, the lesion is severe but focal.

Citation: American Journal of Veterinary Research 73, 10; 10.2460/ajvr.73.10.1540

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Photographs of impact-injured metacarpal condyles (palmar aspect) obtained after euthanasia in 4 horses. Asterisks indicate the medial metacarpal condyles. A—Four months after impact injury, the lesion is of moderate severity and ideally located at the midpoint of the condyle, just palmar to the transverse ridge. B—Four months after impact injury, the lesion is mild and superficial. C—Ten months after impact injury, the lesion is severe and diffuse. D—One month after impact injury, the lesion is severe but focal.

Citation: American Journal of Veterinary Research 73, 10; 10.2460/ajvr.73.10.1540

• Download Figure
• Download figure as PowerPoint slide

Photographs of impact-injured metacarpal condyles (palmar aspect) obtained after euthanasia in 4 horses. Asterisks indicate the medial metacarpal condyles. A—Four months after impact injury, the lesion is of moderate severity and ideally located at the midpoint of the condyle, just palmar to the transverse ridge. B—Four months after impact injury, the lesion is mild and superficial. C—Ten months after impact injury, the lesion is severe and diffuse. D—One month after impact injury, the lesion is severe but focal.

Citation: American Journal of Veterinary Research 73, 10; 10.2460/ajvr.73.10.1540

• Download Figure
• Download figure as PowerPoint slide

Macroscopic assessment—Macroscopic changes associated with the experimental injuries included significant differences between impact-injured and control joints in color and translucency of articular cartilage, among endpoints in effusion and joint enlargement, and between sites in the percentage of abnormal cartilage. Postmortem scores were significantly worse in impact-injured joints than in control joints for translucency at 1 month and for color at the combined 8- and 10-month endpoint. When endpoints were compared within treatment groups, there was a significant difference for effusion in impact-injured joints, with less effusion at 1 month than at 4 or 8 to 10 months, and for enlargement of both the impact-injured and control joints, with decreasing enlargement over time. The percentage of abnormal cartilage on the medial metacarpal condyle from impact-injured and control joints combined, compared with other sites at the same time point, was significantly greater than the lateral aspect of the proximal phalanx at 1 month (P = 0.01); the sagittal groove at 1 month, 4 months, and 8 to 10 months (P = 0.01); the sagittal ridge at 1 month and 8 to 10 months (P = 0.01); the medial proximal sesamoid bone at 4 months (P < 0.05); and the lateral proximal sesamoid bone at 4 months (P = 0.01).

Disruption of the articular surface occurred in a larger area on the medial metacarpal condyle than at other sites. The percentage of damaged area of articular cartilage, as evidenced by India ink uptake, was significantly (P = 0.01) different among areas of interest. Specifically, the percentage of damaged area on the medial metacarpal condyle from impact-injured and control joints combined was significantly larger than the medial and lateral aspect of the proximal phalanx (P = 0.01) and the medial (P < 0.05) and lateral (P = 0.01) proximal sesamoid bone. However, there was no treatment effect (P = 0.90), indicating that the area of cartilage disruption within the medial metacarpal condyle of impact-injured and control joints was not significantly different.

Paravital staining—Cell death was restricted to the superficial zone in most cases. Specifically, in only 2 horses was there decreased cell viability through all 3 zones (superficial, middle, and deep) of cartilage at the site of impact. In the superficial zone, cell viability was significantly (P = 0.01) decreased at the combined 8- and 10-month endpoints, compared with cell viability at the 1- and 4-month endpoints (Table 1). Among all zones, mean cell viability at the lesion site in impact-injured joints (39.8%) was lower than at the distant site in impact-injured joints (53.9%) and at the lesion (54.5%) and distant (67.4%) sites in control joints, but these differences were not significant (P = 0.26).

MTT assay for cartilage metabolism—Cell proliferation and viability in cartilage, as assessed via the MTT assay, declined progressively with time after surgery (P = 0.05). At the lesion site for both impact-injured and control joints, the mean ± SD MTT value decreased from 14.2 ±3.1 MTT/mg of cartilage at 1 month, to 12.2 ± 1.3 MTT/mg of cartilage at 4 months, and to 11.1 ±3.1 MTT/mg of cartilage at 8 to 10 months. The mean MTT value at the distant site followed a similar pattern (12.2 ± 2.5 MTT/mg of cartilage at 1 month and 11.6 ± 2.5 MTT/mg of cartilage at 8 to 10 months) but with a slight increase at 4 months (13.6 ± 2.4 MTT/mg of cartilage). However, there was no significant (P = 0.78) difference in cell viability between impact-injured and control joints.

