Biomechanical Study of Cervical Disc Arthroplasty Devices Using Finite Element Modeling

Many cervical disc arthroplasty devices (CDAs) have been approved by the United States Food and Drug Administration since 2007, as a viable option for anterior cervical discectomy and fusion (ACDF), and a surgical procedure has been in vogue for more than 50 years [1]. The primary objective of the CDA is to allow motion at the surgical or index level that is absent in the traditional ACDF procedure. Other issues such as adjacent segment disease may also be minimized with the use of the CDA due to the allowance or preservation of the motion at the index level [2]. Thus, the internal biomechanical responses are altered at the index and adjacent levels and also between the different intervertebral joint components: discs and facet joints. The magnitude of changes depends on the type of the CDA. As CDAs have used varying approaches in their basic design and material composition to accomplish the primary goal of motion preservation at the index level, it is important to conduct an investigation to determine the motions, intradiscal pressures, and facet forces under physiological loading conditions that represent the normal day-to-day activities. The objectives of this study were, therefore, to compare these biomechanical parameters for four different CDAs (Table 1) using a finite element model of the human subaxial cervical spinal column.

The current research determined the responses of different CDAs by applying a combination of moment and axial load representing the normal day to day activities of the human head-neck complex. The applied load of 75 N is in line with the previous studies and accounts for the combined loading of muscle forces and the weight of the in vivo head on the neck, and the applied flexion, extension, and lateral bending moments are based on previous studies [3]. The cited list is not all inclusive to maintain the low number of citations to literature studies per format. The range of motion (ROM) is a clinical parameter that clinicians routinely use to assess the status of the patient's spine and spine segment, and pre- and postoperative/post-CDA procedure [1,2]. The intradiscal pressure is a measure of the anterior column load, and the facet forces represent the loading on the posterior column [4]. Taken together, these segmental kinematic and force/pressure-related kinetic measures delineate the biomechanical responses of different CDAs and shed light into the clinical biomechanics of artificial disc replacement therapies. Because U.S. Food and Drug Administration (FDA)-approved artificial discs were used in the study, the results are applicable to the current surgical populations in the military and civilian sectors.

A previously validated three-dimensional osteoligamentous finite element model of the cervical spine column (C2-T1) was used in the study. The finite element model simulated the vertebral body, posterior elements, and intervertebral disc and spinal ligaments. Each vertebral body consisted of the cortical shell, cancellous bone, endplate, and posterior bony complex. The cortical bone was modeled as a linear isotropic material with 0.5-mm-thick shell surrounding the cancellous bone, and 0.2-mm-thick endplate was placed on the superior and inferior surfaces of the intervertebral disc. The intervertebral discs were composed of the nucleus pulposus, annulus ground substance, and fibrosus. The finite element mesh was prepared using a preprocessing software (BETA CAE Systems) and imported into a solver software (ANSYS).

The discs were meshed with anteroposterior asymmetry that arise from the posteriorly displaced nucleus in cervical spine segments. The hyperelastic foam ground substance was defined using the Hill strain energy function [5]. The fibers were defined using membrane elements with tension-only directional fibers embedded in the ground substance. The fibers in the anterior annulus region were defined in a crisscross manner, while in the posterior region, they were defined in the vertical direction. The disc anterior region consisted of sixteen layers and the posterior region consisted of eight layers. The anterior annulus fibers did not form a continuous ring with the posterior annulus fibers; however, a gap was formed bilaterally at the uncovertebral clefts. The ligament material property was defined using nonlinear rate-dependent stress–strain relationships derived from the force–displacement relationships of the cervical spine ligaments. The ligament material properties were obtained from literature [6]. Force-deflection curves for each cervical spinal ligament reported in the cited study were converted to stress–strain curves using the geometrical data from another study [7]. In other words, each spinal ligament had its own nonlinear stress–strain curves inputted to the finite element model. Material properties from the literature were used in the model [8,9]. The intact spine model is shown in Fig. 1. Tables 2 and 3 show the details of the finite element model and material properties used in this study.

