Loading rate effect on tradeoff of fractures from pelvis to lumbar spine under axial impact loading

Abstract Objectives: The transmission of impact loading from the seat-to-pelvis-to-lumbar spine in a seated occupant in automotive and military events is a mechanism for fractures to these body regions. While postmortem human subject (PMHS) studies have replicated fractures to the pelvis or lumbar spine using isolated/component models, the role of the time factor that manifests as a loading rate issue on injuries has not been fully investigated in literature. The objective of this study was to explore the hypothesis that short duration pulses fracture the pelvis while longer pulses fracture the spine, and intermediate pulses involve both components. Methods: Unembalmed PMHS thoracolumbar spine-pelvis specimens were fixed at the superior end, and a six-axis load cell was attached. The specimens were mounted on a vertical accelerator, and noninjury and injury tests were conducted by applying short, medium, or long pulses with 5, 15, or 35 ms durations, respectively. Peak axial, shear and resultant forces were obtained. Injuries were documented using posttest x-ray and computed tomography images and scaled using the AIS (2015). Results: The mean age, stature, weight, body mass index, and BMD of twelve specimens were 64.8 ± 11.4 years, 1.8 ± 0.01 m, 83 ± 13 kg, 26.7 ± 5.0 kg/m2, and 114.5 ± 21.3 mg/cc, respectively. For the short, long, and medium duration pulses, the mean resultant forces were 5.6 ± 0.9 kN, 5.9 ± 0.94 kN, and 5.4 ± 1.8 kN, and time durations were 4.8 ± 0.5 ms, 16.3 ± 7.3 ms, and 34.5 ± 7.5 ms, respectively. For the short pulse, pelvis injuries were more severe in 3 out 4 specimens, for the medium pulse, they were distributed between the pelvis and spine, and for the long pulse, spine injuries were more severe in 3 out of 4 specimens. Conclusions: While acknowledging the limitations of the sample size, the results of this study support the hypothesis of the time variable in the tradeoff between pelvis and spine injuries with pulse duration. The tradeoff pattern is attributed to mass recruitment: short pulse biases injuries to pelvis while limiting spinal injuries, and the opposite is true for the longer pulse, thus supporting the hypothesis. It is important to account for the time variable in injury analysis.


Introduction
Analysis of automotive field data has shown an increasing trend in lumbar spine fractures to restrained occupants in frontal impacts (Pintar et al. 2012;Adolph et al. 2013). Analysis of military vehicle impact events has revealed similar findings (Salzar et al. 2006;Ramasamy et al. 2008;Adams 2009;Ramasamy et al. 2011;Danelson et al. 2018;Loftis et al. 2019). In both cases, vertical loading from the seat-to-pelvis-to-spine is attributed to be a load path for fracture to either or both regions (Helgeson et al. 2011;Kang et al. 2012;Patzkowski et al. 2012;Cross et al. 2014;Danelson et al. 2015;Yoganandan et al. 2021). Injury prevention strategies should include the determination of the biomechanics of the pelvis and lumbar spinal column.
Previous isolated pelvis loading studies revealed differing injury patterns with different initial alignments (Salzar et al. 2021). Similar studies with the lumbosacral spinal column revealed compression-related fractures at different vertebral levels and with varying severities (Yoganandan et al. 2018;2020;Ortiz-Paparoni et al. 2021). As a composite system, injury to one body region or component affects the transmissibility of loads to the adjacent body region(s) as the human musculoskeletal components absorb and transmit the energy, and under vertical loading the transmissibility occurs from the pelvis to the sacrallumbar spine. In other words, injuries sustained by the pelvis have implications in the energy transfer to the sacral-lumbar spinal column. As fracture represents the state of, or implies, energy absorption, structural components cranial to the injured or fractured component sustain an altered state of mechanical loading, i.e., often less energy to be absorbed.
