Materials and Methods

Experimental procedures—The study population consisted of 93 client-owned horses that underwent arthroscopic surgery for treatment for OCD of the IRT of 1 or both TCJs (total of 134 joints). A control group of 38 horses (38 TCJs) without disease of the TCJs was also included. The study procedures were part of the normal treatment routines of the horses, and the owners were informed and accepted that their horses were taking part in the study. This project was approved by the Research Ethics Board of Lund/Malmö, which is a Swedish governmental freestanding ethical board not affiliated with any institution or company. The OCD-affected horses were a sample of all horses undergoing arthroscopic surgery at Helsingborg Equine Hospital between October 15, 2001, and June 4, 2004. The sample consisted of horses with OCD of the IRT that were randomly assigned in the routine booking system to one of the surgeons (PB) at the hospital, without consideration of reproductive status, breed, or age of the horse.

The control horses were a sample of all horses that were euthanatized or died at the hospital between March 8, 2002, and May 14, 2004. The control group was a random sample limited by exclusion criteria. Horses were excluded from the control group if, before they were euthanatized or died, they had clinical and clinicopathologic signs of systemic disease, or had gross evidence of disease of the TCJs during necropsy. All control horses were euthanatized for reasons unrelated to their selection and subsequent inclusion in the study.

The first part of the study was performed as a nonblinded, cross-sectional study with 1 observer. The second part was performed as an observational design with 2 observers blinded to the condition of the horse and with repeated assessments made independently of each other.

The degree of joint effusion in the TCJ of each OCD-affected horse, lameness of the affected limb or limbs observed during trotting, and results of a pelvic limb flexion test were subjectively scored before surgery by 1 of the authors (PB). Pelvic limb lameness was evaluated while the horse was led during trotting in a straight line for a distance of 25 to 30 m on a firm surface, both toward and away from the examiner. A pelvic limb flexion test was performed by flexing the tibiotarsal (hock) and femoropatellar-femorotibial (stifle) joints while extending the hoof for 1 minute and then examining the horse for lameness in the limb that was flexed as the horse was led trotting away from the examiner.¹⁶ Joint effusion was graded subjectively as follows: 0 = none, 1 = mild, 2 = moderate, 3 = severe, and 4 = extreme. Grading was based upon the system reported by Bergin et al,¹⁷ except that in our study, we combined grades 4 and 5 into a single grade (4). Lameness during trotting before and after a pelvic limb flexion test was graded by use of a scale of 0 (sound) to 5 (non–weight bearing), with increments of half scores.¹⁸ When allowed by an owner, the site of lameness was confirmed by observing resolution of lameness after administering 10 mL of local anesthetic solution (2% mepivacaine hydrochloride) into the affected TCJ. Radiographic examination of the affected hock joint by the referring veterinarian or at Helsingborg Equine Hospital confirmed OCD of the IRT of 1 or both TCJs. Horses with OCD of the IRT were retained in the study if they also had OCD at the MM of the tibia or LTR of the talus. The horse's reproductive status, breed, age, and specific limb or limbs affected were recorded.

Each OCD-affected horse was anesthetized and positioned in dorsal recumbency. The OCD fragment at the IRT was removed arthroscopically, and the exposed subchondral bone was debrided. The OCD fragments present at the MM of the tibia or LTR of the talus were also removed, and curettage of the exposed subchondral bone was performed.

Two synovial biopsy specimens were obtained during surgery from each of 134 affected joints of 93 horses with a Ferris-Smith rongeur. Biopsy specimens of the synovium were collected from the joint capsular fold directly opposite the IRT of the affected TCJ and fixed in neutral-buffered 5% formaldehyde^a within 20 minutes after sample collection. Biopsy specimens were embedded in paraffin, cut at a thickness of 5 μm, stained with H&E stain, and examined under a light microscope.

Following death or euthanasia (by IV administration of an overdose of pentobarbital sodium) each control horse was placed in lateral recumbency for sample collection. Two synovial tissue specimens from 1 TCJ of each of 38 control horses were collected from the joint capsular fold directly opposite the IRT through a small arthrotomy incision with a Ferris-Smith rongeur. These tissue specimens were similarly prepared and evaluated as described for the aforementioned biopsy specimens. The time from death to sampling of the synovium, with the horse's carcass refrigerated (5°C) was also recorded. Both TCJs of these 38 horses were then examined grossly by enlarging the arthrotomy after the tissue specimens were harvested. One of the authors (PB) harvested all control samples of synovium. Samples were removed from the uppermost TCJ of the dead, laterally recumbent horse.