Four FDA-approved discs were modeled in this investigation. The Bryan disc (Medtronic Sofamor Danek, Memphis, TN) contains a polycarbonate polyurethane nucleus which articulates with a titanium shell on the top and bottom. A polyurethane sheath surrounds the nucleus and it is attached to top and bottom shell. The nucleus and titanium outer shells elements were modeled as solid hexahedral elements whereas the polyurethane sheath was modeled as a quadrilateral shell element. The nucleus and outer shells were defined as surface-to-surface contact with a coefficient of friction of 0.1. The sheath was defined as automatic general contact to the nucleus. The upper outer shell was attached to the inferior surface of the C5 vertebra whereas the lower outer shell was attached to the superior surface of the C6 vertebra. The contact between the outer shell and vertebra was defined by tied contact. Tables 2 and 3 show the modeling details and material properties.

Computer-aided design (CAD) models for the Prestige LP (Medtronic, Minneapolis, MN) were developed based on the design (ball and trough mechanism) and dimensions provided by the manufacturer (Medtronic Sofamor Danek, Memphis, TN). Both components of the Prestige-LP disc were meshed with hexahedral elements. The keel design was included in the mesh to make the disc and thereby complete osteointegration was ensured with the vertebral endplates. The contact between the metal on metal surfaces was modeled as a surface-to-surface contact definition with a coefficient of friction of 0.1 to simulate the articulation between the parts.

A coordinate measuring machine (FaroArm, Irvine, CA) was used to obtain a geometrical representation of the Mobi-C (Biomet Zimmer, Warsaw, IN) and Secure-C (Globus Medical, Inc., Audubon, PA) devices by means of dense clouds of three-dimensional points from the external surface of the discs. The generated three-dimensional points were imported into a software (Dassault Systems, Waltham, MA) and processed with a digitized shape editor to create the surface geometry from the points and meshed in the software, described above. All three components (two endplates and ultrahigh-molecular-polyethylene core) of the Mobi-C and Secure-C devices were meshed with hexahedral elements. The keel design was included to make the disc to complete osteointegration with the vertebral endplates. The surface-to-surface contact was applied between the moving segments of the discs, and a tied contact condition was defined at the bone-implant interface.

The finite element model was modified at the C5-C6 functional spine unit to simulate the Bryan, Prestige LP, Mobi-C, and Secure-C cervical arthroplasty procedures. The intact and surgical models with CDA implanted spines were constrained at the inferior surface of the T1 vertebra in all degrees-of-freedom, and the sagittal and coronal moment loadings were applied at the superior vertebra. The pure moment load levels were 2 Nm each in flexion, extension, and lateral bending modes. The additional follower force of 75 N to simulate the head supported mass was applied to the model (Fig. 2 [10–12]). The overall ROM was obtained for the intact spine. The motion-controlled hybrid loading protocol was used to determine changes in the responses of CDA spines. The applied moment was varied for the CDA models until the overall column range of motion of the spine reached the magnitude determined in the intact spine under the pure moment. The head supported mass loading was maintained in all the CDA models. The segmental ROM, intradiscal pressures, and facet forces were obtained from these simulations for characterizing the responses of cervical arthroplasty. The changes in these parameters in the three modes, i.e., flexion, extension, and lateral bending, were determined by normalizing the respective data with respect to the intact spine.

All the modified models required moments greater than 2.0 N·m to achieve the same overall motion as the intact spine for each loading mode. The moment required for the Bryan disc was the maximum, average 2.6 N·m. The Prestige LP required an average moment of 2.4 N·m to achieve the intact ROM. Mobi-C and Secure-C needed an average moment of 2.3 N·m each. Moments required to achieve the overall intact range of motion in all the models are listed in Table 4.

Figure 3 shows the changes in the motion responses of the four CDAs with respect to the intact spine in flexion at the rostral, index, and caudal levels. At the rostral and caudal segments, the motion decreased for the Prestige LP, Mobi-C, and Secure-C discs, while the Bryan disc showed an increase. At the index level, the Bryan disc decreased motion, while the Prestige LP, Mobi-C, and Secure-C increased the motion.

