Numerous in vitro studies have utilised bone models for the assessment of orthopaedic medical devices and interventions. The drivers for this usage are the low cost, reduced health concerns and lower inter-specimen variability when compared to animal or human cadaveric tissues. Given this widespread exploitation of these models the push for their use in the assessment of spinal augmentation applications would appear strong. The aim of the research was to investigate the use of surrogate-bone vertebral models in the mechanical assessment of vertebroplasty.

Nine surrogate-bone whole vertebral models with an open-cell trabeculae configuration were acquired. Initial μCT scans were performed and a bone marrow substitute with appropriate rheological properties was injected into the trabeculae. Quasistatic loading was performed to determine the initial fracture strength in a manner previously used with human cadaveric vertebrae. Following fracture, vertebroplasty was undertaken in which there was a nominal 20% volume fill. Following augmentation the VBs were imaged using uCT and then subjected to an axial load using the same protocol.

The surrogate models had a substantially thicker cortex than that of human osteoporotic vertebrae. During compression, the surrogate-bone models did not exhibit the characteristic ‘toe-region’ observed in the load-deformation profile of cadaveric vertebrae. The mean initial and post-augmentation failure strength of the surrogate vertebrae were 1.35kN ± 0.15kN and 1.90kN ± 0.68kN, respectively. This equates to a statistically significant post-vertebroplasty increase by a factor of 1.38. In comparison with human osteoporotic bone, no significant difference was noted in the relative increase in fracture strength between the artificial and human VB following augmentation.

Despite the apparent equivalence of the strength and stiffness of the artificial vertebrae compared to that of the cadaveric specimens, there are significant differences in both pre- and post augmentation behaviour. In particular, the load-deformation curve shows significant differences in shape particularly at the toe end and in post failure behaviour. There are also issues surrounding where the marrow and cement flows during the injection process thus affecting the final distribution of the cement.

INTRODUCTION: In the spinal column, bone metastases (BM) and lesions arising from multiple myeloma (MM) can cause severe weakening of the vertebral body (VB) leading to an increased risk of fracture¹. These vertebral fractures may induce severe pain, deformity and increased risk of neurological deficit². At present, however, there is very little known about the mechanical behaviour either of the infiltrated vertebrae or that following vertebroplasty (VP). The purpose of this preliminary investigation was to evaluate (i) the mechanical behaviour of vertebrae with lesion involvement, and (ii) the effectiveness of VP with coblation.

METHODS: Individual vertebrae from two spines, one with MM (n=13) and one with BM secondary to bladder cancer (n=12) were dissected free of soft tissue with the posterior elements retained. Three MM vertebrae with evidence of previous fracture were excluded. Each vertebrae was fractured under an eccentric flexion load from which fracture strength and stiffness were derived³. VBs were then assigned to two groups. In group 1, lesion material was removed by coblation prior to VP and in group 2, no coblation was performed prior to VP. All vertebrae were fractured post-augmentation under the same loading protocol. At each stage microCT assessments were conducted to investigate lesion morphology and cement volume/distribution.

RESULTS: MM vertebrae were characterised by several small lesions, severe bone degradation and multiple compromise of the cortical wall. In contrast, large focal lesions were present in the BM vertebrae and the cortical wall generally remained intact. The initial failure strength of the MM vertebrae were significantly lower than BM vertebrae (L=2200N vs 950N, P< 0.001). A significant improvement in relative fracture strength was found post augmentation for both lesion-types (1.42 ± 0.51, P=0.0006). Coblation provided a marginally significant increase in the same parameter post-augmentation (P=0.08) and, qualitatively, improved the ease of injection.

CONCLUSIONS: Bladder BM and MM vertebral lesions showed significant variations in lesion morphology, bone destruction and the level of cortical wall breach, causing significant changes in the bone fracture behaviour. Account should be taken of these differences to optimise the VP intervention in terms of cement formulation and delivery. Preliminary results suggest the current VP treatment provides significant improvements in failure strength post-fracture.

Introduction: Percutaneous vertebroplasty (PVP) is a treatment option for osteoporotic vertebral compression fractures (VCFs). Short-term results are promising but longer-term studies have demonstrated an accelerated failure rate in the adjacent vertebral body (VB). Limited research has been conducted into the effects of prophylactic PVP in osteoporotic vertebrae. The objective of this study was to investigate the biomechanical characteristics of prophylactic vertebral reinforcement and post-fracture augmentation.

Methods: Human vertebrae were assigned to two scenarios: Scenario 1 used an experimental model for simulating VCFs followed by cement augmentation; Scenario 2 involved prophylactic augmentation using vertebroplasty. μCT imaging was performed to assess the bone mineral density (BMD), vertebral dimensions, fracture pattern and cement volume. All augmented VBs were then axially compressed to failure.

