SPINE Volume 26, Number 6, pp E122–E129
`©2001, Lippincott Williams & Wilkins, Inc.
`
`Load-Sharing Between Anterior and Posterior Elements
`in a Lumbar Motion Segment Implanted With an
`Artificial Disc
`
`Andrew P. Dooris, BS,* Vijay K. Goel, PhD,* Nicole M. Grosland, PhD,*
`Lars G. Gilbertson, PhD,† and David G. Wilder, PhD*
`
`Study Design. A nonlinear three-dimensional finite el-
`ement model of the osteoligamentous L3–L4 motion seg-
`ment was used to predict changes in posterior element
`loads as a function of disc implantation and associated
`surgical procedures.
`Objectives. To evaluate the effects of disc implantation
`on the biomechanics of the posterior spinal elements
`(including the facet joints, pedicles, and lamina) and on
`the vertebral bodies.
`Summary of Background Data. Although several arti-
`ficial disc designs have been used clinically, biomechani-
`cal data—particularly the change in loads in the posterior
`elements after disc implantation—are sparse.
`Methods. A previously validated intact finite element
`model was implanted with a ball-and-cup–type artificial
`disc model via an anterior approach. The implanted
`model predictions were compared with in vitro data. To
`study surgical variables, small and large windows were
`cut into the anulus, and the implant was placed anteriorly
`and posteriorly within the disc space. The anterior longi-
`tudinal ligament was also restored. Models were sub-
`jected to either 800 N axial compression force alone or to
`a combination of 10 N-m flexion– extension moment and
`400 N axial preload. Implanted model predictions were
`compared with those of the intact model.
`Results. Facet loads were more sensitive to the antero-
`posterior location of the artificial disc than to the amount
`of anulus removed. Under 800 N axial compression, im-
`planted models with an anteriorly placed artificial disc
`exhibited facet loads 2.5 times greater than loads ob-
`served with the intact model, whereas posteriorly im-
`planted models predicted no facet loads in compression.
`Implanted models with a posteriorly placed disc exhibited
`greater flexibility than the intact and implanted models
`with anteriorly placed discs. Restoration of the anterior
`longitudinal
`ligament reduced pedicle stresses,
`facet
`loads, and extension rotation to nearly intact levels.
`Conclusions. The models suggest that, by altering
`placement of the artificial disc in the anteroposterior di-
`rection, a surgeon can modulate motion-segment flexural
`
`From the *Iowa Spine Research Center, Departments of Biomedical
`Engineering and Orthopaedics, University of Iowa, Iowa City, Iowa,
`and the †Musculoskeletal Research Center, Department of Orthopae-
`dic Surgery, University of Pittsburgh, Pittsburgh, Pennsylvania.
`Acknowledgment date: March 7, 2000.
`First revision date: July 14, 2000.
`Acceptance date: August 22, 2000.
`Supported by the Sofamor Danek Group, Memphis, TN, The Iowa
`Spine Research Foundation, Iowa City, IA, and The Whitaker Foun-
`dation, Rosslyn, VA.
`The authors thank Dwight T. Todd and Malcolm H. Pope, PhD, for
`their assistance.
`Device status category: 7,9.
`Conflict of interest category: 16.
`
`E122
`
`stiffness and posterior load-sharing, even though the
`specific disc replacement design has no inherent rota-
`tional stiffness. [Key words: artificial disc, biomechan-
`ics, facets, finite element model, lumbar spine] Spine
`2001;26:E122–E129
`
`In many developed countries, the management of low
`back pain is a prevalent problem to the clinician.44 Dis-
`ability of patients with low back pain costs several billion
`dollars in the United States annually because of lost pro-
`ductivity and treatment costs,22 affecting millions of in-
`dividuals.19 Almost three quarters of the United States
`population has at one time experienced back pain, and
`approximately 4% of the population requires surgical
`intervention. As the population ages, the problem will
`certainly grow.12,26,43
`Fusion of painful segments (spinal arthrodesis) is of-
`ten the goal of spinal surgery (25% of 280,000 surgical
`interventions). Fusion typically is done to prevent or cor-
`rect deformity, to stabilize the spine after trauma or
`pathologic destruction, or to eliminate painful move-
`ment of the spinal segments. Despite advances in surgical
`techniques, fusions are neither fully successful nor with-
`out problematic side-effects (e.g., pseudoarthroses and
`donor-site pain). Because fusion— by design—limits mo-
`tion of the fused segment(s), some investigators16,30 be-
`lieve that fusion may induce degenerative changes in the
`neighboring segments, often necessitating additional fu-
`sion surgery.