Figure 4 shows the changes in the motion responses of the CDAs with respect to the intact spine in extension at the three levels. At the rostral and caudal segments, motions decreased for the Prestige LP, Mobi-C, and Secure-C, while the Bryan disc showed a small increase. At the index level, the Bryan disc decreased the motion while the other three CDAs increased the motion. Figure 5 shows the changes in the motion responses of the CDAs with respect to the intact spine in lateral bending at the three levels. At the rostral and caudal segments, motions decreased for the Prestige LP, Mobi-C, and Secure C while the Bryan disc showed an increase in motion. At the index level, the Prestige LP, Mobi-C, and Secure-C increased the motion, while the Bryan disc decreased the motion.

Figure 6 shows the changes in the intradiscal pressure responses of the four CDAs with respect to the intact spine in flexion at the adjacent levels. At the caudal and caudal segments, pressures decreased for the Mobi-C and Secure-C, while it increased for the Bryan disc and Prestige LP. Figure 7 shows the changes in the intradiscal pressure responses of the four CDAs with respect to the intact spine in extension at the adjacent levels. At the rostral and caudal segments, pressures decreased for the Mobi-C and Secure-C, while it increased for the Bryan disc and Prestige LP devices. Figure 8 shows the changes in the intradiscal pressure responses of the four CDAs with respect to the intact spine in lateral bending at the adjacent levels. At the rostral and segments, pressures decreased for the Mobi-C and Secure-C, while it increased for the Bryan disc and Prestige LP devices.

Facet forces for the index and adjacent levels for all four CDAs in extension are shown in Fig. 9. At the rostral and caudal segments, the facet force decreased for the Prestige LP, Mobi-C, and Secure-C, while the Bryan disc increased the force. Compared to the intact spine, all CDA spines showed an increase in facet contact forces at the implanted level. The Mobi-C showed the highest increase among all the CDAs and Bryan disc showed the minimum increases. The facet contact forces for the index and adjacent levels for all four CDAs in lateral bending are shown in Fig. 10. At the rostral and caudal segments, the facet force decreased for the Mobi-C and Secure-C, while the Bryan disc and Prestige LP devices increased the force. The Mobi-C showed the highest increase among all the models and Bryan disc showed the minimum increase.

As stated in the introduction, the objectives of the study were to delineate the biomechanical responses of different CDAs using a finite element model of the human subaxial cervical spinal column. Although not described in this paper because of the focus of the study, human cadaver experimental data were used to validate the present intact spinal model. As is true in any computational models, it is important for the model to be validated with experimental results to have confidence in its predictions. The model-predicted range of motion response was validated with sagittal bending flexion–extension responses from human cadaver cervical columns that were subjected to 2 Nm of pure moment loading [13]. The experimental data in the cited study were obtained by testing five male and two female young normal spines from subjects (mean 33.4 + 11.7 years). The moment-rotation response corridors, expressed as mean ± 1 standard deviation at each segment of the spine, were reported in the experimental investigation. The moment-rotation response corridors expressed as mean ±1 standard deviation for each segment of the spine was reported in the experimental investigation. The finite element model motion responses were within the experimental corridors for each segment [14].

Under the lateral bending mode, the motions of the present finite element model was also within mean ± 1 standard deviation, and for this purpose experimental data (12 spines, mean age 62 years) were used from another study [15]. Similarly, the intradiscal pressures and facet loads were validated with other human cadaver tests that reported these data. All responses were within the mean ± 1 standard deviation limits [10,16]. Thus, the model is considered to be validated with experimental data.

The mesh convergence study was performed, using the element sizes of 0.25 mm, 0.5 mm, 1.0 mm, 1.5 mm, and 2.0 mm for the two fine, baseline, and the two coarse baseline models, respectively, and element qualities were maintained, i.e., aspect ratio >5, warpage >5, Jacobian < 0.6, and skew angle <60 deg. These steps are in line with the recommended practice for the use of finite element models for clinical applications and also by the ASME Standards committee on the verification and validation in computational solid mechanics [17,18].

