Mechanical testing setups affect spine segment fracture outcomes

Abstract

The purpose of the work presented here was to establish an experimental testing configuration that would generate a bending compression fracture in a laboratory setting. To this end, we designed and fabricated a fixture to accommodate a three level spine segment and to be able to perform mechanical testing by applying an off-centric compressive loading to create a flexion-type motion. Forces and moments occurring during testing were measured with a six-channel load cell. The initial testing configuration (Fixture A) included plates connected to the superior potted vertebral body and to the ball-socket joint of the testing system ram. Surprisingly, while all cadaveric specimens underwent a similar off-centric compressive loading, most of the specimens showed extension outcomes as opposed to the intended pure-flexion motion. The extension was due to fixture size and weight; by applying an off-centric load directly on the top plate, unintended large shear forces were generated. To resolve the issue, several modifications were made to the original fixture configuration. These modifications included the removal of the superior plates and the implementation of wedges at the superior surface of the fixture (Fixture B). A synthetic sample was used during this modification phase to minimize the number of cadaveric specimens while optimizing the process. The best outcomes were consistently observed when a 15°-wedge was used to provide flexion-type loading. Cadaveric specimens were then experimentally tested to fracture using the modified testing configuration (Fixture B). A comparison between both fixtures, A and B, revealed that almost all biomechanical parameters, including force, moment, and displacement data, were affected by the testing setup. These results suggest that fixture design and implementation for testing is of extreme importance, and can influence the fracture properties and affect the intended motion.

1. Introduction

Biomechanical testing of the spine has been implemented for many decades to help understand spine biomechanics such as the risk of vertebral fractures in diseases such as osteoporosis and metastatic cancer (Adams and Dolan, 2005; Denis, 1983; Haher et al., 1991). Mechanical testing is a valuable tool that helps investigate and address the biomechanical response of this structure (Adams, 1995; Przybyla et al., 2007; Windhagen et al., 1997). Experimental testing has also been widely used to validate and evaluate the accuracy of computational methods such as quantitative computed tomography-based finite element analysis for predicting bone fracture properties (Buckley et al., 2007; Giambini et al., 2016b; Matsuura et al., 2014; Zeinali et al., 2010). The human spine is a complex structure composed of bony structures, intervertebral discs (IVDs), cartilage, ligaments, and associated muscles, enabling a large range of motion and load bearing capacity necessary for activities of daily living. This complexity of the tissue makes it difficult to obtain reliable and repeatable fracture data of vertebrae in a laboratory setting.

Mechanical testing has been performed on single vertebral bodies by various researchers to evaluate bone and fracture properties (Buckley et al., 2007; Fang et al., 2014; Giambini et al., 2016a; Gustafson et al., 2017; McGowan et al., 1993; Silva et al., 1993; Teoh and Chui, 2008). However, the outcomes from these studies omit important tissues of the spine associated with the characteristics and prevention of fractures. Although custom-made fixtures may be necessary, being able to accurately test a spine segment would provide more realistic and complete outcomes. Several studies have analyzed spine motion segments, incorporating two or three vertebrae, to measure fracture properties of intact or simulated-defect cadaveric spines (Ebihara et al., 2004; Groenen et al., 2018; Lu et al., 2014; Newell et al., 2017; Whealan et al., 2000a; Whyne et al., 2003). Spine segments contain various vertebral bodies and IVDs, which play important roles in transferring loads to the neighboring vertebrae. Addition of the elements, however, requires a larger fixture to accommodate a bigger spine section. These fixtures may negatively influence the outcome of the experimental data. Unfortunately, when evaluating and investigating fracture properties of spine segments, the majority of previous studies have reported only compressive forces and ignored the effect of other reaction forces and moments (Dimar et al., 1998; Ebbesen et al., 1999; Ebihara et al., 2004; Mirzaei et al., 2009; Stemper et al., 2015). Therefore, it is difficult to note from previous literature and reports whether other forces and moments have had a significant impact on the experimental outcomes or whether the mechanical testing setup has been able to generate intended loading and motion scenarios.

