The metacarpophalangeal joint is the most common site of musculoskeletal disease that leads to reduced performance, premature retirement, catastrophic failure, and euthanasia of Thoroughbred racehorses throughout the world.1–5 Condylar fractures of the MC3 are the most common reason for euthanasia of Thoroughbred racehorses in the United Kingdom3 and the second most common reason for euthanasia in North America, after biaxial fracture of the proximal sesamoid bones.2,4
A review of jockey injuries and fatalities associated with Thoroughbred racing identified that 54% of horse falls injure a human.6 The 2017 fatal injury rate for horses in the United States was 1.61 horses/1,000 starts.a If 10 horses/race were to compete in 10 races/d, there would be 1,000 starts in 10 days and, on average, 2 horses would be euthanized and 1 human injured every 10 days. This hypothetical example underscores the urgent need to develop additional surveillance strategies to prevent fractures in racehorses.
Bone is a solid material that can change its internal architecture when stress is imposed (Wolff law).5,7–29 By adapting to increased loads through remodeling, an increase in bone mineral density enables greater tolerance to the high demands of racing.7–9 These physiologic changes are normal and essential if bone is to withstand the increased forces associated with racing. Overuse injuries to bone, referred to as BSI,8–14 are common and disproportionately affect Thoroughbred racehorses.1–5,15–19 These physiologic processes can become pathological if physical stresses continue on a BSI that has not had sufficient time to heal.15–19 Bone marrow lesions provide evidence of a developing BSI at the cellular level and are the earliest opportunity for identification of a BSI by use of advanced imaging.8–14
In a 1992 study,18 the presence of periosteal new bone in acute fractures in 10 of 13 horses with catastrophic humeral fracture provided evidence of chronic preexisting conditions and a timeline for fracture development. This led to the understanding that early identification of the components of a developing fracture provide an opportunity for fracture prevention. Recognizing that a condylar fracture is the culminating event of a BSI that began in the MC3 (or MT3), sMRI has potential for use in the early identification of condylar fractures by noninvasively providing 3-D images of bone architecture, which can be used to identify BMLs.23–25
The purpose of the study reported here was to identify early warning signs of fracture by use of sMRI images of the MC3 of forelimbs obtained from Thoroughbred racehorse cadavers, to grade these bone changes, and to compare grades in racehorses with and without catastrophic condylar fracture. Our hypothesis was that bone changes would be more severe when the fractured limb was compared with the nonfractured contralateral limb in horses with catastrophic condylar fracture (cases) and when the fractured limbs of cases were compared with the limbs of control horses without catastrophic condylar fracture.
Materials and Methods
Sample
Both forelimbs of Thoroughbred racehorses euthanized at racetracks in southern Florida from September 15, 2011, until September 14, 2014, were harvested. A case was defined as a horse euthanized following a complete, displaced forelimb (MC3) condylar fracture. The limb with the condylar fracture was referred to as the fractured limb, and the contralateral unaffected limb was referred to as the nonfractured limb. A control horse was defined as a horse euthanized for any condition not associated with the metacarpophalangeal joint (eg, enteritis or fracture of a hind limb). The distal portion of each study forelimb was packaged by 1 investigator (PAM) and shipped cooled but not frozen to the Equine Medical Center of Ocala, which thereby avoided the effects of a freeze-thaw cycle on the tissues.
sMRI evaluation
The sMRI images typically were obtained ≤ 48 hours and always < 72 hours after horses were euthanized. Images were obtained with a 0.27-T equine sMRI systemb by 1 investigator (LH). Images were collected of both forelimbs for case and control horses, but only a single forelimb of control horses was randomly selected (simulated coin flip for each pair of limbs) for analysis. Information on race statistics was collected from an equine database.c
Images (T1W, T2*W, and STIR) were collected for all limbs. One investigator (JGP) assessed all sMRI images. Dense bone was defined as an area with a decreased signal intensity on T1W, T2*W, and STIR MRI images. A BML, previously referred to as bone marrow edema, was defined as an area of decreased signal intensity on T1W images and increased signal intensity on STIR and T2*W images. The increased signal intensity on STIR (and T2*W is implied) images was used to describe areas with increased amounts of fluid (BML, hemorrhage, necrosis, and inflammation) when compared with results for physiologically normal bone. Because objective assessment was not possible, a binary grading system (present or absent) was used to subjectively characterize BML in either condyle for STIR MRI images (Figure 1).
A transverse STIR MRI image of the distal aspect of the MC3 of a control horse (A) and transverse (B) and frontal (C) STIR MRI images of a horse with a condylar fracture of the MC3. A binary grading system was used to subjectively classify a BML as present or absent. In panels B and C, notice the condylar fracture (white arrow) and generalized accumulation of a BML (red arrows). In the control horse, the medullary cavity is distinctly black owing to a lack of a BML (categorized as absent).
Citation: American Journal of Veterinary Research 80, 2; 10.2460/ajvr.80.2.178
A transverse STIR MRI image of the distal aspect of the MC3 of a control horse (A) and transverse (B) and frontal (C) STIR MRI images of a horse with a condylar fracture of the MC3. A binary grading system was used to subjectively classify a BML as present or absent. In panels B and C, notice the condylar fracture (white arrow) and generalized accumulation of a BML (red arrows). In the control horse, the medullary cavity is distinctly black owing to a lack of a BML (categorized as absent).
Citation: American Journal of Veterinary Research 80, 2; 10.2460/ajvr.80.2.178
A transverse STIR MRI image of the distal aspect of the MC3 of a control horse (A) and transverse (B) and frontal (C) STIR MRI images of a horse with a condylar fracture of the MC3. A binary grading system was used to subjectively classify a BML as present or absent. In panels B and C, notice the condylar fracture (white arrow) and generalized accumulation of a BML (red arrows). In the control horse, the medullary cavity is distinctly black owing to a lack of a BML (categorized as absent).
Citation: American Journal of Veterinary Research 80, 2; 10.2460/ajvr.80.2.178
The percentage of the volume of dense bone for each condyle (defined as the DBVP) was measured (Figure 2). Surface area measurements were calculated with DICOM software.d By use of sagittal T1W images, surface area measurements of dense bone were calculated by tracing the cross-sectional area of a decrease in signal intensity after verifying that this same area also had a decrease in signal intensity on T2*W and STIR images. This measurement was normalized by use of a cross-sectional area measurement that accounted for variability of each horse. The second surface area measurement started at the most proximal palmar aspect of the distal part of MC3, where the curvature of the SCB plate changed abruptly from a convex to a concave shape. A line was drawn dorsally from this point toward the dorsal aspect of MC3 such that the line was perpendicular to the long axis of MC3. The line was continued along the SCB plate until the line intersected with the point of origin. The sum of the 4 measurements of the dense bone surface area of each condyle was divided by the sum of the 4 measurements of the normalizing distal epiphysis surface area to calculate DBVP for the medial and lateral condyles.
