Materials and Methods

Sample population—Forelimbs from the cadavers of 26 mature dogs were used in the study. Dogs had been euthanized for medical reasons unrelated to this study. The dogs were obtained from the Department of Pathobiology, Institute of Pathology and Forensic Veterinary Medicine of the University of Veterinary Medicine, Vienna, Austria. The study was conducted at the Department of Pathobiology, Institute of Anatomy of the University of Veterinary Medicine, Vienna, Austria.

Dogs weighed 3.7 to 74.5 kg and were assigned to 3 groups (large-breed dogs [n = 10], small-breed dogs [8], and chondrodystrophic-breed dogs [5]). Mediolateral and craniocaudal radiographic views were obtained for each limb, and limbs that had joints with evidence of osteoarthritis or other abnormalities were excluded. The final sample population consisted of 19 limbs of 10 large-breed dogs (both forelimbs of 3 Great Danes, 1 Bernese Mountain Dog, 1 German Shepherd Dog, 1 German Wirehaired Pointer, 1 Doberman Pinscher, 1 Weimaraner, and 1 Dalmatian and the left forelimb of 1 Bordeaux Dog), with body weight ranging from 28.5 to 74.5 kg (mean body weight, 48.5 kg); 14 limbs of 8small-breed dogs (both forelimbs of 3 Poodles, 1 Yorkshire Terrier, 1 Miniature Spitz, and 1 Fox Terrier and the right forelimb of 1 Poodle and 1 Beagle), with body weight ranging from 3.7 to 15.0 kg (mean, 9.0 kg); and 8 limbs of 5 chondrodystrophic-breed dogs (both forelimbs of 3 Dachshunds and the left forelimb of 1 Pekingese and 1 Dachshund), with body weight ranging from 5.3 to 8.8 kg (mean, 7.5 kg). The large-breed dogs consisted of 4 males and 6 females (age range, 1.5 to 13.0 years; mean ± SD, 6.7 ± 3.3 years), the small-breed dogs consisted of 5 males and 3 females (age range, 8.0 to 17.0 years; mean, 12.4 ± 3.8 years), and the chondrodystrophic-breed dogs consisted of 2 males and 3 females (age range, 8.0 to 13.0 years; mean, 11.0 ± 1.8 years). The intact limbs from each dog were sealed in a plastic bag and stored in a freezer at −18°C until analyzed.

MRI procedures—Limbs were thawed at room temperature (approx 23°C) on the day before MRI was conducted. Images of joints were obtained at the University Hospital of Diagnostic Radiology, University of Vienna, Vienna, Austria, by use of a 3.0-Tesla unit.^a Limbs were positioned in a volume coil (circular transmit-receive coil) with an inner diameter of 20 cm in a semiextended position with the olecranon positioned on top. For extremely small dogs, a coil with an inner diameter of 7 cm was used to optimize the signal-to-noise ratio. Sagittal images were obtained by use of a 3-D, spoiled gradient-echo, fast-imaging sequence (time repetition–time echo–flip angle, 50 milliseconds–8 milliseconds–45°; 2 acquisitions; 20 minutes of imaging time). Images were aligned parallel to the sagittal ridge of the ulnar trochlear notch,²⁵ which followed the long axis of the ulnar shaft distally, and were obtained at a partition thickness of 0.93 mm and an in-plane resolution of 0.23 × 0.47 mm (field of view, 120 × 120 mm; matrix, 512 × 256). After imaging, the specimens were dissected to confirm radiographic findings. Detection of cartilage-free areas was recorded. Then, the specimens were macerated to confirm integrity of the subchondral bone. Step incongruity was not detected in any specimens, but cartilage-free areas in the center of the trochlear notch (sometimes also involving the lateral coronoid process) were evident in all specimens of the large-breed dogs, whereas these were confined to the lateral coronoid process and seen only in 2 limbs of the small-breed dogs and 6 limbs of the chondrodystrophic-breed dogs.