There exist a few studies on experimental testing that reported both forces and moments acting on the spine during fracture (Alkalay et al., 2008; Buckley et al., 2007). Buckley et al. (2007) performed bending tests on single cadaveric vertebral bodies using a multiaxial load cell and reported the compressive force and the forward bending moment. They showed, in some specimens, that the pattern of recorded bending moments was different from the compressive forces. In a more recent study, Alkalay et al. (2018) intended to apply pure compression on several simulated defect spinal segments. However, their results showed significant bending moments at the onset of fracture, suggesting the unintended presence of a combined bending/compression loading, not pure compression.

The purpose of the work presented here was to establish an experimental testing configuration that would generate a bending compression fracture in a laboratory setting. This testing configuration was being developed as part of a larger study to characterize fracture properties for vertebra, both intact and with simulated lesions. Acquiring relevant and accurate loading data from experiments is crucial for evaluating spine fracture and validating computational models used in fracture prediction. To accomplish this, we designed an instrumented fixture able to accommodate a spine segment, with displacement boundary conditions intended to mimic a variety of activities of daily living. Preliminary testing of our initial fixture design (Fixture A) indicated that the moments acting on the spine segments tested were not the intended physiological loading conditions. Based on this information, we redesigned the test fixture in order to achieve the desired loading conditions (Fixture B). The present study compares fracture test results from Fixtures A and B, and demonstrates that Fixture B produces the desired physiological loading conditions.

2. Materials and methods

2.1. Sample preparation

Cadaveric specimens:

Eight fresh-frozen cadaveric spines (3 females, 5 males) with an average age of 75.6 years were obtained from the Anatomy Laboratory at Mayo Clinic, Rochester, MN, following approval from the Bio-specimens Sub-committee of the Institutional Review Board. Spines were dissected from the torso and spine segments were extracted from each intact spine. Each spine segment was comprised of three vertebrae, posterior elements, two intervertebral discs (IVDs), and all ligaments. From each cadaveric spine, two to four segments were obtained with their middle vertebrae being T6, T9, T12, or L3. In one cadaveric spine, the middle vertebrae consisted of T5, T8, T11, and L2 due to deformity of the spine. In total, 26 3-level spine segments were obtained from the 8 cadavers. Specimens were categorized into two groups: intact (n=14) and simulated defect segments (n=12). In the simulated defect group, holes were created inside the vertebral bodies of the middle vertebrae by drilling anteriorly using diamond drill bits to mimic lytic lesions. The size of the bits depended on the vertebral level and consisted of 8, 10, 12, and 14 mm in diameter for T6, T9, T12, and L3, respectively. The specimens were stored at −20°C and thawed at room temperature before testing procedures.

Synthetic sample:

During the fixture-modification phase of the study following the initial cadaveric testing, we developed and tested one synthetic sample. This reusable synthetic sample allowed for multiple tests to be performed in order to evaluate the fixture configuration, while also minimizing the number of cadaveric specimens used. The overall size of the sample was similar to an L2-L4 spine segment. The sample was made in the Division of Engineering at Mayo Clinic using TangoPlus, which is a proprietary photopolymer material used with an Objet Connex 350 3D printer (Stratasys, Ltd., Eden Prairie, MN, USA). This material is rubber-like and does not fail; therefore, it can be used multiple times. We tested this sample after each modification and compared the results. The purpose was only to compare the results while we were modifying the testing configuration and to learn why applying a flexion type loading could lead to extension results. Therefore, we did not need a validated specimen for this specific task. The use of this sample ensured that several potential confounding variables such as potting locations, spine segment geometry, and spinal levels could not affect the outcomes of the tests. For this sample, we recorded force, moment, and displacement data, which helped modify the fixture for desirable outcomes.

All cadaveric specimens and the synthetic sample were then potted similarly in poly methyl methacrylate (PMMA) (Figs. 1A and 1B) using a custom-built potting fixture. In the cadaveric specimens, the superior and inferior vertebrae were almost entirely potted up-to the IVDs to ensure fracture of the middle vertebral body during testing. Two or three acrylic rods were also used during the potting process of the cadaveric specimens to keep the construct rigid and protect the specimens from any mishandling during preparation. These rods were removed at the onset of mechanical testing.