The present investigation was conducted with one set of material properties for the spinal components that included the hard (bone) and soft tissues (e.g., ligaments). In addition, the properties were the same for all vertebrae, while they can vary in a human. A detailed parametric sensitivity analysis for the material properties was not conducted as this was not the objective of this investigation. Earlier studies examining the role of the material properties of the hard and soft tissues have shown that the variations in the material properties of the intervertebral disc and ligaments, representing the soft tissues have a greater effect on the range of motion than the variations in the material properties of the trabecular bone core, cortical shell, endplates, and posterior element structures, representing the hard tissues [9,19]. In addition, to incorporate region- or level-specific variations in the bone elastic modulus, it may be necessary to use computed tomography scans of specific patients and assign those properties into the current model. The present model allows that flexibility. Despite these limitations, the authors expect the role of the material properties of the bone to be of a secondary nature, especially in the normal physiological loading situations such as those simulated in this study. These will be future study topics to quantitatively determine these effects.

The motion was preserved at the index level in all CDAs, achieving the intended goal of arthroplasty. A similar finding has been reported in literature [3,20,21]. The magnitude of the motion preservation was different, however, between the discs and it can be attributed to the design of the individual CDA. The decrease in the motion with the Bryan disc, unlike the other three that showed an increased motion, is due to the design of the disc itself: one-piece design with a polyurethane core. The decrease, however, was not close to the ACDF that reduces the motion to almost zero [1]. Thus, while the other three increase motion at the index level, even the Bryan disc shows superiority over the ACDF procedure. With the Prestige-LP, Mobi-C, and Secure-C artificial discs, the motion at the implanted level increased and the motion at the adjacent levels decreased, thus protecting the segments from adjacent segment degeneration. In all cases, the reduction in motion at the adjacent levels was less than 15%. The largest increase in motion was in lateral bending. The Mobi C provided supraphysiologic range of motion in all loading modes, approximately 20% in flexion, 25% in extension, and 70% in lateral bending. This larger increase motion in lateral bending is supported by another modeling study [22]. With the Prestige LP, index level increase of 13% in flexion–extension and 17% in lateral bending from a human cadaver study supports the present results [23]. The Bryan disc alone slightly increased the adjacent level motion to compensate the loss of motion at the index level.

The Prestige LP increased the adjacent level disc pressure more than Bryan disc presumably due to metal on metal contact design. Changing the disc to a polymer on polymer design likely reduces the pressure; however, this was not determined in this study because the currently FDA-approved artificial disc for implanting in patients does not come with this option. Although the Bryan disc has a shock absorbing core in the middle, the compensatory adjacent level motion increase may contribute to its slight increase in the disc pressure at the adjacent segments. The mobile cores with metal to polymer contact in the Mobi-C and Secure-C could be the reason for the reduction in the adjacent level disc pressures.

The facet force at the index level increased in all four CDAs, and other studies have reported similar findings, albeit not with same CDAs and same magnitudes of loading [24]. The facet force was the greatest in the Mobi-C disc, presumably due to hypermobility at the implanted level. The adjacent level facet forces were reduced by all CDAs; and the Bryan disc showed minimal increase. In lateral bending, only the Prestige LP disc showed an increase in the force at the adjacent levels, and this may be because of its design that does not allow translation in the lateral direction.

It should be noted that, surgeons often rely on their experience for selecting a particular artificial disc or conventional fusion as there are no specific recommendations, while there is a longer history on fusion outcomes. The ultimate goal of clinical and biomedical studies would be to determine the optimal artificial disc for certain pathologies. This study focusing on the determination of the range of motion, disc pressure, and facet loads at the index level and adjacent segments, and under flexion, extension, and lateral bending addresses the biomechanical issue, and forms a first step. Other loading modes and severities should be included. It would be possible to come up with a recommendation by combining the design/type of the disc (e.g., one piece versus two or three pieces, permissible degrees-of-freedom of the device itself), outcomes from the mechanical loading (such as those presented in the study), pre-operative morphology of the patient's spine, patient demographics, and occupation. The authors are pursuing these efforts to develop a framework for the surgeon to select the optimal disc. Such a process will assist patient counseling/education as this is also a part of modern health care system.