A T1W sagittal image of the distal aspect of the MC3 in the limb of a control horse. A surface area measurement of the DBVP was calculated by tracing the cross-sectional area of dense bone (identified as a decrease in signal intensity; green outline). This measurement was normalized by use of a second surface area measurement (blue outline), which started at the most proximopalmar aspect of the distal portion of MC3 where the curvature of the SCB plate changed abruptly from a convex shape to a concave shape. A line perpendicular to the long axis of MC3 was drawn from this point toward the dorsal aspect of MC3. The line was continued distally and then in a palmar direction along the black line of the SCB plate until the line intersected with the point of origin. The sum of the 4 medial condylar measurements (surface area measurements of dense bone) was divided by the sum of the 4 medial condylar measurements (second surface area measurements) to determine the DBVP of the distopalmar aspect of the medial condyle. The same procedures were used to calculate the DBVP of the lateral condyle.
Citation: American Journal of Veterinary Research 80, 2; 10.2460/ajvr.80.2.178
A T1W sagittal image of the distal aspect of the MC3 in the limb of a control horse. A surface area measurement of the DBVP was calculated by tracing the cross-sectional area of dense bone (identified as a decrease in signal intensity; green outline). This measurement was normalized by use of a second surface area measurement (blue outline), which started at the most proximopalmar aspect of the distal portion of MC3 where the curvature of the SCB plate changed abruptly from a convex shape to a concave shape. A line perpendicular to the long axis of MC3 was drawn from this point toward the dorsal aspect of MC3. The line was continued distally and then in a palmar direction along the black line of the SCB plate until the line intersected with the point of origin. The sum of the 4 medial condylar measurements (surface area measurements of dense bone) was divided by the sum of the 4 medial condylar measurements (second surface area measurements) to determine the DBVP of the distopalmar aspect of the medial condyle. The same procedures were used to calculate the DBVP of the lateral condyle.
Citation: American Journal of Veterinary Research 80, 2; 10.2460/ajvr.80.2.178
A T1W sagittal image of the distal aspect of the MC3 in the limb of a control horse. A surface area measurement of the DBVP was calculated by tracing the cross-sectional area of dense bone (identified as a decrease in signal intensity; green outline). This measurement was normalized by use of a second surface area measurement (blue outline), which started at the most proximopalmar aspect of the distal portion of MC3 where the curvature of the SCB plate changed abruptly from a convex shape to a concave shape. A line perpendicular to the long axis of MC3 was drawn from this point toward the dorsal aspect of MC3. The line was continued distally and then in a palmar direction along the black line of the SCB plate until the line intersected with the point of origin. The sum of the 4 medial condylar measurements (surface area measurements of dense bone) was divided by the sum of the 4 medial condylar measurements (second surface area measurements) to determine the DBVP of the distopalmar aspect of the medial condyle. The same procedures were used to calculate the DBVP of the lateral condyle.
Citation: American Journal of Veterinary Research 80, 2; 10.2460/ajvr.80.2.178
An established MRI grading scale for POD4 was used to evaluate representative sagittal T1W sMRI images of the medial and lateral condyles. Briefly, this grading system focused on the integrity of the SCB plate and bone in the area of contact between the proximal sesamoid bone and the palmar condyles of MC3 during the stance phase of the gallop. Changes in the SCB of the distopalmar aspect of the medial and lateral condyles of MC3 were graded on a scale of 0 (normal) to 5 (markedly abnormal) as follows: grade 0 = no abnormalities identified in the SCB plate or the signal in the SCB subjacent to the SCB plate; grade 1 = normal signal intensity and morphology of the SCB plate, but an increase in densification of the subjacent medullary bone marrow and a linear area of hyperintensity immediately subjacent to the SCB plate; grade 2 = abnormal hyperintensity of the SCB plate and a contour deformity that was suggestive of focal compression or fracturing of the SCB plate or the subjacent trabecular bone, and an increase in the volume of the area of increased densification in the subjacent SCB marrow; grade 3 = full-thickness defect in the SCB plate with a signal intensity similar to that of joint fluid (which implied that the overlying articular cartilage was not intact), and an increase in densification of the subjacent SCB marrow; grade 4 = partial or complete detachment of an osteochondral fragment that was not displaced, and an increase in densification of the subjacent SCB marrow; and grade 5 = the osteochondral fragment identified in grade 4 was displaced from the deeper SCB to leave a large area of exposed SCB of MC3, and an increase in densification of the subjacent SCB marrow.
A condylar fissure has been defined as an abnormality of the distal surface of MC3 (or MT3) within the parasagittal groove.27,28 The MRI inclusion criteria for a condylar fissure was the presence of a hyperintense short incomplete line on T1W images that originated from the palmar aspect of either the medial or lateral parasagittal grove of the distal portion of MC3 and extended a short distance dorsally and proximally into the cancellous bone with associated hyperintensity on STIR and T2*W images. Blood vessels were identified infrequently as distinct linear structures that originated proximal and palmar in close proximity to the physeal remnant and radiated in a circuitous pattern. The MRI inclusion criteria of a blood vessel was hyperintensity of this distinct linear structure on T1W, T2*W, and STIR images.
Statistical analysis
Comparisons of continuous or ordinal variables between the 26 pairs of fractured and nonfractured limbs from case horses were made by use of the McNemar test and conditional logistic regression for density values. Comparisons of continuous or ordinal variables between the 26 fractured limbs from case horses and a single limb from the 88 control horses were made by use of the Wilcoxon rank sum test, and comparisons of categorical variables were made by use of the χ2 test or, when expected values were < 5 for any given cell, Fisher exact test. In addition, associations between the binary outcome variable of case or control horse with continuous, ordinal, and categorical variables were quantified by use of logistic regression analysis with 95% CIs estimated by use of maximum likelihood methods. Comparisons of condylar density measures between case and control horses were made by use of generalized linear regression modeling with a Gaussian link. Post hoc testing among pairs of groups was made with the Sidak method. For all regression analyses, model fits were assessed with diagnostic plots of residuals. Significance for all analyses was set at P < 0.05. All analyses were performed with statistical software.e
Results
Sample
During the study, 114 horses met the criteria for enrollment. There were 26 fractured and 26 nonfractured limbs from horses classified as cases and 88 limbs (36 of the left forelimb and 52 of the right forelimb) selected from horses classified as controls. Of the 26 fractured limbs of case horses, 17 involved the left forelimb (8 fractures of the lateral condyle and 9 fractures of the medial condyle) and 9 involved the right forelimb (8 fractures of the lateral condyle and 1 fracture of the medial condyle); thus, there were 16 fractures of the lateral condyle and 10 fractures of the medial condyle.