`To avoid this cascade, some investigators are pursuing
`alternative treatment measures. Tissue engineering ap-
`proaches include regeneration of the nucleus through in-
`troduction of growth factors or gene therapy, as well as
`cell transplantation and scaffolding. Another approach,
`already used clinically, is to mimic normal disc function
`through artificial mechanical means to circumvent bio-
`mechanical changes normally attributed to rigid instru-
`mentation and fusion. Several artificial disc designs have
`been proposed in an effort to reinstate stabilized physio-
`logic motion in the lumbar spine to manage various ail-
`ments that normally indicate spinal fusion.
`A few disc designs have undergone clinical trials with
`limited published results.11,23,31 Likewise, some of the
`disc designs have been evaluated biomechani-
`cally.3,8,25,28 These designs vary greatly in their ap-
`GLOBUS MEDICAL, INC.
`proach to restoring normal disc joint function. One re-
`EXHIBIT 1013
`curring difficulty with each implant lies in determining
`IPR2015-to be assigned
`(Globus v. Flexuspine)
`1 of 9
`
`

`
`SPINE Volume 26, Number 6, pp E122–E129
`©2001, Lippincott Williams & Wilkins, Inc.
`
`Load-Sharing Between Anterior and Posterior Elements
`in a Lumbar Motion Segment Implanted With an
`Artificial Disc
`
`Andrew P. Dooris, BS,* Vijay K. Goel, PhD,* Nicole M. Grosland, PhD,*
`Lars G. Gilbertson, PhD,† and David G. Wilder, PhD*
`
`Study Design. A nonlinear three-dimensional finite el-
`ement model of the osteoligamentous L3–L4 motion seg-
`ment was used to predict changes in posterior element
`loads as a function of disc implantation and associated
`surgical procedures.
`Objectives. To evaluate the effects of disc implantation
`on the biomechanics of the posterior spinal elements
`(including the facet joints, pedicles, and lamina) and on
`the vertebral bodies.
`Summary of Background Data. Although several arti-
`ficial disc designs have been used clinically, biomechani-
`cal data—particularly the change in loads in the posterior
`elements after disc implantation—are sparse.
`Methods. A previously validated intact finite element
`model was implanted with a ball-and-cup–type artificial
`disc model via an anterior approach. The implanted
`model predictions were compared with in vitro data. To
`study surgical variables, small and large windows were
`cut into the anulus, and the implant was placed anteriorly
`and posteriorly within the disc space. The anterior longi-
`tudinal ligament was also restored. Models were sub-
`jected to either 800 N axial compression force alone or to
`a combination of 10 N-m flexion– extension moment and
`400 N axial preload. Implanted model predictions were
`compared with those of the intact model.
`Results. Facet loads were more sensitive to the antero-
`posterior location of the artificial disc than to the amount
`of anulus removed. Under 800 N axial compression, im-
`planted models with an anteriorly placed artificial disc
`exhibited facet loads 2.5 times greater than loads ob-
`served with the intact model, whereas posteriorly im-
`planted models predicted no facet loads in compression.
`Implanted models with a posteriorly placed disc exhibited
`greater flexibility than the intact and implanted models
`with anteriorly placed discs. Restoration of the anterior
`longitudinal
`ligament reduced pedicle stresses,
`facet
`loads, and extension rotation to nearly intact levels.
`Conclusions. The models suggest that, by altering
`placement of the artificial disc in the anteroposterior di-
`rection, a surgeon can modulate motion-segment flexural
`
`From the *Iowa Spine Research Center, Departments of Biomedical
`Engineering and Orthopaedics, University of Iowa, Iowa City, Iowa,
`and the †Musculoskeletal Research Center, Department of Orthopae-
`dic Surgery, University of Pittsburgh, Pittsburgh, Pennsylvania.
`Acknowledgment date: March 7, 2000.
`First revision date: July 14, 2000.
`Acceptance date: August 22, 2000.
`Supported by the Sofamor Danek Group, Memphis, TN, The Iowa
`Spine Research Foundation, Iowa City, IA, and The Whitaker Foun-
`dation, Rosslyn, VA.