MRI measurements—For each joint, 2 images depicting the elbow joint in sagittal planes were selected. The first image (ie, S1) was defined as the image intersecting the elbow joint adjacent to the sagittal ridge at the center of the laterocranial tubercle of the olecranon, whereas the second image (S2) was located more laterally and passed the joint at its most lateral aspect, immediately before the articular surface of the trochlear notch sloped caudally to result in double contours of the trochlear notch (Figure 1). The MRI scans were digitized by use of a transmitted light scanner.^b Mirror-inverted scans of the right forelimbs were used for further analyses.

View Full Size

Illustrations depicting the cranial view of the proximal aspect of the left radius and ulna and position of S1 and S2 through the cubital (elbow) joint of dogs obtained by use of MRI (A) and a lateral view of the osseus contours obtained for S1 and the position of the reference points (B). Reference points A to C were located on the ulnar trochlear notch, D to F were located on the radial articular fovea, and point M was located at the center of the humeral condyle. The coordinates for the x- and y-axes are indicated.

Citation: American Journal of Veterinary Research 67, 2; 10.2460/ajvr.67.2.211

• Download Figure
• Download figure as PowerPoint slide

Six reference points (A to F) were defined at the bone-cartilage boundary of the elbow joint (Figure 2). By use of an imaging processing software program,^c 4 reference points (A, C, D, and F) were initially defined at the bone-cartilage boundary of the ulnar trochlear notch and the radial articular fovea, respectively. Reference point A represented the tip of the anconeal process and hence the proximal end of the ulnar trochlear notch, reference point C represented the distal end of the ulnar trochlear notch, reference point D was defined as the caudal end of the radial articular fovea, and reference point F represented the cranial end of the radial articular fovea. Then, a circle that optimally intersected these 4 reference points was created and the center (ie, point M) was determined. When we failed to create a circle that passed through all 4 reference points, then, alternatively, a circle was designed that passed through reference point A and 1 of the other 3 reference points (C, D, or F), but the distance of the deviation of any remaining points from the circle had to be minimal (eg, typically 1 point was located slightly outside and 1 point was located slightly inside of the circle). Finally, 2 additional reference points (B and E) were established. Reference point B was defined as the point located at the bone-cartilage boundary within the concavity of the trochlear notch whose distance was maximal to point M, and reference point E was defined as the point located at the bone-cartilage boundary within the concavity of the radial articular fovea whose distance was maximal to point M.

View Full Size

Sagittal, fat-suppressed, gradient-echo MRI scan of the left elbow joint of a female Great Dane for S1. The bone-cartilage boundaries are indicated (arrows). Notice that the subchondral bone has a low-intensity signal and the articular cartilage has an isointense signal, compared with signals for the muscle and surrounding soft tissues. U = Ulna. H = Humerus. R = Radius. Bar = 10 mm.

Citation: American Journal of Veterinary Research 67, 2; 10.2460/ajvr.67.2.211

• Download Figure
• Download figure as PowerPoint slide

The 6 reference points located at the antebrachial (ie, outer) bone-cartilage boundary of the joint were used to create 6 reference lines connecting each of the reference points with point M. The corresponding 6 humeral (ie, inner) reference points were defined as the intersection points between the reference lines and bone-cartilage boundary at the humerus. The Cartesian coordinates of the antebrachial and humeral reference points A to F and point M were recorded.

The interosseous distances between antebrachial and corresponding humeral reference points were determined by use of 2 methods. Measurements of the antebrachial and humeral distances between reference points A and M through reference points F and M and calculations of the interosseous distances (between the antebrachial and humeral surfaces) were performed. Interosseus distances were also calculated by means of trigonometry by use of the coordinates for the antebrachial reference points A through F, humeral reference points A through F, and point M. All measurements and calculations were performed by a single investigator (KJJ) and determined to an accuracy of 0.1 mm.

