Percutaneous vertebroplasty (PVP) is an emerging interventional technique for treatment of vertebral compression fractures. Bone cement is introduced to mechanically augment fracture and pain relief is almost immediate. Recent clinical and biomechanical studies have outlined the phenomenon of fractures occurring in adjacent vertebrae following PVP [1,2]. It is widely believed that rigid cement augmentation may cause a shift in the normal loading pattern of the spine thereby resulting in adjacent fractures. However, very few studies have attempted to quantify this effect [3].

Most biomechanical studies adopt a single vertebral body as a model for PVP analysis. With this approach it is not possible to determine the effect of load distribution on adjacent structures. Where multi-segment vertebrae have been used there is little documentation of the fracture characteristics produced or their repeatability. The purpose of this study was to develop a 3-vertebra model for the biomechanical analysis of PVP. The particular focus was on developing a robust technique for generating repeatable level of fracture severity from specimen to specimen.

An alignment device was developed to fit into standard materials testing machine, which allowed constant axial compression without causing lateral bending or flexion-extension of the specimen’s ends. Porcine 3-segment specimens (T8-L2) were mechanically compressed to failure at a rate of 5mm/min applied vertically at a distance of 35% to the anterior edge of the specimen’s anterior-posterior length. During the test load-displacement data was displayed in real time on a PC. In order to generate uniform fractures, a protocol was devised in which the specimens were compressed for a further 6mm after initial yield point. After the initial fracture the segments were augmented with 3ml of PMMA cement injected through each pedicle and then recompressed. The fracture characteristics generated under these conditions were analysed using quantitative microcomputer tomogragy (μCT).

μCT images showed that fractures were generated in the central vertebra, with some propagation towards adjacent vertebra. The results support the use of a 3-segment specimen as a better representation for PVP analysis. The method will enables the load shift and fracture progression on either side of the augmented vertebra to be observed, thereby providing a more complete picture of load-bearing kinetics. Secondly, the middle, augmented motion segment remains unconstrained by platens and cement impressions; hence its anatomical boundary conditions are less compromised. Although longer segments have been shown to be more anatomically appropriate, it is difficult to apply physiologic levels of load without causing the specimen to buckle. We were able to minimise buckling effect by incorporating an alignment device to position the specimen without constraint. Given the preceding observations, the concepts of 3-segment specimen in PVP biomechanical tests provides a suitable compromise in choosing an appropriate clinical setting for in-vitro testing of biological spine specimens.