`The authors thank Dwight T. Todd and Malcolm H. Pope, PhD, for
`their assistance.
`Device status category: 7,9.
`Conflict of interest category: 16.
`
`E122
`
`stiffness and posterior load-sharing, even though the
`specific disc replacement design has no inherent rota-
`tional stiffness. [Key words: artificial disc, biomechan-
`ics, facets, finite element model, lumbar spine] Spine
`2001;26:E122–E129
`
`In many developed countries, the management of low
`back pain is a prevalent problem to the clinician.44 Dis-
`ability of patients with low back pain costs several billion
`dollars in the United States annually because of lost pro-
`ductivity and treatment costs,22 affecting millions of in-
`dividuals.19 Almost three quarters of the United States
`population has at one time experienced back pain, and
`approximately 4% of the population requires surgical
`intervention. As the population ages, the problem will
`certainly grow.12,26,43
`Fusion of painful segments (spinal arthrodesis) is of-
`ten the goal of spinal surgery (25% of 280,000 surgical
`interventions). Fusion typically is done to prevent or cor-
`rect deformity, to stabilize the spine after trauma or
`pathologic destruction, or to eliminate painful move-
`ment of the spinal segments. Despite advances in surgical
`techniques, fusions are neither fully successful nor with-
`out problematic side-effects (e.g., pseudoarthroses and
`donor-site pain). Because fusion— by design—limits mo-
`tion of the fused segment(s), some investigators16,30 be-
`lieve that fusion may induce degenerative changes in the
`neighboring segments, often necessitating additional fu-
`sion surgery.
`To avoid this cascade, some investigators are pursuing
`alternative treatment measures. Tissue engineering ap-
`proaches include regeneration of the nucleus through in-
`troduction of growth factors or gene therapy, as well as
`cell transplantation and scaffolding. Another approach,
`already used clinically, is to mimic normal disc function
`through artificial mechanical means to circumvent bio-
`mechanical changes normally attributed to rigid instru-
`mentation and fusion. Several artificial disc designs have
`been proposed in an effort to reinstate stabilized physio-
`logic motion in the lumbar spine to manage various ail-
`ments that normally indicate spinal fusion.
`A few disc designs have undergone clinical trials with
`limited published results.11,23,31 Likewise, some of the
`disc designs have been evaluated biomechani-
`cally.3,8,25,28 These designs vary greatly in their ap-
`proach to restoring normal disc joint function. One re-
`curring difficulty with each implant lies in determining
`
`2 of 9
`
`

`
`Load-Sharing in an Implanted Lumbar Motion Segment • Dooris et al
`
`E123
`
`the proper indications and contraindications for disc re-
`placement. Although some investigators view disc re-
`placement as a therapeutic treatment, others see it as a
`safeguard against further degenerative changes.
`After four decades of disc design investigation and
`experimentation, authors have proposed several sets of
`criteria encompassing clinical and nonclinical (e.g., bio-
`mechanical and biomaterial) issues.3,24,31 For example,
`because the replacement disc might last decades, the de-
`sign must minimize wear debris, possibly even more than
`total hip replacement designs.25,28 However, not all of
`the biomechanical parameters can be evaluated by using
`experimental protocols. Some parameters, such as bone
`stresses or joint forces, are difficult to analyze in cadaver
`studies. Cadaver studies involve an array of widely vary-
`ing parameters, many of which are difficult to control,
`characterize, or even measure.
`Not only should healthy posterior elements be re-
`quired before implantation, a disc replacement design
`must also preserve the posterior elements for long-term
`success. Degeneration of the posterior elements, espe-
`cially of the facet joints, has been found to be a source of
`low back pain by various clinical, biologic, and biome-
`chanical studies.1,2,4,5,9,10,13–15,18,32–37,39 – 42 If it is true
`that degeneration of the posterior elements results from
`disc disease in most joint degeneration sequelae,7 then it
`appears reasonable that current protocols would contra-
`indicate implantation with degenerated posterior ele-
`ments.11,17,23 Reports also suggest that in addition to
`preoperative degeneration, the posterior elements may
`be abnormally stressed by disc replacement, and there-
`fore implantation should be avoided.31
`In the present study, a computational investigation
`into the biomechanical changes brought on by disc im-
`plantation in a lumbar spine was undertaken. Because
`flexion and extension are primary displacement modes, a
`half-sagittal plane finite element model of the implanted
`motion segment was built for this analysis. Comparisons
`were made between the intact and the implanted func-
`tional spinal unit (FSU) models. To address issues relat-
`ing to surgical technique, model parameters were varied
`to simulate different anulus window sizes and anteropos-
`terior artificial disc placement.