Previous whole-body PMHS inferior-to-superior impact loading sled experiments revealed the effect of acceleration pulse on the migration of injuries to the pelvis to the lumbar spine complex (Yoganandan et al. 2014). Specifically, intact PMHS tests were conducted by applying inferior-tosuperior accelerations of different pulse shapes to the pelvis of a supine occupant. Although not statistical due to the limited sample size, longer pulses resulted in spine injuries while shorter pulses resulted in pelvis fractures. This was termed as mass recruitment effect on injury migration, in this case it was directed cranially. Changes in the rate of loading, reflected by the pulse shape, changed the injury pattern from the distal to the proximal regions of the specimen. This aspect cannot be examined with isolated pelvis or lumbar spine experiments due to the lack of the spine or pelvic regions in these models. The issue of injury migration or tradeoff between the two body regions, pelvis and or lumbar spine, were the subject matter for this research. Specifically, the objective of this study was to examine the tradeoff between pelvis and lumbar-sacral spine (hereafter termed as spine or lumbar spine) injuries with vertical loading using a combined pelvis-lumbar spine experimental model. The key questions posed in the study were: does a longer pulse induce spine, does a shorter pulse induce pelvis, and does a pulse in between the two durations induce both pelvis and spine injuries?

Specimen preparation and mounting
The protocol for the study was approved by the local institutional review boards and sponsor. The experimental design consisted of isolating the full lumbar-pelvis subsystem from T11 to the acetabulum with musculature and abdominal skin. PMHS were screened to ensure no presence of bloodborne pathogens or spinal or pelvic abnormalities that may affect biomechanical injury outcomes. Pretest x-rays and computed tomography scans were obtained. The inclusionexclusion criteria were such that all subjects were male, no surgical intervention to the lumbo-pelvis, no osteoporosis as determined by the quantitated computed tomography bone mineral density of the lumbar spine that followed the clinical guidelines (threshold 80 mg/cc), and no bridging osteophytes or congenital fusion of the spine-pelvis complex. While preparing the specimen, the contents of the pelvic bowl were removed; however, the ligamentous structures along the lumbar spinal column and within the pelvic ring were maintained intact. A cylindrical 11.5 kg ballistic gel form was contoured to fill the lower and upper abdominal cavities and surround the lumbar spine. The pelvic angulation in the sagittal plane was targeted at a mean of 40 ± 5 degrees. This orientation was defined as the angle of the pelvis plane with respect to the vertical axis. The pelvis plane was defined as the lines connecting the bilateral anterior superior iliac spines with the pubic symphysis. The specimen was embedded in polymethylmethacrylate at the superior level. A six-axis load cell (Denton 3300 J; Humanetics Innovative Solutions, Farmington Hills, MI) was attached to the superior end of the fixation that was embedded in the T12 vertebral body. The cranial end of the specimen was flexed approximately 12 ± 5 degrees. A 2.7 kg and a 8.8 kg mass were added to represent the mass of the pelvic and surrounding abdominal contents, and a mass of 12 kg was placed on the superior end of the preparation to account for the effective mass of the torso (Salzar et al. 2021). As stated, the pelvic mass was within the pelvic cavity, contouring the pelvic bowl from the rami inferiorly to the distal sacrum, and the abdominal mass was a cylindrical mold set atop the previous gel setting, surrounding the ilia and lumbar segments.

Vertical accelerator loading device
The custom vertical accelerator device was used (Yoganandan et al. 2015). Briefly, the device consisted of two components ( Figure 1). The reader is referred to the illustration shown in the citation. The impact component consisted of a stanchion fixed to the laboratory wall, a cart assembly, and a V-shaped lever arm attached to the seat platen. The drop-cart assembly allowed for vertical motion along the stanchion and weight adjustments. The cart mass was released with a predetermined height and impacted the lever arm, accelerating the specimen off the seat platen and up the free-standing impact-receiving component. This component allowed for positioning upon the seat platen and mounting of the superior end of the specimen to a cart and vertical track that constrained the preparation to postimpact vertical translation. The mounting included a six-axis load cell affixed to the specimen potting, which constrained the superior end of the preparation.