Processing of data—To allow for comparisons among specimens that had differing positions, the original Cartesian (x, y) reference frame was translated until point M was located at (x₀, y₀) equal (–5.0 cm, 5.0 cm) and rotated about the fixed center (point M) until antebrachial reference point A was positioned directly above point M (ie, x₀ of point A was −5.0 cm). Reference points B through F were translated and rotated accordingly. Results referring to these data were termed original data. To allow for comparison among specimens of various sizes, an adjustment factor for body size was introduced. The distance between antebrachial reference point A and point M was set to 3.0 cm for S1 in all specimens (ie, the new coordinate of the antebrachial reference point A was set to [x₀, y₀] equal [–5.0 cm, 8.0 cm]) with the adjustment factor equivalent to 3 divided by the original distance between antebrachial point A and point M; all other coordinates in S1 and S2 were adjusted accordingly. Results referring to these data were termed adjusted data. Mean ± SD value for the adjustment factor was 1.38 ± 0.25 for the large-breed dogs, 2.63 ± 0.55 for the small-breed dogs, and 2.33 ± 0.19 for the chondrodystrophic-breed dogs.

Position of the reference points within the Cartesian (x, y) reference frame was characterized by the inclination angles of the y-axis relative to the reference lines from the antebrachial reference point A to point M through the antebrachial reference point F to point M and was calculated by use of trigonometry. Therefore, point M was translated to (x₀, y₀) equal (0.0 cm, 0.0 cm), and the angle increased as we proceeded counterclockwise from reference point A to reference point F.

Reproducibility of measurements—Measurements of the antebrachial and humeral distances of the line from reference points A to M through F to M were performed twice. Intraobserver reproducibility of data was tested by comparing results of these 2 measurements and by comparing the first measurement with the trigonometric calculation of the distances by the use of the coordinates of the reference points; reproducibility was expressed as the CV. For the comparison of the 2 measurements, CV ranged between 0.35% (antebrachial reference point D to point M for S2) and 1.07% (antebrachial reference point C to point M for S1). For comparison of the measurement with the trigonometric calculation, the CV was slightly higher and ranged between 0.57% (antebrachial reference point C to point M for S2) and 2.56% (antebrachial reference point E to point M for S1). The trigonometrically calculated data were used for further analysis.

Statistical analysis—A statistical software program^d was used to perform statistical analysis. Results of the left and right limbs were pooled, and normal distribution of data was confirmed (Kolmogorov-Smirnov test). Student t tests were performed for each of the 3 groups of dogs to compare the widths of the gaps at various reference sites but within 1 image (ie, S1 or S2) and also to compare results of corresponding reference sites between S1 and S2. A 1-way ANOVA (Scheffé test) was used to compare position of the reference points and widths of the gaps among the 3 groups of dogs at corresponding reference sites and images. Tests were applied on both the original and adjusted data set. Correlations between body weight and width of the interosseous gaps (original data) were tested by calculation of the Pearson correlation coefficient. Significance was set at a value of P ≤ 0.05 for all tests.

Results

Position of reference points—For S1, differences among groups in the position of the reference points were limited to point B, which was located more proximally (ie, nearer to point A) in the small-breed dogs, compared with its location in the large- and chondrodystrophic-breed dogs (Figure 3; Table 1). For S2, this same relationship was still valid when comparing the small- and large-breed dogs. In the large-breed dogs, the trochlear notch encircled the humeral condyle to a significantly higher degree, and reference point C was located significantly more proximally in the chondrodystrophic-breed dogs, compared with the location in the small-breed dogs, indicating a shorter trochlear notch in the joints of the chondrodystrophic-breed dogs (ie, smaller proximodistal diameter). Significant within-group differences between S1 and S2 were limited to reference point C in the large-breed dogs, which denoted a higher trochlear notch for S1 (ie, larger proximodistal diameter).

View Full Size

Mean position of the antebrachial and humeral reference points (A to F) and the center of the circle (point M) indicating mean width of the interosseous distances for S1 (panels A, B, and C) and S2 (panels D, E, and F) in the large-breed dogs (panels A and D), small-breed dogs (panels B and E), and chondrodystrophic-breed dogs (panels C and F). Results were determined by use of adjusted data. Solid squares refer to antebrachial reference points (A to F), and open squares represent the humeral reference points (A to F).