Age, sex, number of starts, number of wins, and race distance did not differ significantly between case and control horses (Table 1). The likelihood that the fractured limb was the left forelimb (OR, 2.7; 95% CI, 1.1 to 6.8) was significantly (P = 0.033) greater than for the control limbs. This association of the left forelimb fracture when compared to the control limbs was perhaps simply attributable to a chance distribution of our coin flip method for selecting control limbs. Comparing the observed distribution with that expected by chance alone (fractures in 50% of the right forelimbs and 50% of the left forelimbs) revealed that the likelihood of the left forelimb being associated with injury (OR, 1.9; 95% CI, 0.8 to 4.7) was not significant (P = 0.247).
Characteristics of case and control horses for a study of catastrophic condylar fracture with bony changes of the third metacarpal bone identified by use of sMRI in forelimbs from cadavers of US Thoroughbred racehorses.
| Variable | Case horses (n = 26) | Control horses (n = 88) | P value* |
|---|---|---|---|
| Age (y) | 3 (2–5) | 3 (2–8) | 0.158 |
| Sex | |||
| Female | 11 (42) | 48 (55) | Referent |
| Gelding | 5 (19) | 16 (18) | 0.613 |
| Male | 10 (39) | 24 (27) | 0.233 |
| Sex | |||
| Females | 11 (42) | 48 (55) | Referent |
| All males | 15 (58) | 40 (45) | 0.277 |
| Race starts | 5 (0–33) | 6.5 (0–56) | 0.3014 |
| Race wins | 0 (0–5) | 1 (0–12) | 0.3196 |
| Distance of final race (No. of furlongs) | 6.5 (4.0–12) | 6.75 (4–8.5) | 0.3972 |
| Limb† | |||
| Left | 17 (65) | 36 (41) | Referent |
| Right | 9 (35) | 52 (59) | 0.033 |
Data are reported as median (range) or as number (%) of horses.
P values were derived by use of logistic regression analysis; values were considered significant at P < 0.05.
Represents the limb with the condylar fracture in case horses and a randomly selected limb for the control horses.
BML
A BML was identified as an increase in the STIR signal detected in association with the condylar fracture in 26 of 26 (100%) fractured limbs (Figure 1). In 21 of 26 (81%) fractured limbs, the BML was located toward the periphery of the dense bone, and in 5 of 26 (19%) fractured limbs, the signal increase was located centrally within the dense bone (Figure 3). In all fractured limbs, the BML was located proximally in the epiphysis over the parasagittal groove and radiated in the parasagittal plane, which made the 90° orientation of frontal plane MRI images superior for lesion identification.
Sagittal (A and D), transverse (B and E), and frontal (C and F) STIR MRI images of the distal aspect of the MC3 in 2 horses with a condylar fracture. The condylar fracture (white arrow) and generalized accumulation of a BML (red arrows) are indicated. In 21 of 26 (81%) fractured limbs, the BML was located toward the periphery of the dense bone (A, B, and C), and in 5 of 26 (19%) fractured limbs, the signal increase was located centrally within the dense bone (D, E, and F).
Citation: American Journal of Veterinary Research 80, 2; 10.2460/ajvr.80.2.178
Sagittal (A and D), transverse (B and E), and frontal (C and F) STIR MRI images of the distal aspect of the MC3 in 2 horses with a condylar fracture. The condylar fracture (white arrow) and generalized accumulation of a BML (red arrows) are indicated. In 21 of 26 (81%) fractured limbs, the BML was located toward the periphery of the dense bone (A, B, and C), and in 5 of 26 (19%) fractured limbs, the signal increase was located centrally within the dense bone (D, E, and F).
Citation: American Journal of Veterinary Research 80, 2; 10.2460/ajvr.80.2.178
Sagittal (A and D), transverse (B and E), and frontal (C and F) STIR MRI images of the distal aspect of the MC3 in 2 horses with a condylar fracture. The condylar fracture (white arrow) and generalized accumulation of a BML (red arrows) are indicated. In 21 of 26 (81%) fractured limbs, the BML was located toward the periphery of the dense bone (A, B, and C), and in 5 of 26 (19%) fractured limbs, the signal increase was located centrally within the dense bone (D, E, and F).
Citation: American Journal of Veterinary Research 80, 2; 10.2460/ajvr.80.2.178
A BML was detected significantly more frequently in the fractured limb of case horses (26/26 [100%]) than in the nonfractured limb of case horses (7/26 [27%]). Condylar fractures were significantly associated with BML (7 case horses had a BML in both the fractured and nonfractured limb, and 19 case horses had a BML in only the fractured limb). Evaluation of the fractured limbs revealed that a BML was found in the lateral condyle in 15 of 16 lateral condylar fractures, the medial condyle in 10 of 10 medial condylar fractures, and the medial condyle of 1 lateral condylar fracture. Evaluation of the nonfractured limb of case horses revealed that a BML was identified in the lateral condyle in 5, the medial condyle in 2, and both the lateral and medial condyles in 1.
When comparing the fractured limbs of case horses with the limbs of control horses, a BML was significantly more common in limbs of case (26/26 [100%]) than control (6/88 [7%]) horses. The OR could not be calculated because all case horses had a BML. Sensitivity of a BML for identifying a condylar fracture was 100% (26/26; 95% CI, 87% to 100%), and specificity was 93% (82/88; 95% CI, 86% to 97%).
DBVP
Mean DBVP did not differ significantly between fractured and nonfractured limbs of case horses for the lateral or medial condyle (Table 2). Fractured and nonfractured limbs of case horses were stratified for analysis on the basis of whether the fracture was located in the lateral or medial condyle. For comparison of fractured and nonfractured limbs of case horses, when the fracture was in the lateral condyle (n = 16), the DBVP was significantly higher in the lateral than the medial condyle, and when the fracture was in the medial condyle (10), the DBVP was significantly higher in the medial than the lateral condyle (Table 3).