Data acquisition and analysis
Sensor data were acquired at a sampling rate of 100 kHz, force and acceleration signals were processed using a four pole Butterworth low pass filter at class 1000. The recorded forces were compensated, representing the transmitted loads to the specimen. From the axial and shear forces and resultant force time histories for each specimen, peak forces were obtained, and statistical analysis of the peak forces with pulse type was conducted using the t-test. The statistical significance was set at 0.05 level.
Test protocol, imaging, and injury scoring Pretest x-rays were obtained while the specimen was on the platform of the vertical accelerator to confirm the alignment. The experimental protocol included two tests, a non-injury test and an injury test. After the first non-injury test, the specimen was palpated, and x-rays were obtained to check the radiological integrity by comparing with pretest x-rays. The specimen was also palpated by the same clinician in all specimens. The injury test was conducted at a greater impact level with targeted short, medium, or long pulses (approximately 5 ms, 15 ms, and 35 ms, respectively), and x-rays and computed tomography scans were obtained to identify and score the severity of the injuries. They were scored using the Abbreviated Injury Scale, AIS 2015 version (AIS 2015) by a trained practicing clinician and the other was a certified AIS coder. Both observers were independent and blinded to the test sequence and the association of the specimen identity with the type of injury test: short, medium, and long duration exposures.

Results
The test matrix consisted of four specimens for each exposure type (see online supplement Table A1). The demographics of the 12 male specimens were: mean age, stature, total body mass, body mass index, and bone mineral density data were: 64.8 ± 11.4 years, 1.8 ± 0.01 m, 83 ± 13 kg, 26.7 ± 5.0 kg/ m 2 , and 114.5 ± 21.3 mg/cc. The specimen demographics and bone mineral density data were not significantly different (p < 0.05) for any combinations of pulses.
Representative force-time plots for the three pulses are shown ( Figure 2). The mean axial forces were 5.17 ± 0.91 kN, 4.86 ± 1.61 kN, and 4.55 ± 0.50 kN, respectively, for the short, long, and medium duration pulses. These data for the shear forces were 2.48 ± 0.76 kN, 2.87 ± 1.26 kN, and 3.82 ± 0.99 kN, respectively. The resultant forces were 5.57 ± 0.90 kN, 5.35 ± 1.77 kN, and 5.87 ± 0.94 kN, respectively, for the three pulses ( Figure 3). The axial, shear, and resultant forces were not significantly different (p < 0.05) in any combinations of pulses, except for the shear force in once pulse combination. Figure 4 shows the highest AIS levels for lumbar and or pelvis injuries on a specimen-by-specimen by basis for the short, medium, and long pulses. Table A2 (see online supplement) shows the details of injuries.

Discussion
As stated in the introduction, the objective of the study was to investigate the tradeoff in injuries between the pelvis and lumbar spine from vertical impact loading. It was based on the consideration that the resistance and load paths depend on the applied impact pulse to the human anatomical structural complexes in the pelvis and spine regions. It is well known that the responses of biological materials are time dependent, and regarding the dual regions of the pelvisspine complex, the temporal factor plays an important role, not possible to examine in isolated tests of spinal components such as intervertebral disk, facet joint, or sacroiliac joint segmented spine, intact pelvis, or intact lumbar spinal columns. The issue of mass recruitment effect is relatively minimal or does not exist in isolated subsystem or component tests. Hence, the combined pelvis-spinal column model was chosen in the present study.
The targeted time durations of the short, medium, and long pulses were approximately 5, 15 and 35 ms, respectively. The actual mean durations corresponding to these pulses were: 4.8 ± 0.5 ms, 16.3 ± 2.6 ms, and 34.5 ± 4.0 ms. The range represented a sevenfold change, delineating the responses and injuries of the pelvis-spine complex in a wide temporal domain. The shortest duration may be representative of a seat in a military vehicle without seat cushion (currently in vogue) while the longest duration pulse may be representative of a seat with an energy absorbing material (Danelson et al. 2015;Rupp et al. 2021). As the base platform upon which the specimens were placed on the vertical accelerator device did not deform, the present results   simulated scenarios wherein seat deformations are not expected in the real world. It should be noted, however, that automotive field data have identified residual seat deformations to be a factor that influences the axial loading from the seat of the vehicle to the pelvis-lumbar spine complex. From this perspective, actual deforming seats should be included in future experimental and or finite element modeling studies to gain an improved understanding of the pelvis-spine complex.