Cartilage sGAG—Depletion of sGAG from cartilage due to impact injury was observed. The sGAG concentration from cartilage at the impact-injured joint lesion site (mean ± SD, 44.1 ± 3.0 μg/mg; range, 32.9 to 57.5 μg/mg) was significantly lower than at the control joint lesion site (mean ± SD, 53.3 ± 3.0 μg/mg; range, 37.7 to 70.2 μg/mg), impact-injured joint distant site (mean ± SD, 52.4 ± 3.0 μg/mg; range, 30.4 to 71.1 μg/mg), and control joint distant site (mean ± SD, 50.0 ± 3.0 μg/mg; range, 27.1 to 69.4 μg/mg) at all time points combined.

Histologic evaluation, histologic scoring, and immunostaining—Cartilage defects were observed via histologic evaluation at all sites of impact injury. Histologic evaluation, expressed as the mean of Osteoarthritis Research Society International scores by 3 observers, revealed a significant effect for treatment (P < 0.05) and site (P = 0.01). For all sites combined, scores for control joints (mean score, 0.5) were lower than scores for impact-injured joints (mean score, 1.0). There was a significant (P = 0.01) difference in mean score for both impact-injured and control joints at all time points combined among all comparisons of the 3 sites (medial metacarpal condyle, medial proximal sesamoid bone, and medial aspect of the proximal phalanx). The highest mean score was observed at the medial metacarpal condyle (mean score, 6.7), with much lower scores at the medial proximal sesamoid bone (mean score, 1.7) and medial aspect of the proximal phalanx (mean score, 0.9). The interobserver agreement for scoring was almost perfect, with a coefficient of concordance of 0.81 to 0.99 for the 3 observers at all 3 sites. However, there was evidence of bias between 2 observers (EJR and MBH) at the medial aspect of the proximal phalanx (P = 0.01).

Immunohistochemical analysis revealed mild evidence of apoptosis and mild to moderate presence of COL2–3/4C_short in impact-injured medial metacarpal condyle sections. There was a significantly (P < 0.05) higher apoptosis score for sections from impact-injured joints, compared with control joints, for 1 observer (technician) and a nonsignificant (P = 0.06) difference for the other observer (EJR). Interobserver agreement was moderate, with a weighted κ of 0.58 (P = 0.01). Immunohistochemical analysis, taking the COL2–3/4C_short as evidence of osteoarthritis progression, revealed a significantly higher score in sections from impact-injured joints than for control joints. In addition, there was a significant endpoint effect in impact-injured sections. The impact-injured medial metacarpal sections had the highest mean score at 4 months (mean score, 2.9), with the lowest mean score at 1 month (mean score, 1.2) and an intermediate mean score at 8 to 10 months (mean score, 1.5). There was good interobserver agreement for COL2–3/4C_short (weighted κ = 0.65; P = 0.01).

Bone assessment—Due to variability in the site of impact injury, micro-CT analysis of the lesion site also occurred at different anatomic locations. A macroscopic defect in the bone was identifiable on the micro-CT image in only 1 horse. Micro-CT analysis for BMD revealed a significant site and endpoint interaction but no treatment effect, so impact-injured and control joint values were combined. Mean BMD values were highest at 1 month, decreased by 4 months, and then increased again by 8 to 10 months (Figure 6). Micro-CT analysis for TMD revealed a significant site effect (P < 0.05) and endpoint effect (P = 0.01). Mean TMD values, consisting of impact-injured and control joint values that were combined due to lack of treatment effect, were significantly higher for the superficial site (815.0 mg/mL) than the deep location (799.6 mg/mL). Similar to BMD values, TMD values were highest at 1 month, decreased at 4 months, and increased again at 8 to 10 months (Figure 7). Micro-CT analysis for bone volume fraction did not reveal any significant differences.

Mean ± SE BMD for combined impact-injured and control limbs at superficial (white bars) and deep (gray bars) sites on the medial metacarpal condyle and at endpoints of 1, 4, and 8 to 10 months in 12 horses. ^a–eGroups with the same letter are significantly (P < 0.05) different.

