- Top of page
- Materials and methods
The lumbar vertebrae are major load-bearing structures within the spinal column. The current understanding of the microstructure of these bodies and their full role in load-bearing is incomplete. There is a need to develop our understanding of these issues to improve fracture prediction in musculoskeletal diseases such as osteoporosis. The lumbar vertebrae consist primarily of trabecular bone enclosed in a thin cortical shell, but little is known about how microstructural parameters vary within these structures, particularly in relation to the trabecular compartment. The specific aim of this study was to use micro-computed tomography to characterize the trabecular microarchitecture of the ovine L3 vertebra in cranial, mid-vertebra and caudal regions. The L3 vertebra was obtained from skeletally mature ewes (n = 18) more than 4 years old. Three-dimensional reconstructions of three pre-defined regions were obtained and microarchitectural parameters were calculated. Whereas there was no difference in bone volume fraction or structural model index between regions, trabecular number, thickness, spacing, connectivity density, degree of anisotropy and bone mineral density all displayed significant regional variations. The observed differences were consistent with the biomechanical hypothesis that in vivo loads are distributed differently at the endplates compared with the mid-vertebra. Thus, a more integrative approach combining biomechanical theory and anatomical features may improve fracture risk assessment in the future.
- Top of page
- Materials and methods
There is a need for improved characterization of the microstructure of vertebral bodies (Sun et al. 2004); this will benefit the research community in areas including vertebral fracture, vertebroplasty and osteoporosis. The current clinical standard in assessing the structural integrity of a vertebral body is to use dual energy X-ray absorptiometry (DEXA) scanning to measure bone mineral density (BMD). This measurement is two-dimensional in nature, has inherently low resolution and is, at best, a proxy measurement of strength (Hordon et al. 2000). Thus, it is important to develop other methods of assessment that take account of factors such as bone quality.
The lumbar vertebrae are major load-bearing structures within the spinal column. They consist primarily of trabecular bone enclosed in a thin cortical shell. An essential but poorly understood issue in the full characterization of the vertebral body is the respective roles of the trabecular and cortical compartments in load-bearing capacity (Eswaran et al. 2006). Furthermore, it is not known to what extent microstructural parameters display regional variations within each vertebra, which is particularly relevant in the trabecular compartment. Adopting a more integrative approach combining the functional role of trabecular bone at different sites within these structures will improve non-invasive assessment of vertebral bone strength in vivo.
Despite much data in the literature on the topic, the variation of trabecular microarchitecture within the vertebral body remains unclear. Regional vertebral parameters have been investigated using histological techniques (Hordon et al. 2000), micro-computed tomography (microCT) of standardized bone samples (Gong et al. 2006) and some groups have addressed the situation at the whole bone level (Eswaran et al. 2006). The range of values of relative bone volume (BV/TV) for human vertebral bone is between 6.5% and 16% (Hulme et al. 2007). Regional changes of trabecular microarchitecture have also been investigated directly at an elemental (rod/plate) level (Stauber & Muller, 2006). This method involves characterizing morphometric parameters of specific trabeculae, rather than using average values over a large group. This may help to improve the understanding of bone failure mechanisms, as well as the effects of pharmaceuticals.
It has been shown that human vertebrae have greater bone volume, more connections, reduced trabecular separation and more plate-like isotropic structures posteriorly than their corresponding anterior regions (Hulme et al. 2007). It was also demonstrated that heterogeneity exists between superior and inferior regions. The specific aim of this study was to use microCT to describe the microarchitecture of the ovine L3 vertebra based on the structure–function criteria rather than traditional anatomical divisions. We defined regions based on the hypothesis that the biomechanical demands on bone near the endplates will be different from those in the central region. For this reason, the three regions used in this study were cranial, mid-vertebral and caudal.
- Top of page
- Materials and methods
One specimen was removed from this analysis due to an error in the imaging process. Scanning could not be repeated as specimens were destructively tested in a related study. Figure 2 shows the average data for BMD, Tb.N, Tb.Th and Tb.Sp at the cranial, mid-vertebral and caudal regions. Whereas there was no difference in BMD between cranial and caudal sites, there was a significant difference at the mid-vertebra compared with both of those regions. Tb.N was significantly higher both cranially and caudally compared to the mid-vertebra, and it was also significantly higher at the cranial compared with caudal region. Tb.Th was significantly higher in the mid-vertebral region compared with cranial and caudal regions, but there was no difference between the latter two regions. Tb.Sp was significantly higher at the mid-vertebra compared with cranial and caudal regions, and it was also significantly higher caudally compared with the cranial region.
Figure 2. Average bone mineral density (BMD), trabecular thickness (Tb.N), trabecular thickness (Tb.Th) and trabecular spacing (Tb.Sp) values at cranial, mid-vertebral and caudal regions.