The PMHS selected for the three pulses did not significantly differ (p > 0.05) in demographics or bone mineral densities, although all were male specimens. This shows that the study was not biased about the assignment of a specimen to a specific pulse type. Although the fundamental constituents of the structure are the same between males and females, it would be appropriate to repeat these experiments with female specimens due to the following reasons (citations are limited due to the constraint by the journal). The anatomical shape of the female pelvis is different, and the orientation of the sacral-lumbar spine changes between males and females for the same angulation of the pelvisspine complex. Female lumbar spine and pelvis bone densities and strength are lower than the male, and this is true for both the pelvis and lumbar vertebrae, although the structural bony strength factor may not be as important in the military (population is younger, active, and healthier) compared to the civilian population. Female lumbar spinal columns tend to be shorter than male spines for comparable anthropometry, and their vertebral body and disk sizes are also shorter. In addition, the curvature of the spine affects the internal load paths: greater lordosis induces more bending moments, straight spines tend to maximize the axial force, and kyphotic spines tend to reverse the moment pattern, and these effects should be considered for an improved assessment of the tradeoff in injury patterns. Occupant-specific computational models may be used for this purpose.
The lack of statistical significance for the axial and resultant forces in addition to the randomized specimen selection discussed above, show that the biomechanical responses are also not biased to the pulse type. It should be noted that the axial force is the major or principal resisting load under this mode, and axial forces exceeded the off-axis shear force, considered as a secondary component. The lack of statistical significance in the offaxis force in two out of three pulse combinations reinforces the lack of bias in specimen selection used in the study. Additional samples should be tested in a future study to increase the robustness of these observations. A comparative evaluation of injuries based on AIS severities to the lumbar spine and pelvis appear to support the tradeoff concept in injury patterns, i.e., short pulses biases injuries to the pelvis and limit injuries to the spine to lower severities. For the short pulse, pelvis injury severities were greater in 3 out 4 cases/specimens, and in the other, both pelvis and lumbar spine components had the same severity score of AIS 2, i.e., 1 out of 4 cases. In addition, in one case (in the 3 out of 4 group), the spine fracture was limited to AIS 1, i.e., minor injury. For the long pulse, 1 out 4 specimens responded with lumbar spine-only injuries, the severity of lumbar spine injuries in 3 out of 4 specimens were greater than or equal to pelvis, and one specimen responded with a greater severity to pelvis than spine. This pattern is somewhat opposite to the short pulse and with the limited sample size recognition. Taken together the results from eight specimens support hypothesis that long pulses biases injuries to the spine and limit injuries to the pelvis, while the opposite is true for the short pulse. The mass recruitment issue is discussed later. For the medium pulse, in two cases pelvis injury severities were greater than spine injury severities, in one case the opposite was true, and in the other case, there was only spine injury. The mixed nature of this pattern also supports the hypothesis that medium duration exposure leads to injuries to both regions albeit in a less expected manner, and answers the key questions posed in this study.
The injury results were evaluated in a more general manner, AIS ¼ 1 severity is minor and does not need special attention. If this injury severity is excluded, the pelvis-spine injury counts were such that for the short pulse it was 4-3 (4 pelvis and 3 spine), medium pulse it was 3-4 (3 pelvis and 4 spine), and for the long pulse, it was 3-4 (3 pelvis and 4 spine). The 4-3 to 3-4 switch from short to medium pulse appears to be in line with the injury tradeoff issue acknowledging the limited small size used in this investigation. As another extension of the above analysis, if one were to treat the injury outcomes at the AIS 3þ level (AIS 2 no injury, AIS 3 injury) for the short pulse, the pelvis-lumbar pattern was 3 versus 1 (one specimen had only AIS 2 injuries for both), for the medium pulse it was 2 versus 3, and for the long pulse it was 1 versus 3 specimens. This categorical assignment also supports the tradeoff injury issue.