Citation: American Journal of Veterinary Research 73, 10; 10.2460/ajvr.73.10.1540

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Mean ± SE BMD for combined impact-injured and control limbs at superficial (white bars) and deep (gray bars) sites on the medial metacarpal condyle and at endpoints of 1, 4, and 8 to 10 months in 12 horses. ^a–eGroups with the same letter are significantly (P < 0.05) different.

Citation: American Journal of Veterinary Research 73, 10; 10.2460/ajvr.73.10.1540

• Download Figure
• Download figure as PowerPoint slide

Mean ± SE BMD for combined impact-injured and control limbs at superficial (white bars) and deep (gray bars) sites on the medial metacarpal condyle and at endpoints of 1, 4, and 8 to 10 months in 12 horses. ^a–eGroups with the same letter are significantly (P < 0.05) different.

Citation: American Journal of Veterinary Research 73, 10; 10.2460/ajvr.73.10.1540

• Download Figure
• Download figure as PowerPoint slide

Mean ± SETMD for combined impact-injured and control limbs at superficial (white bars) and deep (gray bars) sites on the medial metacarpal condyle and at endpoints of 1, 4, and 8 to 10 months in 12 horses. ^a–dGroups with the same letter are significantly (P < 0.05) different.

Citation: American Journal of Veterinary Research 73, 10; 10.2460/ajvr.73.10.1540

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Mean ± SETMD for combined impact-injured and control limbs at superficial (white bars) and deep (gray bars) sites on the medial metacarpal condyle and at endpoints of 1, 4, and 8 to 10 months in 12 horses. ^a–dGroups with the same letter are significantly (P < 0.05) different.

Citation: American Journal of Veterinary Research 73, 10; 10.2460/ajvr.73.10.1540

• Download Figure
• Download figure as PowerPoint slide

Mean ± SETMD for combined impact-injured and control limbs at superficial (white bars) and deep (gray bars) sites on the medial metacarpal condyle and at endpoints of 1, 4, and 8 to 10 months in 12 horses. ^a–dGroups with the same letter are significantly (P < 0.05) different.

Citation: American Journal of Veterinary Research 73, 10; 10.2460/ajvr.73.10.1540

• Download Figure
• Download figure as PowerPoint slide

Evaluation of bone remodeling—The fluorochrome calcein green label was distinct in areas of remodeling, but oxytetracycline uptake was not discernible. Therefore, measurement of the amount of remodeling that occurred between fluorochrome injections could not be performed. Histologic evaluation of combined impact-injured and control joint sections for fluorochrome deposition revealed a significant endpoint effect, with a significant difference between sections harvested at 4 months and those collected at 8 to 10 months (P = 0.01) but with no treatment effect (P = 0.23). There was more evidence of remodeling at 4 months (mean, 6.2 remodeling units/field) than at 8 to 10 months (mean, 3.2 remodeling units/field).

Discussion

The impact injury method succeeded in creating mild changes in some outcome variables consistent with the development of osteoarthritis in impact-injured joints (specifically synovial fluid COMP concentrations, cartilage color and translucency, cartilage sGAG concentration, histologic evaluation via Osteoarthritis Research Society International scoring, and immunohistochemical evidence of apoptosis and collagen degradation). Overall, results suggest that cartilage injury, rather than subchondral bone injury, initiated osteoarthritis in this experiment. The lack of consistent evidence of osteoarthritis in all outcome variables at all time points can be attributed to the variability in the impact lesions created, insufficient severity of the method, sampling error, or natural variability in the individual response.

Clinically, all horses had a mild response to the surgical procedure, although significant differences between impact-injured and control limbs were not detected. Postoperative lameness was mild to moderate and decreased with time after approximately 1 to 2 months, which suggested that the osteoarthritic pain created by this method subsided with time or that the lameness was more likely caused by soft tissue damage at surgery than by progression of osteoarthritis. Some horses also had intermittent control limb lameness in which there was not a significant difference from the impact-injured limb lameness values, which further supports the suspicion that soft tissue injury during surgery contributed to the lameness. However, occult preexisting lameness that was not initially detected cannot be ruled out and overload of the contralateral limb, caused by shifting weight from the impact-injured limb to the control limb, may also have contributed to the lameness. Radiographic scores for evidence of osteoarthritis in impact-injured limbs were significantly worse at endpoint than at baseline. However, this difference was not consistent enough to withstand pairwise comparison of impact-injured and control limbs in the same horse, possibly due to a variable degree of preexisting damage. Radiographs often reveal little about cartilage integrity until joint space narrowing and collapse are present.^23,24 In our study horses, as in horses with naturally occurring joint disease, it was difficult to radiographically detect and predict the clinical progression of osteoarthritis.²⁵ This unpredictability in the pathological consequences of joint injury may be due, in part, to the variability in involvement of the different tissues in the joint²⁵ and to the variability in the severity of the injury.