Mean (95% CI) percentage values for DBVP of fractured and nonfractured condyles in limbs of case horses.
| Condyle | Fractured limb (n = 26) | Nonfractured limb (n = 26) | P value |
|---|---|---|---|
| Lateral | 25.9 (18.5–33.4) | 21.2 (15.0–27.3) | 0.197 |
| Medial | 25.2 (19.8–30.5) | 19.8 (13.2–26.4) | 0.052 |
| Difference* | 0.8 | 1.3 | 0.917 |
P values were derived by use of linear mixed-effects modeling to account for within-horse correlations. Values were considered significant at P < 0.05.
Represents the difference between the lateral and medial condyles.
Mean (95% CI) percentage values for DBVP of fractured and nonfractured condyles in limbs of case horses, with density for fractured condyles stratified on the basis of fracture location.
| Condyle | Lateral fractured (n = 16) | Medial fractured (n = 10) | Nonfractured (n = 26) |
|---|---|---|---|
| Lateral | 32.2 (22.5–41.9)a | 15.9 (9.1–22.7)b | 21.2 (15.0–27.3)b |
| Medial | 14.7 (10.2–19.2)a | 41.9 (26.7–57.1)b | 19.8 (13.2–26.4)a |
| Difference* | 15.6a | 26.4b | 1.3c |
Within a row, values with different superscript letters differ significantly (P < 0.05); P values were derived by use of linear mixed-effects modeling to account for within-horse correlations.
See Table 2 for remainder of key.
Mean DBVP was significantly higher in both the lateral and medial condyles when the fractured limb of case horses was compared with limbs of control horses (Table 4). When the fractured limb of case horses was stratified on the basis of the location of the condylar fracture (n = 16 for lateral and 10 for medial), the DBVP was significantly higher in the fractured limb of case horses for either the lateral or medial condyle, compared with the DBVP for the limbs of control horses (Table 5).
Mean (95% CI) percentage values for DBVP of condyles in fractured limbs of case horses and limbs of control horses.
| Condyle | Fractured limb (n = 26) | Control limb (n = 88) | P value |
|---|---|---|---|
| Lateral | 25.9 (19.0–32.9) | 16.2 (14.3–18.1) | 0.002 |
| Medial | 25.2 (17.2–33.2) | 18.1 (15.5–20.8) | 0.042 |
Values were derived by use of linear mixed-effects modeling to account for within-horse correlations. Values were considered significant at P < 0.05.
Mean (95% CI) percentage values for DBVP of condyles in fractured and nonfractured limbs of case horses and limbs of control horses, with density for fractured condyles stratified on the basis of fracture location.
| Condyle | Lateral fractured (n = 16) | Medial fractured (n = 10) | Control (n = 88) |
|---|---|---|---|
| Lateral | 32.2 (22.5–41.9)a | 15.9 (9.1–22.7)b | 16.2 (14.3–18.1)b |
| Medial | 14.7 (10.2–19.2)a | 41.9 (26.7–57.1)b | 18.1 (15.5–20.8)a |
Within a row, values with different superscript letters differ significantly (P < 0.05); P values were derived by use of linear mixed-effects modeling to account for within-horse correlations.
For fractured limbs, the mean DBVP was 25.9% for the lateral condyle and 25.2% for the medial condyle; therefore, the diagnostic sensitivity and specificity of a DBVP cutoff of > 25% for identification of a condylar fracture of the MC3 were calculated. The DBVP in the lateral condyle was > 25% for 10 of 16 limbs with a lateral condylar fracture and ≤ 25% for 6 of 10 limbs with a medial condylar fracture and for 64 of 88 control limbs. Thus, the diagnostic sensitivity and specificity of a DBVP > 25% for detection of a lateral condylar fracture were 63% (10/16) and 71% (70/98), respectively. The DBVP in the medial condyle was > 25% for 7 of 10 limbs with medial condylar fracture and ≤ 25% for 15 of 16 limbs with a lateral condylar fracture and for 70 of 88 control limbs. Thus, the diagnostic sensitivity and specificity of a DBVP > 25% for detection of a medial condylar fracture were 70% (7/10) and 82% (85/104), respectively.
POD
The signal increase associated with POD was evident in close proximity to the SCB plate, palmar to the transverse ridge, and radiated in a frontal plane, which made the 90° orientation of sagittal plane MRI images superior for lesion identification. Most POD lesions were not graded as severe (ie, ≤ 2), so POD was categorized as present or absent for analysis. None of the 16 horses with lateral condylar fractures had POD in the lateral condyle, and none of the 10 horses with medial condylar fractures had POD in the medial condyle. For case horses, although the proportion of fractured limbs with POD (1/26 [4%]) was smaller than the proportion of nonfractured limbs with POD (5/26 [19%]), the difference was not significant. We observed POD in the lateral condyle for 24 of 88 (27%) control limbs and in the medial condyle for 23 of 88 (26%) control limbs. Most of the POD lesions were not graded as severe (ie, ≤ 2), so POD was categorized as present or absent for analysis. The proportion of control limbs with POD in either condyle (29/88 [33%]) was significantly greater than the proportion of fractured limbs with POD (1/26 [4%]).
Condylar fissures
Condylar fissures were significantly more common in the nonfractured limb of case horses (7/26 [27%]) than in the fractured limb of case horses (0/26 [0%]), with 6 fissures occurring in the lateral condyle and 3 in the medial condyle. Two nonfractured limbs of case horses had a fissure in both the lateral and medial condyles, 1 nonfractured limb of a case horse had a fissure in only the medial condyle, and 4 nonfractured limbs of case horses had a fissure only in the lateral condyle. Of the 9 condylar fissures, 7 were BML-negative and 2 were BML-positive structures.
Condylar fissures were rare in the limbs of control horses and the fractured limb of case horses. The proportion of condylar fissures in either condyle of control limbs (4/88 [5%]) was not significantly different from that of fractured limbs in case horses (0/26 [0%]). For the 4 control limbs with condylar fissures, 2 were located medially and 2 were located laterally.