Citation: American Journal of Veterinary Research 67, 2; 10.2460/ajvr.67.2.211

• Download Figure
• Download figure as PowerPoint slide

Width of interosseous gaps—Comparing original data among groups revealed significant differences in the width of the interosseous gaps at reference points B to E as a result of differences in size of the dogs. Thus, it is noteworthy that there were no significant differences among the 3 groups at reference points A and F (ie, the cranioproximal and craniodistal limits of the joint; Table 2). Gap widths at the reference points were not significantly different when comparing S1 and S2, except for the gap width at reference point A, which was significantly wider in the large- and chondrodys-trophic-breed dogs for S2, compared with the width for those breeds in S1.

In the large-breed dogs, interosseous gaps in S1 were significantly narrower at reference point A than at any other reference point and the gaps in S1 and S2 were wider at reference point B than at any other reference point. In contrast, width of the gap at reference point A did not differ significantly from the widths of other reference points in the small-breed dogs, and differences mainly could be confirmed between the narrowest gaps for reference points C and D and the widest gaps at reference point F. Similarly, the gap at reference point B was not particularly wide in the chondrodystrophic-breed dogs, but the joint was especially narrow at reference point D.

Interestingly, body weight and width of the interosseous gaps were correlated only in the large-breed dogs and only at specific reference points. For S1, body weight and gap width were correlated for reference points B (r, 0.834; P < 0.001) and D (r, 0.458; P < 0.050), whereas for S2, body weight and gap width were correlated for reference points B (r, 0.680; P < 0.001) and C (r, 0.517; P < 0.050).

Analysis of adjusted data revealed that in both sagittal planes, the gap width at reference point A was significantly narrower in large-breed dogs, compared with width for the small- and chondrodystrophic-breed dogs, and the gap widths at reference points E and F were significantly narrower in large-breed dogs, compared with gap widths for small-breed dogs (Figure 4; Table 3). There were no significant differences among breeds at reference points representing the center of the joint (ie, points B, C, and D).

View Full Size

Mean width of interosseous distances at reference sites A to F after adjustment of the data for S1 (panels A, B, and C) and S2 (panels D, E, and F) for large-breed dogs (panels A and D), small-breed dogs (panels B and E), and chondrodystrophic-breed dogs (panels C and F).

Citation: American Journal of Veterinary Research 67, 2; 10.2460/ajvr.67.2.211

• Download Figure
• Download figure as PowerPoint slide

Discussion

Studies^10-12,25 in men reveal that the humeroulnar joint is physiologically incongruous in the unloaded state. This has been proven by a pattern of bicentric distribution of subchondral mineralization^11,12 or experimentally by examination of casts.^10,25 The incongruous notch is spread apart by the humerus during loading, the contact areas merge in the depths of the notch, and the joint becomes congruous upon heavy loading^10,12,25 It has been suggested^{10-12,14,26,27} that this physiologic incongruity brings about an optimized distribution of stress over the articular surface and may lead to better nourishment of articular cartilage.

Monocentric humeroradial and bicentric concave humeroulnar incongruity have been determined by use of casts in normal canine elbow joints of mature mixed-breed dogs weighing 19 to 30 kg.^9,28 The 3 contact areas during typical weight bearing are the articular surface of the radial head, lateral surface of the anconeal process, and MCP, whereas the central area of the ulnar trochlear notch does not come in contact with the humerus in the larger dogs investigated in those studies.^9,28 Thus, the potential role of incongruity in the canine elbow joint has been interpreted with caution because the ulna has not yet been considered to be substantially involved in weight bearing.²⁹ However, in a study³⁰ in which investigators used in vitro methods to simulate conditions obtained in dogs (mean body weight, 20 kg) during the midstance phase at a trot, the ratio of the mean force acting on the radius and ulna, respectively, remained close to a 50:50 distribution regardless of the applied load. Interestingly, although not significantly different, the percentage of force acting on the ulnar articular surface increased with increasing loads (from 47.8% to 49.3%),³⁰ which indicated that the ulna is an important weight-bearing structure.