Fig. 1.

(A) A cadaveric spine segment specimen; (B) a synthetic spine segment sample potted in PMMA; (C) Fixture A testing setup; (D) Fixture B testing setup; (E) and (F) close-ups of the top parts from Fixtures A and B showing the differences between the two designs

Each cadaveric spine segment was imaged using quantitative computed tomography (QCT) in air at room temperature using a custom built scanning fixture. A Siemens Somatom Definition Scanner (Siemens Healthcare GmbH, Germany) was used for imaging acquisition. Images were obtained at 120 kVP, 220 mAs, and reconstructed using a sharp algorithm U70 with 0.6 mm isotropic voxels.

2.2. Mechanical testing

Cadaveric spine segments and the synthetic sample were mechanically tested using an MTS 858 Mini Bionix II testing machine (MTS Systems Corporation, Eden Prairie, MN). A custom-made instrumented fixture was designed to hold the spine segments during mechanical testing and to facilitate different loading conditions (Fig. 1C and 1D). The bottom part of the fixture included a 6-channel load cell (Interface Inc., Scottsdale, AZ) mounted on top of an x-y linear bearing stage. The x-y stage allowed the specimens to be accurately placed so that an off-centric loading could be applied, creating a flexion/extension-type loading condition. In order to mimic a flexion-type loading scenario, the anterior-most boundary of the middle vertebra was aligned with the axis of the ram by adjusting the x-y stage. The distances required to move the spine segment posteriorly so that the ram would line-up with the anterior cortex were obtained via the QCT scan images. These images showed the coordinates of the center of the vertebrae and the anterior cortex, thus, providing the metrics needed for the offset loading configuration. Once the spine segment was off-centered, the lower end of the fixture was restrained from movement on all axes.

The top part of the fixture was attached to the crosshead, designed based on the kinematics and biomechanics of the spine to allow the specimens to be compressed unconstrained. An external linear potentiometer was used to record the displacement of the crosshead during testing (Fig. 1). The top portion of the fixture consisted of two different designs:

Fixture A:

This fixture is comprised of two circular plates with slots to allow for anterior and posterior translation of the specimens (Fig. 1E). These circular plates were attached to the top plate holding the superior potted vertebral body. The entire top fixture was attached to the top ram of the MTS. In order to allow for unconstrained motion and loading, a ball was attached to the ram of the MTS machine and placed on a hemispherical hole to form a ball-and-socket joint.

Fixture B:

The two circular rocking plates and their attachments from Fixture A were removed and the ball attached to the ram of the MTS was placed directly on the top plate holding the superior potted vertebral body (Fig. 1F). In order to evaluate loading configurations, several wedge blocks of various slopes (5, 10, 15, 25 and 40 degrees) were also placed separately on top of the fixture to create a flexion-type loading condition. The synthetic sample was used in order to determine the optimal wedge that would create flexion loading, while minimizing other reaction forces and moments.

All specimens were compressed at a rate of 0.083 mm/sec to reach a maximum displacement of 10 mm. In Fixture B, we noticed that the displacement values at the onset of fracture were considerably larger than those from Fixture A. Therefore, we increased the maximum displacement to 14mm in Fixture B for the remaining specimens. Force, moment, and z-displacement data were collected at 100 Hz.

Based on our coordinate system, the force F_z corresponded to the compressive force, while F_x and F_y, corresponded to the shear forces generated by the testing configurations. The moment in the x-axis (M_x) represented flexion if positive, and extension if negative. The moments in the y- and z-axes represented lateral (M_y) and torsional (M_z) moments, respectively. The collected moment data were in the load cell’s coordinate system (CS), and did not represent the moments experienced by middle vertebrae in the spine segments. Therefore, a new CS was defined at the center of the middle vertebra of each segment and the moments were transferred to the new CS. This transformation only affected the moments with no change in the force data. All moment data from cadaveric specimens in the current study are reported based on the CS of the middle vertebrae. Table 1 shows the number of cadaveric specimens used in both fixtures.