Discussion
Although BSI is diagnosed in human runners, up to 30% of marching soldiers, and 20% of track and field athletes,7–14,20 Thoroughbred racehorses develop BSI at a higher frequency and at more locations than human athletes.19 At speeds of up to 60 km/h, racing Thoroughbreds with a typical weight of 500 kg have each limb supporting the entire load unassisted at some point during the 4 beats of the gallop gait. A catastrophic condylar fracture is the culminating event of a BSI that begins in the distopalmar aspect of MC3 (or distoplantar aspect of MT3) where the proximal sesamoid bones of the suspensory apparatus oppose stress created by the downward movement of MC3 (or MT3) during the stance phase of the gallop.15,16
The strain required to break cortical bone in a single event (maximal strain) in tension is high (7,300 μϵ in human cortical bone) when bone fails because of a single episode of loading. Known as a traumatic fracture, the MRI appearance of these peracute lesions are distinct and different from the MRI appearance of chronic repetitive bone injuries.7,8 Bone injury can occur in the absence of a radiographically apparent fracture because lower strain (submaximal strain) creates microscopic damage (ie, microdamage).7–14 Microdamage of BSI is a physiologic event that initiates bone remodeling. Bone adapts to increases in load by increasing its density to prevent microdamage.7–14 Paradoxically, both the 3-stage timeline of remodeling (activation of osteoclasts and osteoblasts, then removal of damaged bone by osteoclasts, and finally formation of new bone by osteoblasts) and the faster speed of osteoclasts (weeks) versus osteoblasts (months) weaken bone and increase the risk of fracture in the short term.7–14 Thus, BSI becomes a pathological event when repetitive stresses continue on microdamaged bone, which causes repair mechanisms to be overwhelmed. There are 3 distinct stages of BSI: stress reaction, stress fracture, and complete (catastrophic) fracture.7–14 A stress reaction is a prefailure event distinguished by the presence of a BML and provides evidence of BSI at a cellular level.8,11 A stress fracture (macrocrack) is a prefailure event distinguished by the presence of an incomplete fracture line and provides evidence of BSI at a macroscopic level.8,11 A complete fracture is a failure event that occurs when a macrocrack propagates during training or racing and results in a complete and stable fracture, whereas a catastrophic fracture occurs when an unidentified BSI propagates and results in a complete and unstable fracture.7–14
The BSIs are subclassified as low- or high-risk injuries.14 A low-risk BSI11 occurs on the compression side of the bending axis of a bone, and recovery is typically uncomplicated. A high-risk BSI14 (eg, femoral neck in humans) occurs on the tension side of the bending axis of a bone and is at high risk for progression to a complete fracture. On the basis of this definition, an MC3 condylar fracture in horses would be considered a high-risk BSI. There is evidence that a transverse tensile force acts at the distopalmar aspect of MC3 between the condyles and the sagittal ridge owing to a lateromedial deformation force in galloping Thoroughbreds.17 The earliest diagnosis possible in the BSI continuum would help to avoid progression to higher grades and longer healing times and yield a reduced risk to horses (and jockeys) attributable to fracture propagation.
The term bone marrow edema has been replaced with BML. The BMLs are characterized by an excessive fluid signal in the marrow space on MRI images. Lesions are not typical of edema by histologic criteria30; rather, they are characterized by fibrosis, lymphocytic infiltrates, and increased vascularization, with the latter likely being responsible for the fluid signal seen on MRI images.30–32 In addition, BMLs are not evident on radiographic or CT images. The MRI characteristics of BMLs include a decreased signal intensity on T1W images and an increase in signal intensity on T2-weighted sequences. The increase in signal intensity of the most fluid-sensitive sequences (STIR) and fat-suppressed T2-weighted sequences created by BMLs is much more pronounced than the decrease in signal intensity created by BMLs on the T1W sequence. The T2*W chemical shift images are also valuable for BML identification.30–32 Identification of a BML with MRI images is not specific for a disease process, but the location and distribution of the marrow abnormality can provide insight to the etiology and, if related to trauma, the mechanism of injury. When generated by chronic and repetitive trauma, histologic lesions are the result of hemorrhage and intertrabecular microfractures.30–32
In the study reported here, limbs were not frozen before MRI, so the integrity of the soft tissues and medullary fluid was not affected by freeze-thaw cycles, which were used in another study.31 The presence of BMLs in fracture formation is supported in the literature on humans8–14,22 and equids.23–25 In the present study, presence of a BML was significantly more common when comparing fractured (26/26 [100%]) versus nonfractured (7/26 [27%]) limbs of case horses and fractured limbs of case horses (26/26 [100%]) versus limbs of control horses (6/88 [7%]), with sensitivity of 100% and specificity of 93%. Magnetic resonance imaging is the advanced imaging modality recommended for evaluation of BSI in human medicine because it has the best combined specificity and sensitivity of the available imaging modalities33 and is the only modality that can be used to identify BMLs. Recognizing that BSI develops as a pathological continuum, with BMLs becoming apparent before a stress fracture, the identification of BMLs provides the earliest warning signs of a forming BSI.
In the present study, an increase in the STIR signal was evident in the palmar condyle for 3 situations: BMLs, POD, and blood vessels. An increased STIR signal was associated with the BML of a condylar fracture localized axially proximal to the SCB plate toward the physeal remnant in the plane of the parasagittal ridge and that radiated in the sagittal plane. This made the 90° orientation of frontal plane images superior for lesion identification. This information is supported by studies that involved the use of MRI22–24 and CT26 to identify that most condylar fractures occurred at sites deep within the trabecular bone, proximal to the articular surface in areas of high bone volume fraction rather than within density gradients at the articular surface of these areas. The increase in STIR signal associated with a POD lesion was localized subjacent to the SCB plate palmar to the transverse ridge and radiated in the frontal plane. This made the 90° orientation of sagittal plane images superior for identification of POD lesions. The location of the signal changes associated with POD is supported by results of gross15,27,28 and MRI4,24 examinations. In the study reported here, the increase in STIR signal from blood vessels was identified as distinct linear structures that originated proximal and palmar in close proximity to the physeal remnant and radiated circuitously. This can be incorrectly identified as a unicortical condylar or fissure fracture. These blood vessels were possibly dormant vasculature that remained from closure of the physis and were recruited and proliferated in horses with advanced POD. The distinct location for each of these 3 situations that caused an increase in the STIR signal may provide information for accurate identification of the inciting cause of injury in racing Thoroughbreds. The pattern of BML observed in traumatic knee injuries of humans have been correlated with the type and inciting cause of injury.22
An increase in bone mineral density of 5% to 8% caused by mechanical loading can improve bone strength by > 64%, recognizing that it is the mineral content of bone that provides stiffness.7–9 Bone with a marked increase in bone mineral density typically is very stiff but also brittle, which results in a reduction in work to failure and increases the risk of fracture. In a 2003 report,34 equine compact bone was found to be stiff and strong in compression but also associated with brittle postyield behavior.