Macroscopic evaluation of the metacarpophalangeal joints revealed some changes associated with the impact injuries, although not all variables were consistently affected. There were abnormalities in the color and translucency of articular cartilage, effusion, and joint enlargement. Loss of articular cartilage translucency occurred early in the study, whereas color abnormalities were observed later as osteoarthritis progressed, indicating the accumulation of nonenzymatic glycation products.²⁶ When impact-injured joints were compared, effusion increased with time because synovial fluid production reflected the development of osteoarthritis, but joint enlargement decreased as surrounding soft tissues remodeled after surgery.

Although there was more India ink uptake and therefore more disruption in the articular cartilage surface on the medial metacarpal condyle than at other sites, there was no difference between impact-injured and control joints, which raises concerns about the validity of the method, the condition of the horses at baseline, or the specimen preparation. The focal nature of the impact injuries and their limited progression may not have been sufficient to influence India ink results for the entire condyle. A study²⁷ involving quantification of India ink uptake found a 10% error in the staining of nondegenerated cartilage in equine proximal phalanx specimens. Although this nonspecific false-positive uptake may have occurred in our study as well, it was much less likely due to the thorough rinsing and blotting to remove excess ink that is routinely performed in our laboratory.²⁰ Measurement of reflectance from the articular surface has been described as an extension of India ink staining,²⁸ but because the distal medial metacarpal condyle has sufficient curvature to require a complex mathematical correction for nonplanar geometry, it was not used in the present study. In addition, India ink staining at sites of preexisting disease or incidental cartilage damage may have obscured differences caused by the impact injury alone. Some preexisting disease could have been present in this cohort of horses, according to a recently published survey of racehorses.²⁰

When combined, the paravital staining and MTT assay results indicated that a mild amount of progressive superficial cartilage damage occurred, although there was no significant effect from impact injury. Cell viability, as determined by paravital staining of cartilage, was lowest in the superficial zone, with a significant decrease at the 8- to 10-month endpoint. However, this finding is unlikely to be indicative of progressive osteoarthritis from the impact injury because there was no significant difference between treatment groups (although preexisting osteoarthritis may have progressed in both impact-injured and control joints). Similarly, the amount of metabolically active cells, as assessed with the MTT assay, decreased over time (albeit not significantly). However, the MTT assay only provides relative comparisons between values, so it is unknown whether this magnitude of decrease is clinically relevant.

Further evidence of mild, focal osteoarthritis occurring after impact injury was provided by COMP and sGAG results. Higher concentrations of the articular cartilage component COMP were present in synovial fluid from impact-injured joints than from control joints, which indicated more cartilage destruction caused by the catabolic effects of osteoarthritis. The decreased concentration of sGAG in cartilage from the impact-injured joint lesion site indicated a local effect of focal osteoarthritis caused by the impact injury. It is interesting that there was a local effect of sGAG depletion in the cartilage at the impact-injured joint lesion site, but there was no change in the sGAG concentration in impact-injured joint synovial fluid. However, it is possible that a difference between impact-injured and control joints may have been observed if the synovial fluid had been evaluated more frequently. This finding indicated that the local effect was not substantial enough to have an effect on the sGAG concentration in the synovial fluid of the entire joint. However, in the experience of the authors, focal changes in the palmar aspect of the metacarpal condyles cause lameness, even without widespread osteoarthritis involving the entire joint. Thus, clinical disease may be caused by focal osteoarthritis in this region. On the basis of these results, use of COMP may have more promise than use of sGAG of being a potential noninvasive method of staging osteoarthritis. However, the results of previous studies^18,19 of COMP concentrations in synovial fluid are conflicting, so more research is needed before this biomarker can be used clinically.

Higher Osteoarthritis Research Society International scores for impact-injured joint sections than for control sections indicated that there was more histologic evidence of osteoarthritis in impact-injured joints. The scores for impact-injured medial metacarpal condyles were overall consistent, with grades ranging from 4.5 to 5.5 by all observers in 10 of the 12 horses. Although scores did not progress over time, it is possible that progression may have occurred with a longer time period, but this seems unlikely due to the lack of progression by 10 months. Alternatively, this unexpected finding may simply indicate that the impact injury used in the present study was insufficient to cause progression of osteoarthritis. Repeated impact injury or strenuous treadmill exercise may contribute to progression in future studies.

