One of the most frequently occurring developmental disorders in the canine elbow joint is the fractured medial coronoid process of the ulna (Boulay, 1998; LaFond et al., 2002; Meyer-Lindenberg et al., 2002). It has been suggested that mechanical overloading of the medial coronoid process (MCP) at a critical age (Grøndalen and Grøndalen, 1981) leads to this condition. If mechanical overloading is one of the main parameters, it is important to know the normal loading pattern in this part of the developing joint. On anatomical grounds (Fig. 1), it can be expected that the following forces are present: a downward/backward force applied by the humeral trochlea; a horizontal, laterally directed force applied by the radial head; and a tensile force applied by the anular ligament.
The microcomputed tomography (micro-CT) technique was used to assess bone volume fraction, trabecular characteristics, and orientation within the MCP. Changes in these parameters are thought to be engendered by the combined processes of increase in bone size and mass and adaptation to external load during growth. Parameters obtained by micro-CT have been used earlier to describe the trabecular structure (Engelke et al., 1993; Rüegsegger et al., 1996; Hildebrand et al., 1999). However, only a few studies have dealt with the developing trabecular structure in skeletally immature individuals, and all of these pertained to long bones or vertebrae (Nafei et al., 2000a, 2000b; Tanck et al., 2001). The question is whether these results can be extrapolated to a bony protrusion like the MCP. Therefore, the aim of this study is to determine the course of development of a functional trabecular structure within the MCP of young golden retrievers. The trabecular structure will be described by standard bone parameters calculated from micro-CT scans.
MATERIALS AND METHODS
The right MCPs of nine golden retrievers aged 4, 6, 8, 10, 13, 16, 18, 22, and 24 weeks were studied. The dogs were not lame and showed no signs of elbow problems at physical examination. The project was approved by the Committee for Animal Experimentation (DEC) of the Faculty of Veterinary Medicine, University of Utrecht, The Netherlands.
After euthanasia, the dogs were perfused intravenously with a 4% buffered formaldehyde solution. The joints were dissected and inspected before the MCPs were separated from the rest of the ulna with a diamond-blade precision saw. Each MCP was embedded in methylmetacrylate (MMA), which was turned into a maximally 3 cm long cylinder with a diameter of 17 mm using a hollow coring bit. It was ensured that the cylinder contained the complete MCP (Fig. 2).
The MCPs were scanned in a micro-CT scanner (μCT 20, Scanco Medical, Bassersdorf, Switzerland) (Rüegsegger et al., 1996). The sample holders with a diameter of 17.4 mm had been marked with a line on the outside to make the anatomical planes of the MCP coincide in a consistent manner with the coordinate system of the scanner. The data were acquired in a voxel size of 34 μm × 34 μm × 34 μm. Due to limitations in the software, volumes of interest (VOIs) were selected for histomorphometric measurements as follows. First, the MCP was divided longitudinally into a medial and lateral half by the software, then a cube of 100 × 100 × 100 voxels was fit in as proximally as possible below the cortex in one half. A second cube was fit under and partly overlapping this first cube. A third cube was positioned in this way below the second one and so on until the total length of the MCP was filled with blocks. This was repeated for the second half.
The threshold for discrimination of bone for each MCP was determined visually by comparing scans obtained at different threshold values with the original one. The optimal threshold is most often a compromise between two possible thresholds: a lower, which tends to make the trabeculae too thick, and a higher threshold, which disconnects the thinner trabeculae.
The following parameters were evaluated: bone volume (BV) fraction (BV divided by total volume, or TV), bone surface (BS) density (BS/BV), trabecular thickness (Tb Th; expressed in mm), trabecular separation (Tb Sp; mm), trabecular number (Tb N; mm−1), and the degree of anisotropy. The parameters were calculated with the mean intercept length (MIL) method based on a count of the number of intersections between a linear grid and the bone-marrow interface as a function of the three-dimensional orientation of the grid in the VOI. Generalization of this originally two-dimensional method to three-dimensional reconstructions can be made on the assumption that a polar plot of the MIL data approaches an ellipsoid (Whitehouse, 1974; Odgaard, 1997). The mean intercept length tensor of this ellipsoid gives the principal direction of the trabecular structure. The eigenvalues of this tensor (H1, H2, H3) are associated with the lengths of the axes of the ellipsoid (Harrigan and Mann, 1984), of which H1 is the largest axis and H3 the smallest. Three degrees of anisotropy for the planes defined by these axes can be calculated: DA13 is the ratio between H1 and H3, DA12 between H1 and H2, and DA23 between H2 and H3. A value of 1.0 means no trabecular alignment in a specific direction (isotropy).