Racing creates changes in density across the palmar aspect of MC3 of Thoroughbreds, with the highest density abaxially in the condyles and lowest density axially at the sagittal ridge.5 Authors of that study5 suggested that this density gradient may be responsible for lateral condylar fractures originating in the parasagittal groove. There is conflicting information on the use of changes in density for predicting fracture formation.26,35–39 In 42 horses with lateral condylar fracture, CT slices were imaged to a depth of 2 mm, and examination of those slices provided an impression of changes in the SCB plate but not the adjacent trabecular bone.35 A focal increase in bone density and increased heterogeneity of density were characteristic of limbs with lateral condylar fractures, compared with characteristics for control and nonfractured limbs.35 In another study,36 a 1.5-T MRI unit was used to measure the depth of dense SCB at an angle 35° palmar to the transverse ridge in the lateral parasagittal groove, which revealed that a dense-bone depth of 16 mm was associated with an increased likelihood of lateral condylar fracture. In contrast, high-resolution CT was used to identify that 7 of 13 (54%) horses had no evidence of focal porosity in the parasagittal groove and that most of the condylar fractures occurred at sites deep within the trabecular bone, which suggested that margin density gradients were not helpful for identifying horses with condylar fractures.26 Investigators of 2 other studies37,38 that involved the use of CT to compare horses with and without condylar fracture were not able to identify significant differences between groups.
Although changes in density in the fractured condyle were significantly associated with injury in the present study, DBVP had only moderate sensitivity and specificity for predicting condylar fracture and was less sensitive and less specific than the identification of BML. There was a significant difference in the mean DBVP in the lateral (26% vs 16%) or medial (25% vs 18%) condyle when fractured limbs in case horses were compared with limbs of control horses. When the fractured limb was stratified on the basis of the location of the condylar fracture, mean DBVP in the lateral (32% vs 21%) or medial (42% vs 20%) condyle was significantly different when fractured and nonfractured limbs of case horses were compared and when the lateral (32% vs 16%) or medial (42% vs 18%) condyle of fractured limbs of case horses was compared with the limbs of control horses. A cutpoint of 25% for DBVP was chosen for comparison because it was higher than the mean value for control horse condyles and was equal to the unstratified value for horses with condylar fracture. In case horses with a lateral condylar fracture (n = 16), a cutpoint of 25% for DBVP had sensitivity of 62% and specificity of 71% for identifying a lateral condylar fracture, when compared with the control limbs, and for case horses with a medial condylar fracture (10), a DBVP cutpoint of > 25% had sensitivity of 70% and specificity of 82% for identifying a medial condylar fracture, when compared with the control limbs. A similar finding was identified in 37 necropsied Thoroughbred racehorses evaluated by use of high-resolution peripheral quantitative CT.38 Authors of that study38 concluded that although the bone volume or total volume of the epiphysis in horses with condylar fracture was significantly associated with injury, it was challenging to make a diagnosis, even with high-resolution 3-D imaging of necropsied horses, and CT had only moderate sensitivity and specificity for the prediction of injury.
In the study reported here, fractured limbs had a significant within-limb difference in mean DBVP. When the fracture was in the lateral condyle, mean DBVP was significantly higher in the lateral versus medial condyle in the fractured limb and the lateral or medial condyle in the nonfractured limb. A similar finding was identified when the fracture was in the medial condyle. The association between the within-limb difference of discordant densities of condyle pairs when evaluating mean DBVP may be an early warning sign of impending condylar fracture that requires further investigation. A prospective cohort study with a large sample size could help to better define the basic pathogenesis of MC3 fractures and the association of discordant densities of condyles as an early warning sign of an MC3 fracture.
Palmar osteochondral disease is another fatigue injury of equine athletes that affects the SCB immediately adjacent to the metacarpal condyle.4,15,24,27,28 Similar signal changes and the location of these changes were identified in the horses reported here and in other studies4,24 that involved MRI evaluation of POD. In the study reported here, the proportion of control limbs (29/88 [33%]) with POD was significantly greater than that for condylar-fractured limbs of case horses (1/26 [4%]). This finding is supported by several studies27,28,38 of equine cadavers, which suggests that the presence of POD may have a protective effect on the formation of linear fissures and condylar fractures. It has been reported27,28 that as the grade of the POD lesion increases, the grade of the linear fissure decreases, which suggests a negative association between the grade of POD and grade of the linear fissure in the parasagittal grove of the condyle. In a study27 of Thoroughbred racehorses euthanized because of serious injury during racing, excluding condylar fractures, POD was identified in 67 of 114 (59%) condyles. In another study,38 POD was observed in 2 of 13 horses with lateral condyle fractures versus 8 of 16 horses with other fatal limb injuries. It has been suggested that different patterns of loading in individual animals may predispose to a linear and condylar fracture formation or to POD.27,28 This association requires further evaluation.
Fissures in the parasagittal groove are commonly identified in the contralateral limb of horses with condylar fractures during postmortem examination, and it has been suggested these fractures are evidence of preexisting disease.27 In a study38 of horses with condylar fractures, 5 of 13 limbs had fissures in a parasagittal groove of the contralateral limb. Because many exercise-induced SCB lesions occur bilaterally, it is reasonable to assume that the presence of a fissure increases the risk of a complete fracture.38 A study38 involving CT of equine cadavers identified parasagittal fissures or short incomplete fractures in the parasagittal groove of the contralateral limb of 5 of 13 horses with condylar fractures, whereas no fissures were observed in any of the 8 control horses. The bilateral occurrence of BSI has been reported previously in racing Thoroughbreds in studies that involved the use of scintigraphy19 and sMRI.24,25 In the present study, bilateral BMLs and condylar fissures were detected in 7 of 26 (27%) nonfractured limbs of horses with a catastrophic condylar fracture in the affected limb.
Identifying BSI by use of radiography is limited, with up to 85% of stress fractures unidentified in the early stages of fracture development and up to 50% undetected during follow-up radiography.10 Scintigraphy has been used for the diagnosis of BSI and has a similar sensitivity but inferior specificity to those of MRI, and it can provide false-negative results when conducted too early in the course of disease.33 Identification of cancellous BSI by use of CT relies on changes in bone density and a formed fissure or fracture33; therefore, CT is helpful later than MRI during the course of disease for the detection of a BSI. In the study reported here, density was significantly associated with injury but was not as discriminating as the identification of BML by use of STIR MRI images.