It should be noted that injuries at the fixation were not coded as it was attributed to, or influenced by, the rigidity of the specimen in the fixative material, and this occurred in specimen 2, associated with the long pulse. Likewise, the potential for cord injury was not included as the neurological issue was deemed to be an inference rather than from the direct medical image readout from a PMHS test. It should be noted that complete transection of the spine is unlikely in the in vivo situation because of the surrounding tissues and their tethering to the human spinal column, Figure 4. AIS levels for the spine and pelvis based the the three duration pulses. Short1 refers to short pulse, specimen 1, etc. Med9 refers to medium pulse, specimen 9, etc. Long 12 refers to long pulse, specimen 12, etc.
while the isolated lumbar PMHS column can transect due to the instability resulting from fracture. It should also be noted that bleeding issues cannot be easily inferred or included in PMHS experiments. Thus, injuries such as pelvis fractures that receive more severe scoring from AIS 4 to AIS 5 for the same structural injury cannot be replicated with the PMHS biological model, although it may be a relatively easier exercise to rescore or elevate the severity of some spinal injuries based on the bony encroachment space into the canal. Whole body PMHS studies may provide some insights into the added instability and potential for hemorrhagic and neural involvements.
By testing four specimens for each pulse, the present results support the role of the temporal variable in the tradeoff between pelvis and spine injuries under vertical loading. As briefly stated in the introduction, the tradeoff issue is due to the mass recruitment with time availability (Yoganandan et al. 2014). This phenomenon was found to be true in the combined pelvis-spine model. The absorption of the same transmitted resultant forces (insignificant difference at p > 0.05 level) in all specimens at the superior end of the spinal column fixation, allowed to delineate the rate effect/pulse type on injury tradeoff between the two body regions. The longer pulse allowing the mass of the inferior structures, in this case pelvis, to vertically accelerate placed an additional demand on the spinal column and exposed it to injuries. In contrast, the lack of the time availability in the short pulse exposed only the inferior or the immediately in-contact structure, i.e., pelvis, to the impact energy and resulted in biasing injuries to this region. Such effects may be more pronounced in a whole-body PMHS and in vivo situations. The greater lag with the longer pulse explains the mass recruitment effect and applies to whole body models. In addition to the mass recruitment effect as a factor for the injury tradeoff issue, other variables may be involved. It is known that the human lumbar spine and other biological materials are viscoelastic (Virgin 1951;Yoganandan et al. 2008). This property may have a greater influence for the long pulse than the short pulse. Disk injuries with or without bony injuries may be more influenced by this material behavior, as disks are more viscoelastic than vertebrae.
It should be noted that previous vertical loading lumbar spine tests categorized into single and multilevel injuries (Yoganandan et al. 2020). Table 1 shows the number of specimens with one and 1 level spinal injuries. They were further classified as (i) any fracture that included body, disk, endplate, and/or transverse process-related injuries, and (ii) only body/disk, ignoring the endplate and transverse process-related injuries. The reason to exclude the endplate and/or transverse process in the later definition is that they are generally clinically insignificant. It should be noted that the AIS coding scheme allows scoring these relatively often minor fractures, with little impact to the treatment regimen, perhaps with a view to code all pathologies (AIS 2015). A more appropriate treatment-based scale such as incapacitation may be needed to improve the scoring of these injuries. As can be seen, for the former definition of injuries, more specimens with long pulses had most 1 level injuries, followed by medium and then, short pulses. In the later definition, the hierarchy was long and medium with equal number of specimens in each category and followed by the short pulse. These findings, albeit from very limited sample size show that the spinal injuries occur more frequently with medium and long pulses than short pulses, offering support via PMHS experiments to the general tradeoff hypothesis. Table 1. Injury analysis with the number of specimens with single and 1 level spine injuries for or the three pulses.

Short
Medium Long