Results: Product of BMD value and endplate surface area gave the best prediction of failure strength when compared to actual failure strength of specimens in scenario 1. Augmented VBs showed an average cement fill of 23.9%±8.07% S.D.. In scenario 1, there was a significant post-vertebroplasty factorial increase of 1.72 and in scenario 2 a 1.38 increase in failure strength. There was a significant reduction in stiffness following augmentation for scenario 1 (t=3.5, P=0.005). Stiffness of the VB in scenario 2 was significantly greater than observed in scenario 1 (t=4.4, P=0.0002).

Discussion: Results suggest that augmentation of the VB post-fracture significantly increases failure load, whilst stiffness is not restored. Prophylactic augmentation was seen to increase failure strength in comparison to the predicted failure load. Stiffness appears to be maintained suggesting that prophylactic PVP maintains stiffness better than PVP post-fracture.

Introduction: A feature of osteoporosis is vertebral compression fractures (VCF). Experiments looking at predicting compressive strength of human lumbar vertebrae have showed a correlation between compressive strength, bone density and size of vertebral endplates. The objective of this study was to compare the actual versus predicted failure strength of osteoporotic human vertebrae in relation to creating a validated experimental model for a vertebral compression fracture.

Methods: Twenty-six human vertebrae underwent CT scanning to evaluate bone mineral density (BMD) from a large and small region of interest (ROI) within the vertebral body (VB). Cranial, caudal and verage endplate surface area (SA) measurements were recorded. Specimens were axially compressed to failure and a regression analysis undertaken in which the failure load was fitted using both BMD alone and the product of the BMD and endplate SA.

Results: Measurements of BMD from a large or small ROI showed a poor correlation when compared to vertebral failure strength. The product of BMD and endplate SA showed significant correlations with failure strength. The regression explains a significant proportion of the variation of the response variable.

Discussion: Results from this study are consistent with published data which have established a good correlation between the product of endplate SA and BMD to vertebral compressive strength. BMD values from a large ROI and average or caudal endplate area provide the best prediction of failure strength. Experience from this study suggests that the experimental model is reproducible and accurate, however, further work is required on a larger data set to verify initial findings.

Introduction: Spinal cord injury (SCI) continues to challenge the healthcare and the adjunct social welfare systems. Significant advances have been made in our understanding of the pathological cascade following the initial insult. However, this has yet to be translated into clinically significant treatments and one possible reason for this is that little is known about the actual interaction between the cord and the spinal column at the moment of impact; a factor that is becoming increasingly recognised as important. Burst fractures are a common cause of SCI and are sufficiently well defined to allow significant advances to be made in developing laboratory models of the fracture process. Following on from these advances an in-vitro model of the interaction between the cord and burst fracture fragment was developed and used to perform preliminary experiments to establish those factors that are important in determining the extent of probable cord damage.

Methods: A rig was developed that reliably reproduced a range of fragment-cord impact scenarios previously observed in the development of a model of the burst fracture process. In summary, a simulated bone fragment of mass 7.2 g was fired, transversely, at explanted bovine cord (within 3 hours of slaughter) with a velocity of 2.5, 5.0 or 7.5 ms-1. The cords were mounted in a tensile testing machine using a novel clamping system and held at 8 % strain. A surrogate posterior longitudinal ligament (PLL) was included and simulated in three biomechanically relevant conditions: absent, 0 % strain and 14 % strain. The posterior elements were represented by an anatomically correct surrogate. The impacts were recorded by using either a high speed video camera (4500 frames/s) or a series of fine pressure transducers.

Results: The fragments were recorded to undergo the same occlusion profile as previously reported in the burst fracture model, except that the cord itself reduced the level of maximum occlusion possible. All tests displayed the fragment recoiling following maximum occlusion. The maximum occlusion and the time to this position were found to be significantly dependent on both the fragment velocity and the condition of the PLL. Similar results were observed for peak pressure. One surprising result was that maximum occlusion or time to this event did not change with or without the cord being encased in the dura mater; a structure that is thought to protect the cord from external impacts.

Discussion: The model developed here of the cord-column interaction for the burst fracture produced useful initial insights into the factors that affect the impact on the cord. The PLL has a significant role to play in both reducing the peak pressures and the spreading the energy imparted over a longer period. The model has several areas in which it could be improved and these include 1) the incorporation of the perfusion pressure which tends to hydraulically stiffen the cord and 2) the inclusion of the cerebrospinal fluid, which may operate in unison with the dura in protecting the cord from impacts. Future work includes the incorporation of the CSF into the model, the development of surrogate cords and the generation of computational models using novel programming techniques.