The preferred imaging modality for the identification of a BSI in humans is MRI.10,33 A BML confirms the presence of a BSI, the stage of injury, a prediction of the course of healing, and an appropriate schedule for return to competition.20,21 A BSI has 4 categories that map the location of a BML alone until the final stage of injury: grade 1, increase in periosteal STIR signal; grade 2, increase in periosteal and medullary STIR signal; grade 3, increase in medullary STIR signal plus decrease in T1W signal; and grade 4, all of the components of grade 3 plus a visibly discrete fracture line.20,21 Meta-analysis of 3,724 imaging evaluations was used to examine the diagnostic accuracy of various imaging modalities for the definitive identification of stress fractures in the lower extremities of humans.33 Authors of that study33 reported 95% CIs for sensitivity and specificity of conventional radiography, scintigraphy, MRI, and CT, and they concluded that MRI was the most sensitive and specific imaging test for diagnosing BSI of the lower extremities in humans because it can be used to identify a BML, the earliest indication of a BSI.8–14,20,33
When euthanasia is an inclusion criterion, there will be a bias toward fractures of a higher severity. Histopathologic data were unavailable for the study reported here; however, this information has been reported elsewhere.15,35 The MRI images were obtained from a motionless limb, so the quality of the images was likely higher than that obtained by use of standing subjects. The presence of a catastrophic condylar fracture meant that the investigators were aware of group assignment, which might have introduced bias in the assessment of BMLs and DBVP grades. A cohort study would be needed to better determine the pathogenesis and progression of MC3 (or MT3) condylar fracture and POD as distinct entities and in combination, the prevalence and distribution of these 2 conditions within racing populations, their potential role in the development of lameness, and progression or improvement of these conditions with altered training and racing regimens as monitored with MRI. It is possible that MRI findings associated with catastrophic MC3 fracture in the present study were confounded by other variables.
Early diagnosis of fatigue injury to bone prior to development of a fracture in the distal aspect of a limb should be a focus for future research.39 A BSI involves an imbalance between load-induced microdamage and bone removal and repair. This physiologic process becomes pathological when bone has undergone excessive stress and repair mechanisms become overwhelmed by the accumulation of microdamage that leads to the development and continuum of BSI. A BML provides evidence of fracture formation at the cellular level and the earliest opportunity for identification of a BSI with advanced imaging. The sMRI is available for evaluation of BSIs in the metacarpophalangeal or metatarsophalangeal joints of racehorses,4,24,25 which allows BMLs to be identified without the need to anesthetize horses. Longer acquisition times, motion artifact, and low signal-to-noise ratios associated with STIR sMRI images have caused equine researchers to consider the use of CT as an imaging modality for providing early warning signs of a BSI in the metacarpophalangeal or metatarsophalangeal joints.39,40 Standing CT offers shorter acquisition times, but it relies on changes in bone density and thus delays the ability to make a diagnosis until a discrete fracture line is evident.20,21,33 Standing MRI has longer acquisition times, but it provides the ability to make an earlier diagnosis of the high-risk nature of an MC3 BSI14,17 through identification of a BML. The merit for each of these 2 imaging modalities to identify a BSI before it becomes a catastrophic condylar fracture in racing Thoroughbreds requires further investigation.
Acknowledgments
Presented in part as an abstract at the British Equine Veterinary Association Congress, Liverpool, England, September 2017.
The authors thank Nina Ubide and P. J. Campo for technical assistance.
ABBREVIATIONS
| BML | Bone marrow lesion |
| BSI | Bone stress injury |
| CI | Confidence interval |
| DBVP | Dense bone volume percentage |
| MC3 | Third metacarpal bone |
| MT3 | Third metatarsal bone |
| POD | Palmar osteochondral disease |
| SCB | Subchondral bone |
| sMRI | Standing MRI |
| STIR | Short tau inversion recovery |
| T1W | T1-weighted |
| T2*W | T2*-weighted |
Footnotes
Equine Injury Database [database online]. Lexington, Ky: The Jockey Club, 2008. Available at: www.jockeyclub.com. Accessed Feb 1, 2018.
Hallmarq Veterinary Imaging, Guilford, England.
Equibase [database online]. Lexington, Ky: Equibase Company LLC, 2018. Available at: www.equibase.com. Accessed Jan 10, 2018.
Osirix Imaging Software, Geneva, Switzerland.
S-PLUS, version 8.2, TIBCO Inc, Seattle, Wash.
References
1. Estberg L, Stover SM, Gardner IA. Fatal musculoskeletal injuries incurred during racing and training in Thoroughbreds. J Am Vet Med Assoc 1996;208:92–96.
2. Peloso JG, Mundy GD, Cohen ND. Prevalence of, and factors associated with musculoskeletal racing injuries of Thoroughbreds. J Am Vet Med Assoc 1994;204:620–626.
3. Parkin TDH, Clegg PD, French WP. Risk of fatal distal limb fracture among Thoroughbreds involved in the five types of racing in the United Kingdom. Vet Rec 2004;154:493–497.
4. Peloso JG, Vogler JB III, Cohen ND, et al. Association of catastrophic biaxial fracture of the proximal sesamoid bones with bony changes of the metacarpophalangeal joint identified by standing magnetic resonance imaging in cadaveric forelimbs of Thoroughbred racehorses. J Am Vet Med Assoc 2015;246:661–673.
5. Riggs CM, Whitehouse GH, Boyde A. Structural variation of the distal condyles of the third metacarpal bone and third metatarsal bone in the horse. Equine Vet J 1999;31:130–139.
6. Hitchens PL, Hill AE, Stover SM. The role of catastrophic injury or sudden death of the horse in race-day jockey falls and injuries in California, 2007–2012. Equine Vet J 2016;48:50–56.
7. Warden SJ, Hurst JA, Sanders MS, et al. Bone adaptation to a mechanical loading program significantly increases skeletal fatigue resistance. J Bone Miner Res 2005;20:809–816.
8. Warden SJ, Davis IS, Fredericson M. Management and prevention of bone stress injuries in long-distance runners. J Orthop Sports Phys Ther 2014;44:749–765.
9. Oftadeh R, Perez-Viloria M, Villa-Camacho JC, et al. Biomechanics and mechanobiology of trabecular bone: a review. J Biomech Eng 2015;137:1–15.