Histologic evaluation—To develop and investigate a scoring system, the first 100 of the 172 synovial specimens obtained from OCD-affected horses (n = 54) and control (25) horses were thoroughly examined for evidence of pathological changes. In these 100 specimens, we looked for the occurrence of the following 6 changes: proliferation of emarginated and palisading synoviocytes, proliferation of scattered synoviocytes, focal infiltration by inflammatory cells, perivascular infiltration by inflammatory cells, focal mineralization, and villus proliferation. These 6 abnormalities have been used to characterize histopathologic signs of inflammation of the synovium of humans,^19–23 cattle,²⁴ and dogs.^25–28 Two of these 6 abnormalities—infiltration by inflammatory cells and proliferation of emarginated synoviocytes—were found to be descriptive markers for inflammation of equine synovial membranes in our specimens. Only these 2 markers were used further in the study.

All of the 172 tissue samples were then examined and scored by 2 experienced pathologists (pathologist A [MS] and pathologist B [JT]). These pathologists used the 2 selected pathological abnormalities to grade the degree of inflammation of the synovium. Both pathologists examined all tissue specimens twice in a randomized, blinded design, and the results were recorded and summarized after all 4 examinations were completed. The degree of infiltration by inflammatory cells and the degree of proliferation of emarginated synoviocytes were scored as follows: 0 = none, 1 = mild, 2 = moderate, and 3 = severe. Regarding infiltration by inflammatory cells, the scoring reflected the number of mononuclear cells within the synovial membrane (none, only few scattered cells; mild, a limited number of cells; moderate, scattered areas with accumulation of cells; and severe, large areas with diffuse infiltrations of cells). Regarding proliferation of emarginated synoviocytes, the scoring reflected the number of synoviocytes on the membrane: score 0 = scattered low synoviocytes on the synovial membrane, score 1 = 1 or 2 layers of synoviocytes, score 2 = 3 or 4 layers of synoviocytes, and score 3 = ≥ 5 layers of synoviocytes.

Statistical methods—The scoring system for histopathologic evaluation of inflammation of the synovium that we developed was assessed with agreement analysis by use of WK to take into account the magnitude of the discrepancy of the ordered categories.^29–37 The weights 1, 0.67, 0.33, and 0 were used. The intraobserver agreement reflected the repeatability of scoring the same sample twice by the same pathologist, and the interobserver agreement reflected the repeatability of scoring the same sample by each of 2 pathologists. The strength of agreement expressed as WK was graded as poor (0.00 to 0.20), fair (0.21 to 0.40), moderate (0.41 to 0.60), good (0.61 to 0.80), or very good (0.81 to 1.00).²⁹

Mean values for proliferation of emarginated synoviocytes were calculated in control and OCD-affected horses. The P value for difference was based on a linear regression with group (control vs OCD-affected) as an independent variable, along with the age of the horses, to take into account the difference in age between groups. These estimations were calculated separately for pathologist and observation. The cluster option was used to account for > 1 joint within a horse. Similar analyses were performed to assess the relationship between joint effusion or lameness and positive results for a flexion test with proliferation of emarginated synoviocytes or infiltration by inflammatory cells. Adjustments were made for age, reproductive status, and breed. In all multiple regression analyses, the scored findings (0 to 4 for joint effusion and 0 to 5 for lameness) were used as the dependent variables.

In a subgroup, it was assessed whether the interval between death and the collection of samples of synovial tissue influenced the results. In these analyses, the time variable was logarithmically transformed to become sufficiently close to a normal distribution.

Because 2 OCD-affected horses represented a single breed and 9 control horses were older (> 9 years of age) than the OCD-affected horses, analyses were done without these horses. The results were similar to the results without the exclusion of these horses. Therefore the results were reported including all horses.

Because of age variation (1 to 9 years) in the group of OCD-affected horses, age was included in the model. Horses with OCD of the MM of the tibia in addition to OCD of the IRT were not analyzed separately because the number of horses in this group was too small. A value of P < 0.05 was considered significant.

Results

Examination of the first 100 synovial samples from 79 horses included study specimens from 75 TCJs of 54 (68%) horses affected with OCD of the IRT and specimens from 25 (32%) control horses. The number of control and study specimens in the first samples reflected the distribution of horses in the study population: 93 (71%) OCD-affected horses and 38 (29%) control horses.