Configuration of the elbow joint at the site of the MCP was not investigated in the study reported here because it is impossible to obtain sagittal sections of the joint that contain the radial head and MCP or the anconeal process in combination with the MCP. However, in the large-breed dogs, the interosseus gaps were narrowest at those sites (reference points A, D, and E) described as contact areas in other studies^9,28 and widest at the site (reference point B) described as a noncontact area in those studies.^9,28 This, in turn, leads us to conclude that the contact areas in the small-breed dogs are located more centrally in the humeroantebrachial joint (reference points C and D), whereas the condition seen in the chondrodystrophic-breed dogs reflects a transition type between the condition seen in the large- and small-breed dogs.

In humans, specimens with a continuous layer of articular cartilage have a lower degree of incongruity than those with a divided articular surface.^10-12 Such divided joint surfaces have also been detected in dogs and have been interpreted divergently as cartilage atrophy–cartilage defect^4,31 or as cartilage-free areas potentially indicating physiologic incongruity.¹³ Histologic examination of the cartilage-free areas revealed no signs of inflammation, which suggests that cartilage alterations at these specific locations (ie, most commonly in the center of the trochlear notch or adjoining lateral areas) are not pathologic and do not interfere with joint function.³² In the study reported here, various-sized cartilage-free areas in the center of the trochlear notch or adjoining lateral areas were evident in all joints of the large-breed dogs, whereas this was evident only at the site of the lateral coronoid process (if at all) in the other dogs. It is noteworthy that the specimens used in the study reported here were selected because of their normal anatomic appearance. Thus, dogs with subclinical changes may have been included in our study.

Magnetic resonance imaging is a reliable tool for use in examining all joint components, especially the bone-cartilage boundary.^17,20 Accounting for the thickness of the cartilage layer in the study reported here would have more accurately reflected joint geometry because the thickness of cartilage has important influences on the congruity between components of the joint.¹⁰ Thus, because of the narrowness of the canine elbow joint, a clear differentiation of the opposing cartilage surfaces is only possible on the surfaces of the joint that are not in contact.²⁴ When articular surfaces are in close contact, the cartilage signal is composed of both adjacent layers.²⁴ Because a clear differentiation was not possible at all reference points, we decided to determine the interosseous gaps instead, which depend on the thickness of the articular cartilage and the joint space as well (similar to the condition seen in radiographs but without the effect of superimposition of bone).

Other authors²¹ have stated that with a high-resolution, 3-D, fat-suppressed gradient-echo sequence, cartilage volumes determined with MRI are generally in good agreement with those measured from anatomic sections. The advantage of fat-suppressed, 3-D, gradient-echo sequences is a substantial increase in contrast to noise between cartilage and joint fluid and cartilage and subchondral bone, revealing articular cartilage as a band of high signal intensity.^15,33 Also, the signal intensities are higher in live animals than in cadaveric specimens, but the contrast between cartilage and its adjacent structures is similar.²¹ However, MRI may also introduce some bias. In the evaluation of a small joint, such as the elbow joint, the slices must be as thin as possible to minimize partial volume effects.²⁴ This artefact results when the structure of the tissue within the slices differs, which causes averaging of the signal from various tissues and blurring of the edges.²⁴ When images are orientated strictly perpendicular to the bone-cartilage interface, this procedure results in a sharper contrast of the cartilage edges as a result of a decrease in partial volume effects.²¹

Compared with our measurements (CV < 2.6%), the mean CV for measurements of focal cartilage thickness in anatomic specimens in other studies ranged from 2.3% to 5.8%²¹ and, for MRI, from 3.2% to 7.7%²¹; it also replicated volume measurements from 1.3% to 3.4%.³³ However, these validation studies were performed in human knee joints in which the maximal cartilage thickness ranges from > 2 mm (femoral trochlea) to > 6 mm (patella).³³ Even in the large-breed dogs investigated here, mean thickness of both articulating cartilages (plus joint space) was only approximately 2 mm.