10. Datir AP, Saini A, Connell A. Stress-related bone injuries with emphasis on MRI. Clin Radiol 2007;62:828–836.
11. Pathria MN, Chung CB, Resnick DL. Acute and stress-related injuries of bone and cartilage: pertinent anatomy, basic biomechanics, and imaging perspective. Radiology 2016;280:21–38.
12. Tenforde AS, Kraus E, Fredericson M. Bone stress injuries in runners. Phys Med Rehabil Clin N Am 2016;27:139–149.
13. Kahanov L, Eberman LE, Games KE, et al. Diagnosis, treatment, and rehabilitation of stress fractures in the lower extremity in runners. Open Access J Sports Med 2015;6:87–95.
14. McInnis KC, Ramey N. High-risk stress fractures: diagnosis and management. PM R 2016;8:S113–S124.
15. Pool RR, Meagher DM. Pathologic findings and pathogenesis of racetrack injuries. Vet Clin North Am Equine Pract 1990;6:1–30.
16. Norrdin RW, Kawcak CE, Capwell BA, et al. Subchondral bone failure in an equine model of overload arthrosis. Bone 1998;22:133–139.
17. Rubio-Martínez LM, Cruz AM, Inglis D, et al. Analysis of the subchondral microarchitecture of the distopalmar aspect of the third metacarpal bone in racing Thoroughbreds. Am J Vet Res 2010;71:1148–1153.
18. Stover SM, Johnson BJ, Daft BM, et al. An association between complete and incomplete stress fractures of the humerus in racehorses. Equine Vet J 1992;24:260–263.
19. Arthur RM, Constantinide D. Results of 428 nuclear scintigraphic examinations of the musculoskeletal system at a Thoroughbred racetrack, in Proceedings. 41st Annu Conv Am Assoc Equine Pract 1995;41:84–87.
20. Nattiv A, Kennedy G, Barrack MT, et al. Correlation of MRI grading of bone stress injuries with clinical risk factors and return to play: a 5-year prospective study in collegiate track and field athletes. Am J Sports Med 2013;41:1930–1941.
21. Fredericson M, Bergman AG, Hoffman KL, et al. Tibial stress reaction in runners. Correlation of clinical symptoms and scintigraphy with a new magnetic resonance imaging grading system. Am J Sports Med 1995;23:472–481.
22. Sahoo K, Garg A, Saha P, et al. Study of imaging pattern in bone marrow oedema in MRI in recent knee injuries and its correlation with type of knee injury. J Clin Diagn Res 2016;10:TC06–TC11.
23. Zani DD, De Zani D, Biggi M, et al. Use of magnetic resonance imaging in the diagnosis of bone marrow edema in the equine distal limb: six cases. Vet Res Commun 2009;33:225–228.
24. Powell SE. Low-field standing magnetic resonance imaging findings of the metacarpo/metatarsophalangeal joint of racing Thoroughbreds with lameness localized to the region: a retrospective study of 131 horses. Equine Vet J 2012;44:169–177.
25. Ramzan PH, Palmer L, Powell SE. Unicortical condylar fracture of the Thoroughbred fetlock: 45 cases (2006–2013). Equine Vet J 2015;47:680–683.
26. Whitton RC, Trope GD, Ghasem-Zadeh A, et al. Third metacarpal condylar fatigue fractures in equine athletes occurs within previously modeled subchondral bone. Bone 2010;47:826–831.
27. Barr ED, Pinchbeck GL, Clegg PD, et al. Post mortem evaluation of palmar osteochondral disease (traumatic osteochondrosis) of the metacarpo/metatarsophalangeal joint in Thoroughbred racehorses. Equine Vet J 2009;41:366–371.
28. Pinchbeck GL, Clegg PD, Boyde A, et al. Pathological and clinical features associated with palmar/plantar osteochondral disease of the metacarpo/metatarsophalangeal joint in Thoroughbred racehorses. Equine Vet J 2013;45:587–592.
29. Firth EC. Current concepts of infectious polyarthritis in foals. Equine Vet J 1983;15:5–9.
30. Eriksen EF. Treatment of bone marrow lesions (bone marrow edema). Bonekey Rep 2015;4:755.
31. Del Grande F, Farahani SJ, Carrino JA, et al. Bone marrow lesions: a systematic diagnostic approach. Indian J Radiol Imaging 2014;24:279–287.
32. Starr AM, Wessely MA, Albastaki U, et al. Bone marrow edema: pathophysiology, differential diagnosis, and imaging. Acta Radiol 2008;49:771–786.
33. Wright AA, Hegedus EJ, Lenchik L, et al. Diagnostic accuracy of various imaging modalities for suspected lower extremity stress fractures: a systematic review with evidence-based recommendations for clinical practice. Am J Sports Med 2016;44:255–263.
34. Les CM, Stover SM, Keyak JH, et al. Stiff and strong compressive properties are associated with brittle postyield behavior in equine compact bone material. J Orthop Res 2002;20:607–614.
35. Loughridge AB, Hess AM, Parkin TD, et al. Qualitative assessment of bone density at the distal articulating surface of the third metacarpal in Thoroughbred racehorses with and without condylar fracture. Equine Vet J 2017;49:172–177.
36. Tranquille CA, Murray RC, Parkin TD. Can we use subchondral bone thickness on high-field magnetic resonance images to identify Thoroughbred racehorses at risk of catastrophic lateral condylar fracture? Equine Vet J 2017;49:167–171.
37. Bogers SH, Rogers CW, Bolwell C, et al. Quantitative comparison of bone mineral density characteristics of the distal epiphysis of third metacarpal bones from Thoroughbred racehorses with or without condylar fracture. Am J Vet Res 2016;77:32–38.
38. Trope GD, Ghasem-Zadeh A, Anderson GA, et al. Can high-resolution peripheral quantitative computed tomography imaging of subchondral and cortical bone predict condylar fracture in Thoroughbred racehorses? Equine Vet J 2015;47:428–432.
39. Dubois MS, Morello S, Rayment K, et al. Computed tomographic imaging of subchondral fatigue cracks in the distal end of the third metacarpal bone in the thoroughbred racehorse can predict crack micromotion in an ex-vivo model. PLoS One 2014;9:e101230.
40. Smith MR, Kawcak CE, McIlwraith CW. Science in brief: report on the Havemeyer Foundation workshop on subchondral bone problems in the equine athlete. Equine Vet J 2016;48:6–8.