Table 1.

The number of samples used in both testing configurations

Testing configuration		Cadaveric spine segments n=26		Synthetic samples n=1
		Intact	Simulated defect
Fixture A		8	10	-
Fixture B	0º no wedge	1	1	1
	5º wedge	-	-	1
	10º wedge	-	-	1
	15º wedge	3	3	1
	20º wedge	-	-	1
	40º wedge	-	-	1
Total		12	14	1*

^*The same synthetic sample was implemented on Fixture B with various wedges after analyzing the data from Fixture A to modify and optimize the fixture and process for a flexion-type loading and fracture condition.

2.3. Statistical analysis

All statistical analyses were performed using JMP Pro version 14.0.1 (SAS Institute Inc., Cary NC). Linear regression analysis was performed with compressive force or stiffness values as dependent variables and the bending moment M_x as an independent variable. A T-test was performed to compare the outcomes from the two fixtures, and to compare intact and simulated-defect specimens. The level of significance was set to 0.05.

3. Results

3.1. Fixture A

Eighteen cadaveric samples were tested to fracture using Fixture A. The force and moment data of the specimens under an off-centric type loading were successfully recorded. In all cases, the specimens were off-centered so that the loading axis of the ram was aligned with the most anterior cortex of the vertebrae and a flexion-type loading condition was expected.

Three patterns of forces and moments were observed during data analyses. In the first pattern, a combination of flexion moment (+M_x) and compression force (F_z) was observed. This loading response was expected as the loading configuration was intended to enforce a flexion-type loading scenario. The maximum values for the compressive force and the bending moment were reached at similar times, and both curves showed similar waveform characteristics (Fig. 2A). However, a M_y moment was also present, indicating an unwanted lateral bending. The three moments were given in the CS of the middle vertebrae. The second observed pattern showed the measured flexion moments at the time of fracture to be very small or negligible. This means that fracture resulted mostly due to a pure compression load as all other moments in the y- and z-axis were also negligible (Fig. 2B). In the third pattern, fracture occurred under a combination of compression, extension (-M_x), and lateral bending (M_y). A large number of specimens presented this pattern during testing, with the outcomes showing dissimilar curves for F_z and M_x. Results from a characteristic cadaveric specimen are shown in Fig. 2C. It can be observed that from the beginning of the test, the specimen and loading configuration induced an extension scenario.

Fig. 2.

Force and moment data from three cadaveric spine segments using Fixture A; vertical dashed lines show fracture points. (A) Specimen #1 fractured under expected flexion/compression loading; (B) specimen #2 experienced a small amount of moments at fracture; and (C) specimen #3 experienced an unexpected extension moment

Figure 3 shows the forces and moments for all specimens at the onset of fracture. It should be noted that the forces F_x and F_y and the three moments in this figure were not the maximum values experienced during testing; rather, the values shown here were recorded at the maximum F_z (representing the fracture load). This figure shows the variation of the force and moment data in the entire cohort using Fixture A. For intact and defected specimens, on average, the maximum F_z values were −2833.7 N [±1851.0], and −1582.8 N [±656.8], respectively. This difference was found to be insignificant (p=0.07).

Fig. 3.

Forces and moments recorded at maximum compression force (F_z) from all the specimens; the solid and hollow circles represent intact and simulated defect specimens, respectively.

Figure 4 describes the measured compressive forces (Fig. 4A) and stiffness (Fig. 4B), and the associated bending moments (±M_x) at the time of fracture. This figure illustrates that those specimens undergoing extension or a small amount of flexion at the time of fracture experienced significantly smaller compressive fracture forces than those specimens undergoing larger flexion moments (+M_x) (p<0.0001) (Fig. 4A). A similar trend was observed for the stiffness outcomes (Fig. 4B), where the specimens being fractured under extension or under a small amount of flexion, showed significantly smaller stiffness values than those being fractured under substantial flexion (p<0.0001). For intact and defected specimens, on average, the maximum stiffness values were 1885.7 N/m [±1404.2], and 827.0 N/m [±261.1], respectively, and this difference was found to be significant (p=0.0423).