Reasons for euthanasia of 35 of the 38 control horses included the following: fracture (n = 13), cartilage damage to a joint other than the TCJ (9), tendon injury (3), rupture of a cruciate ligament (3), ataxia (2), laminitis (2), infection of a joint other than the TCJ (2), and recurrent airway obstruction (1). Two horses died as a result of cardiovascular collapse during anesthesia, and 1 horse died as a result of rupture of the aorta. The findings at necropsy did not exclude any of the control horses from the study.

The flexion test was omitted during examination of 8 OCD-affected horses (10 affected TCJs) because the horses were difficult to handle. One synovial biopsy specimen from 1 TCJ of a horse affected with bilateral OCD of the TCJ was lost during transport of the biopsy specimen to the laboratory.

The mean ± SEM age of the 93 OCD-affected horses was 2.8 ± 0.2 years (range, 1 to 9 years). Mean age of the 29 control horses was 4.2 ± 0.4 years (range, 1 to 9 years). Nine control horses were > 9 years old. The breed distribution among the 93 OCD-affected horses was 55 (59%) Swedish Warmbloods, 36 (39%) Standardbreds, 1 (1%) Andalusian, and 1 (1%) American Paint Horse. The breed distribution among the 38 control horses was 21 (55%) Swedish Warmbloods, 3 (8%) Standardbreds, 8 (21%) Thoroughbreds, 2 (5%) Icelandic horses, 2 (5%) Welsh ponies, 1 (3%) Coldblooded Trotter, and 1 (3%) American Paint Horse.

Of the 93 OCD-affected horses, 42 (45%) were mares, 29 (31%) were stallions, and 22 (24%) were geldings. Of the 38 control horses, 15 (39%) were mares, 7 (18%) were stallions, and 16 (42%) were geldings.

Of 93 horses with OCD of the IRT, 28 (30%) were affected in the left pelvic limb, 24 (26%) were affected in the right pelvic limb, and 41 (44%) were affected bilaterally. Of specimens from the 38 control horses, 15 (39%) were collected from the left pelvic limb and 23 (61%) were collected from the right pelvic limb. Concurrent OCD of the MM of the tibia was identified in 7 (8%) affected horses, 1 of which was bilaterally affected. No horse in the study had OCD that affected both the IRT and the LTR of the talus.

Of the 134 TCJs with OCD of the IRT, 112 had synovial effusion. Twenty-two TCJs had no effusion (grade 0), 59 had mild effusion (grade 1), 32 had moderate effusion (grade 2), 19 had severe effusion (grade 3), and 2 had extreme effusion (grade 4).

Twenty-seven (29%) horses with 30 affected TCJs were lame in the affected limb or limbs. Fourteen of the 27 (52%) lame horses became temporarily sound in the affected limb or limbs after injection of 10 mL of local anesthetic solution into the affected TCJs. Three horses were bilaterally lame, and 2 of these were judged to have a lameness score of 0.5 for each pelvic limb. One horse was assigned a lameness score of 0.5 for 1 pelvic limb and a score of 1 for the contralateral pelvic limb. The lameness of all 3 horses was resolved by anesthesia of the TCJ of the affected limb. Intra-articular anesthesia of the TCJ of the lamer limb revealed lameness in the contralateral pelvic limb. Of the 24 unilaterally lame horses, 12 (7 with a lameness score of 0.5, 4 with a lameness score of 1, and 1 with a lameness score of 2) had OCD of the IRT in the TCJ of the contralateral limb but were not lame in that limb. Thirty-three horses (with 38 affected TCJs) were lame, had positive results of pelvic limb flexion, or both.

Examination of the first 100 of the 172 synovial samples by pathologists A and B determined that the synovium of OCD-affected horses (n = 54) and the synovium of control horses (25) did not differ with regard to 4 markers of inflammation (proliferation of scattered synoviocytes, perivascular infiltration by inflammatory cells, focal mineralization, and villus proliferation). These markers were not analyzed further. Scorings of the degree of proliferation of emarginated synoviocytes and extent of infiltration by inflammatory cells (Figure 1) were determined for all samples of synovium, and these 2 descriptive markers were selected for intra- and interobserver agreement evaluations.

View Full Size

Representative photomicrographs of tissue sections of synovium from a TCJ of each of 3 horses with OCD of the IRT. A—Notice the scattered low synoviocytes on the synovial membrane (arrows; score 0) and that almost no inflammatory cells have infiltrated the membrane (score 0). B—Areas with 3 layers of synoviocytes are evident on the synovial membrane (long arrows; score 2) and a moderate number of inflammatory cells have infiltrated the membrane (short arrow; score 1). C—Areas with ≥ 5 layers of synoviocytes are evident on the synovial membrane (long arrows; score 3) and numerous inflammatory cells have infiltrated this membrane (short arrow; score 3). H&E stain; bar = 50 μm.