Immunohistochemical evidence of apoptosis was mild (mean score, 1.5) with a weak treatment effect, but there was no difference among endpoints. Joint damage may not have been severe enough to cause progression of apoptosis over time. Interestingly, the presence of COL2–3/4C_short was significantly higher at 4 months than at the other endpoints. There may have been some progression of osteoarthritis from 1 to 4 months, but it would be unlikely for osteoarthritis to improve by 8 to 10 months because the disease is progressive. Improvement may be possible if the initial focal cartilage degeneration becomes covered by scar tissue and normal homeostasis returns to the joint. It is still unknown at what point the process becomes irreversible and at what point focal osteoarthritis becomes generalized. Alternatively, it could be concluded that the joints evaluated at 4 months may have sustained more severe impact injuries (ie, the ideal perpendicular impact, rather than a glancing shear impact). However, this conclusion was not supported by the macroscopic classification of the 4-month samples as mild in 1 horse and moderate in 3 horses. Another explanation for the increased values of COL2–3/4C_short at 4 months could be that these horses had more preexisting disease, although this theory was not supported by baseline lameness or radiographic findings. Alternately, osteoarthritis progression can be phasic, and without ongoing injury from high-speed training, collagen degradation may have been normalizing after the 4-month time point.

Micro-CT data were challenging to normalize due to the difficulty of standardizing lesion location, which made it hard to differentiate between true sclerosis and a normal variation in bone density associated with the anatomic location of the site. The variability in density of the distal third metacarpal bone according to anatomic location has been well documented.^29–31 A confounding factor was that a macroscopic defect in the bone was identifiable in only 1 horse, so correlating the lesion location between the 2-D postmortem photographs and the 3-D micro-CT images may have been a source of error. In future studies, a radiopaque marker could be placed over the site of the cartilage defect observed on the postmortem specimen, immediately prior to obtaining micro-CT images, to accurately identify the impact site.

The BMD and TMD values from the medial metacarpal condyle were lower than those reported by Rubio-Martinez et al²² for Thoroughbred racehorses with mild to severe subchondral bone disease. However, sections analyzed in that study were located slightly closer to the sagittal ridge than sections in the present study. Alternatively, the lower values in the present study may have been due to a lower amount of exercise. It is interesting that both BMD and TMD values were highest at 1 month, decreased at 4 months, and increased slightly again at 8 to 10 months. This finding suggests that initial sclerosis was undergoing remodeling at 4 months and the amount of remodeling had decreased by 8 to 10 months, which would be consistent with the COL2–3/4C_short immunohistochemical data. These findings may have been influenced by lunging exercise being performed immediately prior to euthanasia at 4 months, but not at 8 to 10 months. Although the exact timing of the peak of remodeling after subchondral bone injury has not been determined, results of the present study could be interpreted as representing delayed or prolonged remodeling, which would most likely be caused by exercise or repeated injury. However, there was no other evidence of ongoing damage.

Characterization of subchondral bone remodeling with oxytetracycline as a fluorochrome label was disappointing, although calcein green was readily visible. We recommend that subsequent studies use higher doses, although there may be some triggering of enteritis.

A similar method of inducing post-traumatic osteoarthritis in equine femorotibial joints¹⁴ caused consistent degeneration of the cartilage surface, whereas this did not occur in the present study. This difference was most likely due to the anatomic differences between these 2 joints (specifically their size, which allowed better visual evaluation when creating the impact injuries in the femorotibial joint). In the pilot study, fairly consistent injuries were created on the palmar aspect of the metacarpal condyle, but joint manipulation into hyperextension may have been easier due to increased laxity of the soft tissues in cadaveric limbs, which had been separated from the proximal portion of the limb at the level of the distal aspect of the radius. The cartilage and subchondral bone in the femorotibial and metacarpophalangeal joints would be expected to have different structural and functional properties, but these differences were accommodated by determining the threshold for injury of the target tissue during preliminary in vitro research. The lower number of horses per endpoint in the present study (3, 4, or 5 horses), compared with 5 horses/endpoint in the previous study, may also have caused a decrease in power of this study Four horses were originally assigned to each endpoint, but preliminary results (inconsistent macroscopic abnormalities in the first horse euthanized at 1 month) encouraged us to put more emphasis on the later time points. The 1-month endpoint was originally chosen to investigate the early changes in osteoarthritis, which are currently not well described. The 4- and 8-month endpoints were chosen to be between and after, respectively, the endpoints of 2.5 and 6 months in the previous study.¹⁴