`The disc implant studied in this investigation was de-
`signed by the Sofamor Danek Group (Sofamor Danek,
`Memphis, TN). This implant uses a tested tribologic de-
`sign adapted from hip replacements, including two,
`matching, ceramic ball and cup components (Figures 1
`and 2). Because the cup is shallow, and its curvature
`decreases near its perimeter, the joint is slightly “slop-
`py.” The contacting surfaces are made of polycrystalline
`alumina and are surrounded by a titanium sleeve, which
`is grooved and beaded for bone in-growth. With this
`implant design and computational model, the authors
`hypothesized that the implanted FSU would resemble the
`intact FSU under sagittal-plane loading, but that surgical
`technique would affect segment mechanics, including
`facet loading, rotational stiffness, and bone stresses.
`
`Figure 1. The Sofamor Danek (Sofamor Danek, Memphis, TN)
`artificial intervertebral disc. Ball and cup components are made of
`polycrystalline ceramic with a grooved and beaded titanium
`sleeve.
`
`Methods
`
`A two-part investigation was undertaken. An in vitro study21
`quantified changes in load-displacement behavior after disc im-
`plantation in fresh ligamentous spine specimens. A finite ele-
`ment (FE) model with the implanted disc was loaded in a man-
`ner similar to that used in the in vitro study protocol, and FE
`model predictions were compared with in vitro results for FE
`model validation. Subsequently, the implanted FE model was
`loaded under more strenuous conditions, and the results of
`loading the implanted and intact models were compared.
`
`In Vitro Experimentation. Seven fresh cadaver spines (L1
`through S1) were cleaned of muscle and connective tissue and
`prepared for loading by fixing the sacrum and inserting rigid
`crossbeams through the L1 vertebra. Three infrared light-
`emitting diodes (LEDs) were attached noncolinearly at each
`vertebral level and connected to an active optical tracking sys-
`tem (Selspot II System, Partille, Sweden). Pure bending mo-
`ments, from 0 N-m to 6 N-m, in increments of 1.5 N-m, were
`applied to L1 via a loading frame attached to the crossbeams.
`For each specimen, the intact spine was loaded in flexion and
`extension, and the three-dimensional displacements of each
`vertebral level at each loading increment were recorded
`simultaneously.
`
`Figure 2. Geometry of the Sofamor Danek (Sofamor Danek, Mem-
`phis, TN) artificial disc. The ball is eccentrically located to the
`posterior of the implant.
`
`3 of 9
`
`

`
`E124 Spine • Volume 26 • Number 6 • 2001
`
`Figure 3. Three-dimensional, nonlinear, finite element half-model
`of the intact L3–L4 motion segment. Moment loads were applied to
`a pair of crossbeams fixed to the superior endplate of L3. Nodes
`at the bottom of L4 were fixed.
`
`Surgery was performed to excise the anterior longitudinal
`ligament at L4 –L5, the anterior portion of the anulus, and the
`nucleus. The joint was distracted, and the ball and cup compo-
`nents of the Sofamor Danek artificial disc were inserted. The
`load-displacement characteristics then were recorded in flexion
`and extension to 6 N-m, as was done previously with the intact
`spine. The average and standard deviation of the motions were
`computed. The data for the implanted spine were compared
`
`with the those of the intact spine to assess the ability of the
`artificial disc to restore normal motion.
`
`Finite Element Analysis. The FE model used in the present
`investigation expands on a previously validated intact model
`produced from computed tomography scans.20,27 Three-
`dimensional geometry was made by layering border designs of
`1-mm–spaced scans. The model (Figure 3) included one FSU,
`L3–L4 with the entire L3 vertebral body, intervertebral disc,
`ligaments, and the superior portion of L4. All seven spinal
`ligaments were represented by nonlinear cable elements, and
`cortical and cancellous bone was represented by eight-node
`continuum elements. Softened, contact, unidirectional ele-
`ments mimicked cartilaginous facet joints. The anulus con-
`sisted of layers of fiber-reinforced continuum elements, with
`fiber orientations at 30° and 120° to the horizontal, and fiber
`modulus changing from the center to the periphery of the disc
`space. Finite element model details are provided in Table 1.