Fig. 4.

(A) Compressive force and (B) stiffness values versus bending moment M_x

Due to the inconsistency in these data, we calculated the bending moment M_x-calc, based on a static equilibrium state, using the available force data from the load cell. We then compared the calculated bending moment with the bending moment measured directly by the load cell. We noticed that the two forces that contributed to M_x-calc were F_y and F_z, acting through their moment arms r_z and r_y, respectively (Fig. 5). The force F_y, with its moment arm r_z, resulted in a bending moment (M_x-y=r_z×F_y), and the force F_z, with its moment arm r_y, created a second component of a bending moment (M_x-z=r_y×F_z). The total calculated moment M_x-calc was equal to M_x-z + M_x-y (Fig. 5A). Both bending moments measured by the load cell and calculated using the force data are shown in Fig. 5B. It is important to note that the moment arms r_z and r_y changed during deformation of the specimen and we could not calculate these during mechanical testing. Therefore, the initial values of the moment arms were used to estimate M_x-z and M_x-y. This limitation of our calculations can be observed between the dash and solid lines as the displacement gets larger (Fig. 5B). In most of the specimens tested using this fixture, M_x and M_x-calc showed similar waveforms, illustrating that both forces, F_y and F_z, were equally important for the observed bending moments.

Fig. 5.

(A) Schematic diagram describing the forces leading to the flexion/extension bending moments; (B) mathematically-calculated (dashed line) and load cell-measured (solid line) bending moment

In this fixture configuration, the top part of the fixture, which included the circular plates used for translation of the specimen, created a major issue. Due to its weight (~150N) and size, when the spine segment was translated posteriorly to induce a flexion-type loading, these plates tilted anteriorly at the onset of testing, thus increasing the moment arm r_z and inducing additional and unwanted extension. We also calculated the ratio of F_y/F_z at the onset of fracture. This ratio ranged from 0.11 to 0.27, with an average value of 0.2. This indicates that in Fixture A, F_y was about 20% of F_z. The simulated defect specimens (n=8) with an average defect of 17.2% [±6.13] fractured at compressive force of −1544.1 N [±703.94], while intact specimens (n=10) fractured at 2833.7 N [±1851.0]. The difference between fracture force from the two groups was insignificant (p=0.06). To minimize the effect of F_y and r_z, both of which contributed to an extension moment, and to be able to induce a flexion-type only loading configuration, we modified the fixture to create Fixture B.

3.2. Fixture B

In order to apply a pure flexion/bending loading to the spine segment, we modified the fixture to reduce the moment arm r_z and the initial force F_y. To minimize the number of cadaveric specimens to be used during the evaluation, we first tested the synthetic sample (Fig. 1B) using different configurations from Fixture B: no wedge, directly placing the ball on the plate holding the superior and potted vertebra; and five wedges of different slopes (from 5° to 40°) attached separately atop the plate holding the superior PMMA block, and directly placing the ball on the wedge. Experimentally measured outcomes from these testing configurations for no wedge, a 10° wedge, and a 15° wedge are shown in Fig. 6. All tests were performed on a single synthetic sample. It can be observed how the measured compressive force F_z reduced as the angle of the wedge was increased (from no wedge to 15°-wedge). More importantly, the force F_y, which played a major role in generating undesirable moments, decreased considerably with an increasing wedge angle. When no wedge was implemented, M_x started as positive, representing a flexion scenario, for the first few millimeters of displacement and then became negative. This negative behavior is representative of an extension-type motion (Fig. 6A). With a 10° wedge (Fig. 6B), more flexion was observed for the first 6 mm of displacement, but a subsequent downward M_x was observed. Finally, with a 15° wedge, the moment M_x increased linearly (flexion) with displacement for the entire displacement period (Fig. 6C) with no observed additional measured force F_y. A further increase in the slope of the wedge did not change the slope of M_x, yet increased unintended shear forces (data not shown).