Download figure to PowerPoint
Figure 3 shows the average data relating to trabecular porosity, Conn.D, SMI and DA. Porosity is defined as 1–BV/TV, meaning the relationship between volume measurements provides no new information; therefore the BV/TV data are not shown. Porosity was not different between any of the three regions. Conn.D was significantly higher cranially and caudally compared with the mid-vertebral region, and it was also significantly higher at the cranial compared with caudal region. SMI was not different between any of the regions measured. Whereas there was no difference in DA between the cranial and caudal regions, it was significantly higher in both locations compared with the mid-vertebra.
Figure 3. Average porosity, connectivity density (Conn.D), structural model index (SMI) and degree of anisotropy (DA) values at cranial, mid-vertebral and caudal regions.
Download figure to PowerPoint
A substantial number of significant correlations were found among the trabecular microarchitectural parameters (Table 1). Interestingly, those parameters which correlated well in one of the regions did not always correlate well in other regions. For example, BMD was negatively correlated with Conn.D in the cranial, mid-vertebral and caudal regions, but it was positively correlated with Tb.Th at the mid-vertebral region only. Similarly, trabecular porosity was correlated with Tb.N, Tb.Th (negatively) and Tb.Sp and SMI (positively) in all regions; however, it was positively correlated with DA only at the mid-vertebra. Correlation coefficients among the parameters are given in Table 1.
Table 1. Correlation coefficients between trabecular microarchitectural parameters and BMD in cranial, mid-vertebral and caudal regions
| Porosity|| ||−0.771***||−0.859****||0.873****||−0.0178||0.806***||0.391|
| Tb.N|| || ||0.571**||−0.957****||0.609**||−0.364||−0.642*|
| Tb.Th|| || || ||−0.719***||−0.0242||−0.560**||−0.376|
| Tb.Sp|| || || || ||−0.443||0.476||0.602**|
| Conn.D|| || || || || ||0.475||−0.616**|
| SMI|| || || || || || ||−0.0430|
| Porosity|| ||−0.768***||−0.863****||0.844***||0.372||0.953***||0.224****|
| Tb.N|| || ||0.394||−0.972****||0.225||−0.726**||−0.0459|
| Tb.Th|| || || ||−0.537*||−0.62**||−0.771***||−0.322|
| Tb.Sp|| || || || ||−0.115||0.763***||0.0495|
| Conn.D|| || || || || ||0.457||0.402|
| SMI|| || || || || || ||0.253|
| Porosity|| ||−0.525*||−0.634**||0.573*||0.368||0.614**||−0.21|
| Tb.N|| || ||0.461||−0.956****||0.00897||−0.65**||−0.364|
| Tb.Th|| || || ||−0.561*||−0.827***||−0.873****||−0.384|
| Tb.Sp|| || || || ||0.0708||0.636**||0.32|
| Conn.D|| || || || || ||0.729***||0.205|
| SMI|| || || || || || ||0.408|
- Top of page
- Materials and methods
In the mammalian spine, vertebral bodies experience a variety of loads, comprising compressive, bending and torsional forces. Vertebral bodies are generally hyperboloidal in shape, i.e. they are wider at both ends than at the mid-section, and as such, they are well designed to deal efficiently with this loading environment (Zhou et al. 2000). The ratio of trabecular to cortical bone in vertebral bodies is approximately 95 : 5, which explains its load-bearing capacity when considering the function of the trabecular network (Antonacci et al. 1997). Its design meets the requirements for optimal load transfer by ensuring maximal strength with minimal weight (Huiskes et al. 2000). This efficiency exists via the orientation and spacing of the three-dimensional arrangement of trabeculae. Vertically orientated trabeculae attenuate axial forces, while the horizontal elements respond to the shear stresses transferred from the intervertebral disc, and also serve to prevent the vertical ones from buckling (Smit et al. 1997). The vertebral body system is such that under normal circumstances, loads are transferred through the trabecular centrum, which is the primary load-bearing structure, out towards the cortical shell at the mid-vertebra (Rockoff et al. 1969). This theory is reinforced by the observation that vertebral centrum size increases cranio-caudally as body weight percentage increases (Singer et al. 1995; Fazzalari et al. 2001; Daggfeldt & Thorstensson, 2003).
Given the variety of loading patterns on vertebral bodies in the spine, the microstructural form–function relationship must be equally multifaceted, yet this has not been fully characterized. This may have important implications in terms of fracture risk prediction and drug treatment assessments. The goal of this study was to use microCT scanning to quantify trabecular microarchitectural parameters at different regions within the ovine L3 vertebral body. Regions were chosen based on a biomechanical hypothesis that in vivo forces near the vertebral endplates will be different to those at the mid-vertebra. This may be the reason for the characteristic hyperboloid shape of the vertebrae, and as a logical progression of this, the microstructural parameters may also be different in those regions.