The selection procedure resulted in a threshold (27.2% of the gray value) that appeared to be identical in all VOIs of all dogs. The findings of the high BV/TV and the small Tb Sp led us to evaluate the used thresholding procedure for the VOIs of the youngest dogs. According to the method proposed by Hara et al. (2002), first, the optimal threshold was determined and then the bone parameters were calculated with the thresholds set 0.5% and 1% above and below the optimal one. This was done for the samples of 4, 6, and 8 weeks.
The MCPs of all dogs appeared to be normal; there were no irregularities of the articular cartilage and no signs of fracture or secondary joint disease. Figure 3 shows the reconstructions of a representative VOI of the MCP of dogs aged 4, 10, and 18 weeks. Due to the thresholding procedure, the soft tissues (articular cartilage and intertrabecular marrow) are not visible. Four weeks after birth (Fig. 3, left), the trabecular structure is dense, with small rod-like trabeculae. The trabecular structure at 10 weeks is more open (Fig. 3, middle), but the trabeculae appear still rod-like. At 18 weeks after birth (Fig. 3, right), the trabeculae form more plate-like structures, and small rods are no longer present.
The values of the main bone parameters are depicted in Figures 4 and 5. The bone parameters of the VOIs of the medial and lateral halves were not significantly different at any age; for all parameters, the mean of all VOIs was used to describe the bone structure. Bone volume and mean trabecular separation show a reciprocal pattern in time, reflecting a high initial bone density (and low trabecular separation), a sharp drop in density at 8–10 weeks, followed by a gradual increase to high values at 24 weeks (Fig. 4). Although this is suggestive of an identical structure at 4 and 24 weeks, this is not true as the older bone shows thicker trabeculae and a more plate-like structure. The difference is further reflected by the much smaller numbers of trabeculae and the lower surface/volume ratio at the higher age (Fig. 5). Although trabecular thickness drops temporarily at 6–8 weeks, the final values at 24 weeks are also much higher than at 4 weeks (Fig. 4).
An anisotropic orientation of the trabeculae, as evidenced by a value of the DA13 above unity and an orientation in the direction of the proximodistal axis of the ulna, is already present at 6 weeks after birth (Table 1). The mean alignment for all ages made an angle of 20.1° (SD 8.3) with the proximodistal axis more or less perpendicular to the humeroulnar articular surface. The second degree of anisotropy (DA12) was present from 13 weeks after birth (Table 1). The trabecular orientation was directed along the craniocaudal axis of the MCP with a mean angle of 33.0° (SD 11.2), thus toward the attachment of the anular ligament (Fig. 6).
Table 1. Values of the degrees of anisotropy for the ages studied
The differences in BV/TV, BS/BV, and Tb Th were < 4% per step of 0.5% threshold; the differences in Tb N and degree of anisotropy were < 1% per step of 0.5% for the ages evaluated with this method.
For ethical and practical reasons, this study was performed with a group size of n = 1. However, the measured parameters on the nine animals, all offspring of unrelated dams, show either stability (in case of the degrees of anisotropy) or very clear trends with age. It can therefore be argued that these values and trends are representative for the golden retriever breed. In fact, golden retrievers were used as a model for growth of the MCP, because in the Netherlands this breed suffers only to a limited extent from fractures of the MCP and can therefore provide a standard of normal development for more affected breeds with the same general characteristics of skeletal growth, such as Labrador retrievers.
Because only a single specimen was investigated for each age, it is important to know that all elbows had developed normally. This was confirmed by inspection and clinical assessment of the elbows. The MCPs were macroscopically normal, and no signs of fracture were found in the reconstructions either.
Variation in the trabecular structure associated with variation in sampling sites has been reported (Behrens et al., 1974; Townsend et al., 1975). In order to prevent this unwanted scatter and to make a valid comparison between the trabecular structures of unequally sized MCPs, we opted for fitting in as many VOIs as possible to cover the complete trabecular structure, as such excluding dependency on topography.
Contrary to what could be expected, the BV/TV ratio was high at 4 weeks after birth; at 8 weeks, the density had dropped considerably, after which a gradual return to a high value was observed. In pig vertebrae, the bone volume fraction at 6 weeks was much lower (< 0.2) (Tanck et al., 2001). In human fetuses, the range of BV/TV values varied from 0.3 to 0.54, allowing the conclusion that bone is much denser than in young adult BV/TV (0.15) (Nuzzo et al., 2003). Since the newborn puppy is born after a short gestation period and is relatively immature at birth (the average puppy is able to stand by the 10th day and to walk with a steady gait by the 21st after birth) (Mosier, 1978), it might be possible that a primary (fetal) trabecular structure is still present in the first weeks after birth. Unfortunately, it cannot be assessed from micro-CT scans whether the tissue present is calcified cartilage, woven bone, or lamellar bone.