Citation: American Journal of Veterinary Research 71, 1; 10.2460/ajvr.71.1.47

• Download Figure
• Download figure as PowerPoint slide

Based upon assessment of 93 OCD-affected horses and 38 control horses, analysis of intraobserver agreement from 2 evaluations by each pathologist resulted in mean ± SEM WK of 0.76 ± 0.05 to 0.81 ± 0.06 within pathologist A and WK of 0.56 ± 0.05 to 0.63 ± 0.06 within pathologist B. Generally, there was higher agreement when scoring infiltration by inflammatory cells than when scoring proliferation of emarginated synoviocytes.

Analysis of interobserver agreement resulted in mean ± SEM WK of 0.34 ± 0.04 to 0.52 ± 0.05 when scoring synoviocyte proliferation and WK of 0.38 ± 0.06 to 0.48 ± 0.05 when scoring infiltration by inflammatory cells (Table 1). Exclusion of the control horses > 9 years old and the breeds represented by a single horse yielded no substantial difference in the calculated WK values.

Mean interval from death to the collection of samples from synovial tissue was recorded in 38 control horses. Mean interval was 376 minutes (range, 5 minutes to 6 days). Half of the tissue samples were obtained within 17.5 minutes after death. The logarithmically transformation gave a Shapiro-Wilk W of 0.88. The logarithmically transformed time interval was not associated with scores for proliferation of synoviocytes or infiltration by inflammatory cells. This applied to both observations and to both pathologists.

When scores for proliferation of emarginated synoviocytes in the TCJs of the OCD-affected horses were compared with those of control horses, a significantly (P = 0.03) higher proliferation of emarginated synoviocytes was identified in the TCJ of the OCD horses by only pathologist A at the first of the 2 evaluations (Table 2). Infiltration of the synovium by inflammatory cells in the TCJs of OCD-affected horses was significantly higher than that of control horses for both observations and for both pathologists.

Histologic signs of inflammation in the synovium of OCD-affected horses were compared with clinical signs of joint inflammation in those horses. A significant association was identified between joint effusion and proliferation of emarginated synoviocytes and between joint effusion and infiltration by inflammatory cells (Table 3). No significant (P = 0.13 to 0.86) association was identified between clinical signs of lameness and histologic changes in the synovial membrane when histologic signs of synovial inflammation in OCD-affected horses that were lame, that reacted to a flexion test, or that were lame and reacted to a flexion test were compared (Table 4).

No significant differences were observed for lameness and reaction to flexion testing between horses with both OCD of the IRT and OCD of the MM of the tibia and those with only OCD of the IRT. Evidence of OCD of the MM of the tibia did not influence the results of the histopathologic scoring.

Discussion

To describe and evaluate OCD-associated histopathologic changes in the synovial membrane of the TCJ in horses and to compare these changes with clinical signs of lameness, we developed and tested a system of scoring those histopathologic changes. Such a scoring system for grading the synovium in horses has not been developed to our knowledge. Several scoring systems that characterize inflammation in the synovium have been published,^{14,15,19–28} but criteria used by those systems are vague and do not attempt to quantify or grade the inflammatory changes. To increase reliability and strength of the scoring system, we also included control horses that were euthanatized for reasons other than disease of the TCJ. Mean age of horses in the control group was somewhat higher than that of the group of OCD-affected horses because younger control horses that met the selection criteria were difficult to obtain. However, we believed that the older control horses were representative of the population of horses with normal TCJ synovium. Horses that were euthanatized because of disease unrelated to the TCJs could not be matched perfectly with the relatively young population of OCD-affected horses, but the data were statistically adjusted for any confounding effect of age, breed, or reproductive status.

The scoring system was developed by examining the first 100 samples of synovial membrane. Horses from which these samples were obtained were enrolled in the study in chronological order and represented a random sample because OCD-affected horses with differing severity of clinical signs undergoing arthroscopic surgery, as well as control horses, were admitted to the hospital in an arbitrary order.

Although it is difficult to develop and standardize a scoring system for histopathologic changes in equine synovium, scoring inflammatory changes in the joint and comparing those changes with clinical signs is relevant to treatment and prognosis. Our scoring system was based on a few relatively simple criteria developed so that histologic changes in the synovium can be easily and practically quantified by other researchers and clinicians. We determined that infiltration by inflammatory cells and proliferation of emarginated synoviocytes were the 2 most dominant and consistent markers for inflammation of equine synovium. In studies^14,15 of horses with mild synovitis caused by intra-articular fractures and osteochondrosis of various joints, these 2 signs of inflammation have been used to characterize the histopathologic changes in the affected joints.

