Spine traumas include bony fractures, intervertebral disc and ligaments lesions, and spinal cord injuries. They imply a high cost for the health system, as well as a highly incapacitating surgical management (length of hospitalization, physical handicap), and result in aftereffects such as sagittal curve deformity or tetraplegia. Their principal causes are motor vehicle accidents, falls and sports accidents. Multiple mechanisms of occurrence exist for vertebral fractures (compression, flexion-extension, shear, torsion) and spinal cord injury (contusion, traction, shear), which makes them complex to study. Despite major investments in research on this topic, there is a lack of the understanding of some aspects regarding these traumas. This slows down the progress in protection devices and surgical stratégies development. In this context, the main objective of this Ph.D. is to improve fundamental knowledge of vertebral fractures and spinal cord injuries mechanisms. After identifying the main observations and lacks of the literature, to research hypotheses were postulated.
H1. There is a link between traumatic mechanical loading and clinical cartographies of vertebral fracture found in the literature.
H2. During a medullar contusion, CSF has a protective role which can be quantified, and the traumatic behavior is vertebral level dependent.
Thus, this Ph.D. has two main objectives. The first one aims at understanding the mechanical causes of the different vertebral fracture patterns, and the second one aims at understanding the mechanisms of occurrence of spinal cord contusion. In each of these two parts, imaging, experimental, and numerical methods were used to (i) enhance mechanical or geometrical properties understanding for the studied biological tissues, or (ii) reproduce and characterize vertebral or spinal cord injuries.
Reaching the first objective, that is to establish a link between the known fracture patterns and their mechanical cause, required to achieve two specific research objectives. The first step was to create a prediction model for anisotropic mechanical properties of osteoporotic vertebral trabecular bone based on microstructural parameters. This model will then be used to compare structural anisotropy and mechanical anisotropy. Microstructural parameters were calculated on μCT images of cubes of human vertebral trabecular bones. The same cubes were compressed up to rupture in a mechanical testing machine. A linear regression analysis was then performed to obtain a prediction model of anisotropic mechanical properties from microstructural parameters. The prediction model resulting from the regression analysis accounted for 51 to 85% of the variance in the experimental measures. The second step was to use this model on micro-computed tomography images of 13 intact human cadaveric vertebrae. Microstructural parameters were calculated in 69 volumes of interest for each vertebra (three slices in the rostro-caudal direction, 23 areas per slice). Anisotropic Young’s modules and yield strengths from the microstructural parameters were defined in all 897 resulting volumes of interest. The cartography of microstructural and mechanical properties was proposed by calculating means and standard deviation of the different parameters over the 13 specimens. Student tests were used to evaluate differences between anterior and posterior, between central and peripheral, and between proximal, medial and distal areas of the vertebral body. Results showed that overall, distal slice is significantly different from medial and proximal slice. Also, it was shown that central versus peripheral subdivision of the vertebral body is more representative compared to anterior versus posterior subdivision. Fundamental knowledge acquired through this study will allow characterizing finite element models (FEM) such as the “Spine Model for Safety and Surgery” (SM2S) that was used to reach the first objective. A segment of three vertebrae and intervertebral connective tissues was extracted from SM2S and subjected to 51 different dynamic loading conditions divided into four categories: compression, shear, distraction and torsion. Fracture initiation and propagation were analyzed, and time and energy at fracture initiation were computed. To each fracture pattern described in the clinical literature was associated one or several of the simulated fracture patterns and corresponding loading conditions. When compared to each other, torsion resulted in low-energy fractures, compression and shear resulted in medium energy fractures, and distraction resulted in highenergy fractures. Increased loading velocity resulted in higher-energy fracture for similar loadings. The use of a finite element model provided quantitative characterization of fracture patterns occurrence complementary to clinical and experimental studies, allowing to fully understand spinal fracture biomechanics.