In this study the highest average value of Tb.N was in the cranial region, followed by the caudal and finally the mid-vertebral region. A regional difference in Tb.N in the vertebral body makes intuitive sense from a biomechanical perspective. The trabecular network is designed to allow efficient distribution of a combination of compressive and shear forces. Towards the mid-vertebral region, a large trabecular number would not be required as the forces will already have been dissipated out towards the cortex (Mizrahi et al. 1993). This is also the situation in other bones such as the human femur, where the amount of trabecular bone reduces with increasing distance from the joint surface (Silva et al. 1994). This idea is also supported by other work which has consistently shown that the role of the cortical shell is maximum, and the role of the trabecular centrum is minimum, in the human mid-vertebral region (Cao et al. 2001; Homminga et al. 2001; Eswaran et al. 2006).
The highest value of Tb.Th was in the mid-vertebral region. On average, trabeculae at this location were thicker than in cranial and caudal regions. The regional difference of this parameter has not been well characterized previously. This point is arguably even more important in relation to Tb.Th, than Tb.N, because the magnitude of the difference in parameters between regions is inherently smaller. From a design perspective, the property of having fewer, thicker mid-vertebral trabeculae is advantageous. The multi-directional forces from vertebral discs are largely attenuated by the network of smaller, more numerous trabeculae closer to the endplates. Therefore, we can assume that the nature of the loading distribution in the mid-vertebra is some combination of compression and bending (Crawford & Keaveny, 2004). This, combined with the fact that the load fraction in the trabecular compartment is lowest at the mid-vertebra (Eswaran et al. 2006), suggests that the optimal design is to have fewer thicker trabecular elements. It is also possible that thinner trabeculae toward the joint surface allow high local strains there, which would not be desirable at the mid-vertebral region where bending stresses in human thoracolumbar vertebrae are highest (Crawford & Keaveny, 2004). Differences were also present in Tb.Sp between all regions, with the highest average values being in the mid-vertebral region. This parameter displayed an inversely proportional relationship with Tb.N and thus this result was not unexpected.
BMD was significantly higher at the mid-vertebral region compared with both cranial and caudal regions. As mentioned previously, it has been shown that the percentage of total vertebral load taken by the trabecular compartment is lowest around the mid-vertebral region (Eswaran et al. 2006). However, given that there are fewer, thicker trabeculae in the mid-vertebral region, this would mean that there is a lower surface area compared with the regions near the endplates. Lower surface area would mean less potential space for trabecular surface remodelling; therefore it follows that lower remodelling would lead to longer mineralization times and thus higher values of BMD.
Porosity was observed to be unchanged across all regions in this study. This finding was expected, as increased porosity is normally associated with disease states such as osteoporosis, which is not addressed in this study. Interestingly, other workers did find regional differences in this parameter within human vertebrae (Hulme et al. 2007). They showed that BV/TV was different between anterior and posterior regions both superior and inferior to the mid-vertebral point. However, that study was carried out on osteoporotic vertebrae, which are known to experience changes in BV/TV with disease progression (Lane et al. 1999). This point is also reflected in their BV/TV data, which had an average over all regions of 12.5%. This is quite low compared to the BV/TV data in the present study, which had an average over all regions of 38.2%.
The average value for Conn.D was significantly higher in cranial and caudal regions compared with mid-vertebral regions. This parameter was also significantly higher in cranial compared with caudal regions. The distribution of these data is the same as was seen in the distribution of Tb.N across the same regions. This was expected because as the number of trabeculae in any given area increases, so too must the number of connections between them. SMI was found to be unchanged across all three regions; this suggests that although other factors may change within these structures, the general shape is the same, i.e. elements are no more rod- or plate-like in one region than in another. This finding was not unexpected because changing SMI is more often a feature of diseased bone, which was not the focus of this study.
We found strong correlations between many of the microarchitectural parameters which we measured. This compares well with data from other research (Giesen & van Eijden, 2000; Renders et al. 2007). For example, BMD correlated strongly with Tb.Th at the mid-vertebral region only. This suggests that the amount of mineral present in the mid-vertebral region of a given vertebra is related to the average Tb.Th of the trabeculae. Thus if adaptive remodelling is required to increase or decrease the mineralization in that region, then it will be the thickness of trabeculae which will be adjusted accordingly.
Studying the variation of vertebral microarchitecture has provided a unique appreciation of the complexity of the regional load-bearing role within the structure. Trabecular bone is often sampled and tested as a surrogate for bone strength. This work has shown that the microstructure of trabecular bone varies with location, and therefore the site of a bone core should be considered carefully. In fact, clinical samples for analysis are taken most often from the mid-vertebral region (Cann et al. 1985), which in some respects is the least appropriate region as its role in load-bearing is least there (Eswaran et al. 2006).
In conclusion, microarchitectural parameters within the trabecular compartment of the ovine L3 vertebra vary between the cranial, mid-vertebral and caudal regions. These variations are consistent with the load distributions throughout the structure and can be explained biomechanically. A more integrative approach that combines biomechanics with microanatomy will improve our understanding of these complex structures and may improve assessment of vertebral bone strength in vivo.