Synovium of horses with OCD of the IRT and synovium of control horses did not differ with regard to 4 markers of inflammation (proliferation of scattered synoviocytes, perivascular infiltration by inflammatory cells, focal mineralization, and villus proliferation). Other investigators evaluating synovial inflammation in cattle,²⁴ dogs,^25–28 and humans^19–23 also determined that these markers were inadequate to describe pathological changes in the synovium.

In the present study, the scoring system was evaluated to determine intraobserver and interobserver agreements. Each agreement was assessed by use of WK to take into account the magnitude of the discrepancy of the ordered categories.^29–37 This is a chance-corrected index of agreement that penalizes interobserver disagreement and is appropriate for testing a classification scheme that yields categorical or ordinal data.³⁰ Intraobserver agreement was good to very good, and interobserver agreement was moderate to good. Agreement was greater for scoring of infiltration by inflammatory cells than it was for scoring of proliferation of emarginated synoviocytes. One of the intraobserver WK values reached 0.81, which some researchers consider to be almost perfect agreement.³³ These results were considered acceptable for further analysis and comparison with clinical signs.

To minimize errors in evaluation, 1 clinician (PB) evaluated all horses and performed all surgeries. Conducting a blinded investigation would have been difficult because we believed that signs of previous arthroscopic surgery of the TCJ would be easily detected. To eliminate this bias, the lameness examination could have been video recorded to provide a double-blinded study, but this was not feasible with our available resources. Other studies^38–41 have revealed that the results of evaluation of lameness, both before and after flexion tests, are highly variable among observers but are relatively constant within the same observer. The flexion test can also yield false-positive reactions in sound horses.⁴² Although the examinations were conducted in a nonblinded manner, the large number of horses that underwent surgery and were examined postoperatively by the same veterinarian likely provided data that were sufficiently consistent to be compared with the histopathologic scoring.

When synoviocyte scores for OCD-affected and control horses were compared, only 1 pathologist identified a significantly higher score for proliferation of emarginated synoviocytes in OCD-affected horses. The inflammatory cell score for OCD-affected horses was significantly higher than that of control horses for all observations, which indicated that infiltration by inflammatory cells might be the most reliable indicator of inflammation. This is in agreement with findings of studies^{14,15,20–28} in which investigators used the degree of infiltration by inflammatory cells as a guideline for scoring synovial inflammation. In 1 of those studies,²⁰ investigators determined that proliferation of emarginated and palisading synoviocytes were useful descriptors of synovial inflammation.

Of the 93 horses with OCD of the IRT, only 2 horses were not Standardbreds or Swedish Warmbloods, which prevents extrapolation of the results to all breeds of horses. The right (26%) and left (30%) TCJs were equally affected, and 44% of affected horses had bilateral OCD of the IRT of the TCJ.

Joint effusion was highly associated with proliferation of emarginated synoviocytes and infiltration by inflammatory cells. We were able to use linear regression to investigate this association because the variables, although not normally distributed as determined by use of the Shapiro-Wilk W test (W = 0.97), were considered sufficiently similar to a normal distribution for such an analysis. However, no significant association was found between lameness or positive reaction to the flexion test and proliferation of emarginated synoviocytes or infiltration by inflammatory cells. Lameness resolved temporarily in all lame horses that received local anesthetic solution in the affected TCJ, but this diagnostic test was only performed in half of the lame horses before surgery.

No significant association was found between histopathologic scores and the clinical signs evaluated (lameness, reaction to a flexion test, or both). The regression was based on logarithmically transformed variables because they were normally distributed (Shapiro-Wilk W = 0.99), but the statistical power to detect an association of the observed magnitude was only 0.64. Power calculations suggested that doubling the sample size would increase power to 0.91.

The mean interval to postmortem sample collection of the TCJ synovium from the control horses was 6.3 hours (mode, 5 minutes; median, 18 minutes). No significant association was found between interval to postmortem sample collection and the histopathologic score. The lack of influence of interval to sample collection indicated that the sampling protocol was adequate for the study; however, samples should be collected as soon after death as possible.

In the study reported here, a valid and reliable scoring system for assessment of inflammation of equine synovium was developed and was useful in evaluating TCJ inflammation associated with OCD of the IRT in horses. Histopathologic signs of synovial inflammation correlated with joint effusion but not with lameness.