Fig. 6.

The force and moment data obtained from mechanical testing of a synthetic sample using Fixture B; (A) no wedge with load directly applied on a horizontal surface; (B) using a 10° wedge; and (C) using a 15° wedge attached to the horizontal surface

Once a flexion-type loading scenario was established by obtaining M_x, F_z, and minimizing F_y and all other unwanted moments, two cadaveric spine segments were tested using Fixture B with no wedge to ensure outcomes similar to those obtained using a synthetic sample. The two specimens did show a similar pattern to the synthetic material. Figure 7A shows the results from one specimen with no wedge. In this figure, the bending moment M_x was initially positive (flexion) up to almost 2mm, then became negative (extension) (similar to Fig. 6A). Because the specimens fractured, the waveforms of the compressive force F_z and bending moment M_x were dissimilar. The remaining spine segment specimens (n=6) were tested using a 15° wedge. As expected, the measured bending moments were all in flexion, with all other forces and moments being significantly smaller. Figure 7B shows the results from a representative sample tested using a 15° wedge, with the maximum compressive force F_z and maximum flexion moment M_x occurring at similar displacements.

Fig. 7.

Data from two cadaveric specimens fractured using Fixture B; (A) when no wedge was used; (B) when a 15° wedge was used.

The forces and moments for the spine specimens (n=6) tested using Fixture B with a 15° wedge were recorded at the time of fracture, when the maximum compressive force F_z occurred (Fig. 8). All bending moments M_x were positive, showing a flexion-type loading configuration at the time of fracture. Additionally, the ratio of F_y/F_z in Fixture B ranged from 0.03 to 0.11 with an average of 0.08, meaning that F_y was, on average, 8% of F_z in this fixture. This ratio was significantly smaller than the value of 20% from Fixture A (p<0.0001). The defect specimens with defect size of 15.78% [±2.58] fractured at compressive force of 2608.1 N [±194.1], while the intact specimens were fractured at 3361.76 N [±552.9]. This difference in force was found to be insignificant (p=0.13). Also the maximum stiffness values were 994.1 N/m [±402.0] for intact and 754.1 N/m [±219.3] for defected specimens, and this difference was found to be insignificant (p=0.41).

Fig. 8.

Force and moment data recorded at maximum F_z for all the specimens fractured using Fixture B and a 15° wedge

4. Discussion

The purpose of this study was to establish an experimental testing fixture to generate a combination of bending and compression loading on ligamentous spine segments for fracture analysis. To this end, we designed and fabricated a fixture to accommodate a three level spine segment (three vertebrae and two IVDs), and to be able to perform mechanical testing by applying an off-centric compressive loading and creating a flexion motion. The initial testing configuration consisted of a fixture (Fixture A) that included plates that connected the superior potted vertebral body to the ball-socket joint of the MTS ram. These plates allowed for a posterior translation of the specimen so that the ram could be positioned anteriorly and a flexion motion could be induced. Using this fixture, when all test specimens underwent a similar off-centric compressive loading, the majority of the specimens showed extension instead of flexion, as measured by the 6-channel load cell. To resolve this issue and only produce the desired flexion motion and outcomes, several modifications were made to the fixture. These modifications included the removal of the superior plates and the implementation of wedges at the superior surface of the fixture. A single synthetic sample was used during this modification phase to minimize the number of cadaveric specimens while optimizing the process. Finally, cadaveric specimens were tested using the above mentioned modifications (Fixture B), obtaining the desired flexion motion of the spine and the flexion moments. It is interesting to note that almost all measured biomechanical parameters, including force, moment, and displacement data, were affected by the testing setup, suggesting that test fixture design and implementation is of extreme importance because it can influence measured fracture properties. Additionally, is important to state that the implementation of a 6-channel load cell allowed us to dissociate these effects and minimize the unwanted variables for a successful fracture process.