The major limitations of the study were the variability in location (site), depth, and area of the impact lesions and the corresponding variability in the impact force delivered to the joint surface, both of which were most likely due to an inability to directly observe creation of the lesions. Samples were obtained from the area with macroscopic evidence of damage, but the inconsistent location of the injury site caused variation in the location of sample collection. This variation in sample location made interpretation of results difficult because of the highly site-specific variation in structural and functional properties observed in the metacarpal condyles.³² The most important consequence of the variability in lesion location was that only a few of the lesions were located at the site of naturally occurring palmar osteochondral disease. As described, the small joint volume and thick periarticular soft tissue of the palmarodistal aspect of the metacarpophalangeal joint severely interfered with viewing and positioning of the impactor tip and may have interfered with energy transfer to the articular surface. Although the technique described by Byron and Goetz¹⁵ allowed instrument access to the site of interest on the palmarodistal aspect of the medial metacarpal condyle, it was unlikely that a 70° arthroscope, as used in that study, would allow visual evaluation of this location during hyperextension of the joint, as is required when placing the impactor tip perpendicular to the articular cartilage. An alternative method for creating impact injury to that location could be to use the proximal sesamoid bones as the mechanism for delivering the impact force, in a similar manner to impacting the patella against the distal portion of the femur in other species.³³ Use of the proximal sesamoid bones as the impactor could be expected to decrease soft tissue injury because it is a closed joint model. It may also allow more accurate localization of the impact injury because of the normal anatomic positioning of the bones in relation to each other and the minimal restrictions placed on the procedure by the small joint space because arthroscopic viewing and associated instrumentation are not necessary.

Due to the curved geometry of the condyles,³² small alterations in the impact site may have also altered contact between the impactor tip and the articular surface, resulting in a shear force rather than a perpendicular force. This < 90° angle of impact may have caused the impactor tip to slide along the curved articular surface and thus disperse the force applied to the underlying osteochondral unit, which may have had more influence on the outcome variables than the differences in lesion location. Macroscopic evidence of this motion was observed on postmortem examination. Inadequate visual evaluation also contributed to placement of the 4 impact injuries in overlapping sites in some horses, rather than in a cloverleaf pattern. Although this overlap increased the force at that site, it decreased the area of the lesion to less than that of a critically sized defect, such that the articular cartilage could have had the ability to heal spontaneously. In addition, the number of samples that could be obtained from the damaged area in small lesions was limited, so sampling of tissue from the less affected periphery of the impact site may have been a source of error.

The impact injuries appeared sufficient to cause a localized reaction at the site of injury but were not severe enough to cause global osteoarthritis in the joint. The fact that the single impact method used in this study did not cause the same degree of injury seen in naturally occurring palmar osteochondral disease supports the theory that this disease is caused by repetitive injury. However, only a few of the lesions in the present study were located at the site of naturally occurring disease, so a more consistent method is required before ruling out the possibility that it could be caused by a single injury in certain cases. Post-traumatic osteoarthritis would be expected to result from single impact injury, but the injury used in the present study may not have been severe enough to consistently cause this disease. A more substantial injury, created with repetitive impacts or higher impact energy, may be required to cause joint dysfunction and continuing lameness at this site in the metacarpophalangeal joint. The impact injury alone may not be sufficient to cause progression of osteoarthritis, so additional trauma in the form of high-speed exercise on a hard surface, such as a treadmill, may also be required. Although treadmill exercise allows more uniform intensity, lunging was chosen as the form of exercise in this study because it involves a more natural surface, allowing more natural hoof movement. Two surfaces, grass and dirt, were used for lunging, but these surfaces were overall similar with regard to concussive forces on the limbs and all horses were exercised on the same surface on any given day.

The impact injury method resulted in mild evidence of focal osteoarthritis in the palmar aspect of the medial metacarpal condyle. Some cartilage damage was inflicted, but minimal bone damage occurred, so there was limited evidence for progression of osteoarthritis over time. Therefore, additional research is required to develop a more reliable and consistent method for induction of osteoarthritis of the palmar aspect of the metacarpus in horses. An improved method would incorporate the outcome variables used in the present study, expand the time frame and exercise protocol, and improve the consistency of injury location.