The material properties chosen for the different components are not unique. For example, studies have shown that the bone behaves as an anisotropic material [25]. The present model assumed the properties of the bone to be isotropic, and this assumption has been used in previous modeling studies of the cervical and lumbar spines [26,27]. The effect of ignoring the anisotropy is considered to be small because the loading used in this study is in the physiologic range, i.e., the severity of traumatic loads that may cause the fractures to the bones of the cervical spine and bone stresses were not the topics of the investigation. In addition, the study-reported parameters include the angular motion, disc pressure, and facet loads are considered to be less affected in the physiological loading range with the assumptions of the linear isotropic material behavior for the bone. This assumption can be quantitatively evaluated in a separate parametric study that includes the anisotropic material law. From this perspective, the present results serve as a first step in the analysis of different CDAs on spine responses.

The objective of the study was to compare the responses of the CDAs with mode specificity under flexion, extension, and lateral bending, as these modes are distinct and clinically relevant. A separate study that includes the factors such as varying geometry and material properties would be needed to perform a statistical analysis, and this is considered as future research.

In general, the CDAs may minimize adjacent segment degeneration. This is central to the clinical reason for choosing CDA over ACDF to reduce future cervical spine surgery for the same patient due to adjacent segment degeneration. Our study also validates another important clinical issue: Facet arthropathy often coexists with cervical disc herniation, an indication for CDA or ACDF. CDA is contraindicated if there is facet arthropathy present at the index level, since it can worsen after CDA and lead to chronic facetogenic neck pain. This investigation offers insights into the local internal mechanics of the index and adjacent levels from different CDAs under the normal physiological loading modes. This study demonstrates that certain artificial discs induce more loads on the facet column than others. For example, the Mobi-C disc increased the load at the index level more than the other three discs. Surgeons considering this CDA option should carefully review the patient's history for facetogenic neck pain and imaging studies for signs of facet arthropathy, since these could worsen after the CDA. While neck pain is prevalent in civilians and military personnel, due to added head supported mass (helmet), its incidence is more in the military populations. As for the adjacent segment degeneration, the Prestige LP and Bryan discs increased the intradiscal pressure at the adjacent levels, in contrast to the Mobi-C and Secure-C discs. The surgeon should carefully review the adjacent level discs for any signs of degeneration. In younger patients with early signs of adjacent disc degeneration (but not yet severe enough to be surgical), it may be appropriate to avoid the use of the Prestige LP and Bryan disc. In these patients, the Mobi-C and Secure-C may serve to protect the adjacent levels by reducing the intradiscal pressure. It should be acknowledged that the foregoing discussion that is based on the current findings should be explored with other factors. They include variations in the spine geometry/curvature, actual status of the intervertebral joints (disc and facet) at the index and adjacent levels (as the same material property was used in this study), occupation (military personnel placing more demand than common civilians, as head supported mass and neck posture such as check six prevalent in the pilots), and postsurgical issues such as return to original occupation, especially among the younger populations. Experiments with human cadaver spines for the ultimate load-carrying capacity may also need a study. The authors are pursing this line of work.

The finite element modeling study demonstrated the influence of different CDA designs on the range of motions, disc pressures, and facet loads at the index and adjacent cervical spinal levels via single level C5-C6 arthroplasty. In addition, this study validated some key clinical observations: CDA procedure is contraindicated in cases of facet arthropathy and may be protective against adjacent segment degeneration. Consideration of facet arthropathy and disc degeneration at the index and adjacent levels should be a factor in the clinical decision-making process for the selection of the CDA.

The research was supported by the National Center for Advancing Translational Sciences, National Institutes of Health, Award Number UL1TR001436, and by the Office of the Assistant Secretary of Defense for Health Affairs, through the Broad Agency Announcement under Award No. W81XWH-16-1-0010. It was also supported by the Department of Veterans Affairs Medical Research. This material is the result of work supported with resources and use of facilities at the Zablocki VA Medical Center, Milwaukee, WI, and the Medical College of Wisconsin. Dr. Yoganandan is an employee of the VA Medical Center. No relationship exists between the authors and the companies of the products being studied in this paper. The opinions, interpretations, conclusions, and recommendations are those of the authors and are not necessarily endorsed by the NIH, Department of Defense or other sponsors.

• National Institutes of Health (Grant No. UL1TR001436; Funder ID: 10.13039/100000002).

• U.S. Department of Defense (Grant No. W81XWH-16-1-0010; Funder ID: 10.13039/100000005).