Concerns have been expressed that determination of cartilage thickness by use of MRI is inherently less reproducible than determination of volume because of difficulties in reselecting identical section location in longitudinal evaluations.^21,33 This reservation is certainly also fully justified when trying to select identical section locations in specimens of differing size with obvious differences in the conformation of the elbow joint, notably the distinct inclination of the limb axis in chondrodystrophic-breed dogs. However, the cartilage volume represents an integral of cartilage thickness and size of the joint surface areas³⁴ and therefore was considered to be inappropriate when attempting to investigate dogs of various sizes. Nevertheless, we consider our landmarks to be reliable, and because serial sections were available and slice thickness was only 0.93 mm, we were able to select the most appropriate section; thus, the deviation when selecting the slices from various dogs should be < 0.93 mm.

The extent to which our results can be accepted as revealing differences in joint geometry and higher degrees of geometric incongruity in the large-breed dogs, compared with results for the other groups of dogs, is subject to certain limitations. Some limitations arise from the fact that we were not able to position all the limbs in an identical manner. Generally, we attempted to obtain a semiextended position simulating the angle of the elbow joint during the stance phase of dogs, which, in Greyhounds, is reported³⁵ to be an angle of 132°. Variations in positioning in the study reported here were mainly attributed to the size of the coils available. Thus, the extent to which the measurements obtained in our study allowed for extrapolation to the situation in standing small- and chondrodys-trophic-breed dogs remains unknown.

The second limitation was the low number of small- and chondrodystrophic-breed dogs, compared with the number of large-breed dogs. Additionally, differences in age (ie, the large-breed dogs were considerably younger) also may have introduced some bias in our study. On the basis of results of studies^26,27 in human hip joints, a joint that is physiologically incongruous in young adults becomes more congruent with advancing age.

Adjusting data also introduces some bias because the adjustment factor depends on precise localization of the center of the circle (ie, point M). Although at least 10 attempts were made to localize the center, it was our experience that in almost none of the attempts was it possible to design a circle that passed exactly through all 4 reference points (ie, A, C, D, and F). Hence, we most often created a circle that passed through reference point A and one of the other reference points (C, D, or F) with a minimum distance in the deviation from those points. In the case in which point M would have been falsely located nearer to point A, this would have resulted in overestimation of the gap widths after data adjustment. Hence, this bias may particularly affect comparison of corresponding gaps among groups of dogs after adjustment of data, although these adjusted data still allowed better comparison among dogs of obviously differing sizes. When interpreting our results, it is important that both data sets are considered. In fact, we expected that the gaps originally measured would be wider at all reference points in the large-breed dogs, which was not true for the site of the anconeal process and the cranial limit of the humeroradial joint. In fact, almost the inverse situation was evident after data adjustment because significant differences were limited to the border sites of the joint. Although some limitations in exact quantification of the difference were accepted, it could be documented in this study that differences in joint geometry existed among various constitutive types of dogs. Additionally, correlation between body weight and width of the interosseous gaps was evident only in the large-breed dogs and was found in both sagittal planes at the center of the ulnar trochlear notch, which suggested that the degree of incongruity in nonarthritic humeroulnar joints increases with increase in body weight. The narrowness of diametrically positioned aspects of the joint in the large-breed dogs also suggested less ability to compensate intra-articular steps (eg, in short ulna syndrome) in the large-breed dogs.

It is also important to be aware that the articular cartilage thickness and hence the interosseous distances can only be fully and accurately determined in images that transect the bone-cartilage interface at an angle of 90°.^21,33 However, when analyzing the elbow joint, it is impossible to orientate all sectional images perpendicular to the articular surfaces because these surfaces are curved in more than 1 plane. It must therefore be taken into account that the angles between the section images and cartilage layer may distort the apparent thickness.²¹ For our measurements, these effects were mainly neglected when comparing reference sites among groups of dogs because an identical section location was chosen for all dogs and all measurements were performed in the direction in relation to the center (point M).

It must be emphasized that our data reflected the apparent width but not the true width of interosseous gaps of the canine elbow joint and that width of the interosseus gaps does not correspond to width of the joint space. However, the main objective of the study reported here was to document differences in joint geometry among 3 constitutive types of dogs and to reveal geometric incongruity to be physiologic within a certain range. These data provide a benchmark against which to compare these variables in dogs with specific lameness conditions.