One of the disadvantages of using micro-CT is the fact that the obtained results depend on the chosen threshold for the assignment of bone. However, it has been convincingly shown that micro-CT-derived results can be compared with those from conventional histomorphometric techniques (Kuhn et al., 1990; Uchiyama et al., 1997; Müller et al., 1998). Furthermore, the influence of small threshold variations in the region of the optimal threshold on trabecular bone parameters appeared to be very small (Hara et al., 2002). The results of the samples of the youngest dogs led us to evaluate the threshold method. The changes per percentage point threshold were comparable with the ones found by Hara et al. (2002), implying that the values for 4, 6, and 8 weeks calculated in our study are not strongly threshold-dependent and are as reliable as the values at other ages. Measuring bone volume fractions with the micro-CT method (Kuhn et al., 1990; Rüegsegger et al., 1996; Odgaard, 1997) is generally accepted. However, by comparing the results of micro-CT and of Archimedes' principle, Ding et al. (1999) found that the amount of bony tissue was underestimated in low-density bone and overestimated in high-density bone, while no differences appeared in bone densities between 0.2 and 0.3. However, the observed relative changes in bone volume fraction are much larger than this 10% error. It was not possible to establish bone volume fraction accurately using Archimedes' principle, because the required defatting procedure would exclude histological examination after testing (Sharp et al., 1990).
After 8 weeks, the BV/TV increased with age. This result is in agreement with the results of micro-CT studies of the tibia and lumbar vertebrae in pigs (Tanck et al., 2001), with results of serial reconstruction studies in sheep (Nafei et al., 2000a) and with histological growth studies of the ilium in dogs (Fukuda and Iida, 1994). Increasing body weight produces a larger bone strain, which, according to the mechanostat hypothesis of Frost (1996), would in turn stimulate the reinforcement of the MCP by the formation of extra bone to reduce the strain to a preset level. The bone volume fraction plays a major role in the determination of the mechanical properties of trabecular bone (Behrens et al., 1974; Vesterby et al., 1991).
It has been shown in bone of aging men that the Tb Th is maintained at a steady value as long as possible, while the bone volume fraction and the elastic modulus decrease (Parfitt, 1987; Ding and Hvid, 2000). In human fetal bone, the trabeculae are almost as thick as in mature bone (Nuzzo et al., 2003). This would suggest that the Tb Th is maintained at the same level from fetal until old age. However, in our study, it was found that in canine bone the trabeculae grow thicker, while the bone volume fraction increases and trabecular anisotropy develops.
The structure of the trabecular bone was characterized by the BS/BV ratio in this study and not by the structure modeling index (SMI) method, which was described by Hildebrand and Rüegsegger (1997). We calculated the SMI, but the obtained values were largely negative for most of the VOIs, and not between 0 and 3. This might be an inherent problem with the SMI method when it is applied to bone tissue with a high trabecular bone volume.
Already at week 6, the trabeculae start to orient along an axis, perpendicular to the humeroulnar articulation surface. This orientation should be conceived as an adaptation to force, exerted by the humerus to the articular surface during weight bearing. Early alignment in the principal direction was not found by Tanck et al. (2001) in pigs of 6 and 23 weeks. The conclusion of these authors that there is a time lag between bone formation and architectural adaptation is not corroborated by our data. The described adaptation process apparently takes place earlier in life in dogs, or earlier in life for specific bony projections such as the MCP.
The second axis (of orientation), situated in the transverse plane, had a craniocaudal orientation, pointing to the attachment area of the anular ligament at the MCP. The values for the DA12 are lower but increase steadily during growth. Thus, the orientation is less outspoken, albeit clearly above detection level.
The occurrence of a functioning radioulnar joint might lead to movement-induced compressive loading of particularly the inner portion of the MCP. If high pressures in the transverse plane occur in the proximal radioulnar joint, the axis of orientation for bony trabeculae could be expected in a transverse direction, perpendicular to the articular surface of this joint. However, this was not found. Furthermore, there was no difference between the bone parameters of the VOIs of the medial and lateral halves of the MCP, which also indicates that the influence of the radioulnar joint is limited.
In conclusion, the relatively fast development of a functional trabecular structure of the MCP reflects significant mechanical loading at early age. The MCP is loaded primarily in a direction perpendicular to the humeroulnar surface during normal weight bearing, as evidenced by the primary axis of trabecular orientation. The secondary axis of orientation points to the existence of tensile stress, evoked by the anular ligament. The extent to which these forces are present during the development of the MCP may be derived from its trabecular structure and trabecular orientation, which reflect the normal mechanical loading pattern according to Wolff's law (Goldstein, 1987; Roesler, 1987; Frost, 1996).
The authors thank F. Gorter for technical assistance and the Orthopedic Research Laboratory, Nijmegen University, for use of the micro-CT system.