Reaching the second objective, that is to characterize the spinal cord contusion biomechanics, required to achieve four specific research objectives. First, it was necessary to provide morphological characteristics for the complete healthy human spinal cord. This allowed proposing “invariant” parameters that can be used as normative data for spinal cord injury studies. Absolute metrics of SC (transverse and antero-posterior diameters, anterior and posterior horns width, cross-sectional SC area and white matter percentage) were measured using semi-automatic segmentation of high resolution in vivo T2*-weighted transverse images acquired at 3T, at each SC level, on healthy young (N=15) and older (N=8) volunteers. Robustness of measurements, effects of subject, age, and gender, as well as comparison to previously published post mortem data were investigated using statistical analyses (Analysis of variance, Tuckey-HSD, Bland-Altman). Normalized-to-C3 parameters were evaluated as invariants using a leave-one-out analysis. Spinal canal parameters were measured and occupation ratio border values were determined. This work allowed providing observations that will benefit to biomechanical and clinical studies of spinal cord pathologies. In order to use a numerical model to study medullar contusion, the previously described morphological properties had to be completed by the characterization of mechanical properties for the spinal cord under dynamic loading adapted to traumatisms study. The next study thus aimed at defining dynamic mechanical properties for porcine spinal cord samples under transverse compression. Single compression and DMA tests at different strain rates and frequencies modalities were run on a total of 252 spinal cord segments. Stress strain curves and DMA parameters were compared between the different spinal levels and loading modalities. Samples showed a nonlinear viscoelastic behavior and damage occurred between 60 and 80% strain depending on the strain rate (significantly higher at lower strain rate). Increase of stress values with strain rates was not linear. This was explained by DMA results, which showed that spinal cord’s viscosity (and thus sensitivity to strain rate) was not constant over the tested frequencies. The results allow understanding the mechanical behavior of the spinal cord, and gives data which can be used in finite element models. Morphological and mechanical properties that were defined in this part of the thesis can be used to create FEMs. Thus, a FEM of the spinal cord and surrounding anatomical structures (pia mater, dura mater, nerve roots, and cerebrospinal fluid) was developed and used to understand medullar contusion biomechanics. A thoracic and a lumbar segment of the spinal cord were used. Segments were impacted at 4.5m.s-1 to reproduce spinal contusion following thoracolumbar burst fracture. Compression percent, Von Mises stress, and injury kinematics were observed and compared to reference experimental data. The results were similar to data from the literature. The spinal cord underwent 7% more compression at the thoracic level compared to the lumbar level, due to differences in the white and gray matter geometries. Cerebrospinal fluid presence lowered the compression percent by up to 14%. The highest stress levels were located in the gray matter anterior horns, and were consistent with central cord syndrome descriptions. This study confirmed the protective role of cerebrospinal fluid on medullar lesions, and the need for early decompression following spine trauma. Finally, a similar model was created to reproduce spinal cord contusions that were experimentally performed on mice specimens. Comparison of the simulation results with MRI and histology follow-up of each specimen allowed establishing a mechanical injury criterion for the spinal cord. This last study is preliminary and will allow, after some developments, creating a link between vertebral fracture and spinal cord injury.
Results generated in the Ph.D. first allow confirming the existence of a link between mechanical loading conditions during vertebral traumas and fracture patterns. However, it appeared in some cases that different loading conditions can lead to similar fracture patterns. Thus, some fractures observed in clinic must be analyzed with care, as their history and the potential associated spinal cord injuries can be of different natures. In the same way, the contribution of cerebrospinal fluid in spinal cord protection, and influence of vertebral level on injury mechanisms were quantified for medullar contusion. The interpretation of these results supports the choice of an early decompression during surgical management of a spinal cord injury. Finally, the results obtained through this thesis allowed giving recommendations regarding protection devices development and surgical techniques. Moreover, fundamental knowledge was acquired, that will allow enhancing tools used for spinal injuries biomechanics understanding, such as finite element models.
| Date | 3 Feb 2014 |
|---|
| Original language | French |
|---|
| Awarding Institution | - École de technologie supérieure
|
|---|
| Supervisor | Yvan Petit (Supervisor) & Pierre Jean Arnoux (Co-supervisor) |
|---|