Fixture B, our optimized setup, used a 15° wedge to naturally apply a flexion-type bending moment, which reduced unintended shear forces applied to the spine. The forces and moments measured with Fixture B are representative of the forces and moments expected with a flexion only type loading configuration in an ex-vivo spine. The use of the wedge in this fixture also ensured fracture of the anterior vertebral cortical region, named a wedge compression fracture, which is the most common type of spine fracture (Whyne, 2014). In contrast, we noticed the large size and complexity of the initial fixture (Fixture A), which would eventually be found to play a substantial role in generating erroneous data when a flexion-only type loading configuration was desired. Fixture A also imposed over-constrained boundary conditions to the specimens due to the top part of the specimen not being able to freely move in the X and Y directions, changing the mechanics of spine fracture. These characteristics generated a considerable amount of shear forces to the spine, leading to an extension motion instead of flexion. A previous study has also reported similar inconsistencies (Alkalay et al., 2008). This research group experimentally tested single cadaveric vertebrae in a combined compression-flexion loading, and reported all six forces and moments. While the authors stated that the specimens were fractured under flexion, even though the fixture/specimen construct displaced as flexion, the measured outcomes by the load cell reported showed extension rather than flexion outcomes (Alkalay et al. (2008); Fig. 3). These results were similar to the ones observed by our group in the current study when using the original fixture (Fixture A). These observations highlight the importance of measuring and analyzing all the forces and moments during fracture testing to ensure desirable outcomes. Unfortunately, we noted that the majority of previous studies only reported fracture loads (F_z in our study), which is not sufficient to understand the mechanical response of the spine construct during loading.

One important application of mechanical testing on spine specimens is the prediction of fracture in metastatically involved vertebrae (Windhagen et al., 2000). Several prior studies have compared simulated defect samples with intact or augmented specimens for fracture properties (Matsuura et al., 2014; Whealan et al., 2000b). In the current study, we also tested intact segments and those with a simulated defect using both fixtures. As expected, in both fixtures, intact specimens were considerably stronger than those with a defect. Knowing that our original fixture (Fixture A) created inconsistent and erroneous outcomes, these results (intact vs defects) from Fixture A are not acceptable, as the difference in outcomes between both groups might be completely different if the correct fixture (Fixture B) was used. This leads to an important question: is it possible to compare mechanical testing results from prior studies, acknowledging that the majority of the mechanical testing data in the literature have been obtained using custom-made fixtures, each with a different design, size, shape, and where most have either not implemented a 6-channel load cell, or have not reported forces and moments in other directions? The results of the current study emphasize the variables affecting the experimental outcomes and a possible need to standardize mechanical testing setups so that reliable outcomes can be obtained, independent of the fixture configuration.

In our elderly cadaveric cohort, several spine segments presented with disk degeneration and osteophytes, which could profoundly impact the kinematics and biomechanics of the spine (Ferguson and Steffen, 2003). Including IVDs and ligaments in the spine segments for mechanical testing allowed for transferring of loads to the middle vertebral bodies in a more physiological manner.

The current study has some limitations. First, the calculated moments at the CS of the middle vertebra may not be very accurate because this calculation depended on the distance between the two CSs (the load cell CS and the middle vertebra CS). This distance continually changes during deformation, and it is not possible to track its location during testing. Second, only six specimens were experimentally tested using Fixture B. Although the number is not large, due to some specimens being used during fixture evaluation (Fixture A), the measured data showed consistent and desired outcomes. Additionally, we mixed the data from the thoracic and lumbar regions because of the relatively small sample size, even though the biomechanics of the thoracic spine is different from the lumbar region. Finally, using a follower load could have provided a more natural boundary condition, but this would have required attachments to the vertebral bodies possibly affecting the fracture properties by weakening the cortical shell.

In conclusion, the testing fixture configuration can produce undesired loading effects and generate data that is not related to the intended loading condition. The implemented fixture can also affect fracture force, toughness, and stiffness outcomes of the spine. Our study suggests that in order to generate a flexion motion of the spine, while inducing a flexion-only type fracture of the middle vertebra, an unconstrained top plate needs to be implemented with an appropriate wedge. Also all the forces and moments should be recorded and analyzed for desirable outcomes.