`The model was altered to mimic the anterior surgical ap-
`proach used to implant the Sofamor Danek disc. This required
`removal of the anterior longitudinal ligament and the anterior
`portion of the anulus. Nucleus pulposus elements also were
`also removed. Implant geometry was obtained from PRO/
`Engineer CAD drawings (PTC, Needham, MA) imported into
`MSC PATRAN (MSC Software, Los Angeles, CA) for meshing,
`obtaining accurate geometric proportions in a refined mesh.
`The mesh (Figure 4) then was exported to ABAQUS/Standard
`(HKS, Pawtucket, RI) for analysis. Surfaces were created for
`ball and socket contact and for contact in overextension and
`flexion (along the flat faces).
`Specific assumptions and approximations facilitated model
`generation and analysis. Although the actual implant has
`
`Table 1. Summary of Finite Element Models Listing Element Type, Quantity, and Properties for the Model Entities. The
`Changes in These Parameters Due to Surgery are Also Listed
`
`Implanted
`
`Element Type
`
`E (MPa), v
`
`100, 0.3
`1200, 0.3
`100, 0.3
`1200, 0.3
`3500, 0.25
`3500, 0.25
`
`0–20, 0.3
`0–20, 0.3
`0–19.5, 0.3
`0–12, 0.3
`0–15, 0.3
`0–33, 0.3
`0–58.7, 0.3
`
`1E–6, 0.5
`Ground Substance 4.5, 0.45
`Fibers 357–550, 0.3–0.5
`
`380, 0.26†
`114, 0.32‡
`
`4 of 9
`
`Total nodes
`Total elements
`Bone*
`L3 Cancellous
`L3 Cortical
`L4 Cancellous
`L4 Cortical
`L3 Pedicle
`Lamina
`Ligaments*
`Anterior longitudinal
`Posterior longitudinal
`Ligamentum flavum
`Interspinous
`Supraspinous
`Capsular
`Intertransverse
`Intervertebral Disc*
`Nucleus
`
`Annulus
`
`Implant
`Aluminum oxide
`Titanium
`Socket component
`Ball component
`
`Intact
`
`3805
`3765
`
`832
`368
`520
`200
`56
`29
`
`40
`24
`8
`14
`4
`28
`10
`
`392
`
`896
`
`0
`0
`0
`0
`
`7451
`5661
`
`832
`368
`520
`200
`56
`29
`
`0
`24
`8
`14
`4
`28
`10
`
`0
`
`8 node continuum
`
`2 node cable
`
`8 node continuum
`
`560 (A⫺ model)
`656 (A⫹ model)
`
`8 node continuum with 2 node
`rebar
`
`1856
`800
`1658
`998
`
`8 node continuum
`
`Material property references denoted by superscript: *27; †38, ‡29.
`
`

`
`Table 2. Comparison of FE Predictions and In Vitro Results. Predictions Generally Fell Within One Standard Deviation
`After Accounting for Uncontrolled Implantation Procedures
`
`Load-Sharing in an Implanted Lumbar Motion Segment • Dooris et al
`
`E125
`
`Rotation Displacement Under Load Flexion ⬎ 0, Extension ⬍ 0; (°)
`
`In Vitro
`
`AD&A⫺
`
`AD&A⫹
`
`PD&A⫺
`
`PD&A⫹
`
`Applied
`Moment
`(N-m)
`
`⫺6
`⫺4.5
`⫺3
`⫺1.5
`0
`1.5
`3
`4.5
`6
`
`⫺4.4 ⫾ 1.1.8
`⫺3.1 ⫾ 1.7
`⫺2.1 ⫾ 0.7
`⫺0.8 ⫾ 0.3
`0
`1.5 ⫾ 1.4
`2.8 ⫾ 1.5
`3.7 ⫾ 1.4
`5.1 ⫾ 1.5
`
`⫺6.0
`⫺4.8
`⫺3.6
`⫺2.0
`0
`1.6
`2.5
`3.4
`4.0
`
`⫺4.2
`⫺3.4
`⫺2.5
`⫺1.3
`0
`1.6
`2.5
`3.3
`4.0
`
`⫺6.0
`⫺3.6
`⫺4.8
`⫺2.0
`0
`3.0
`4.9
`4.9
`5.8
`
`⫺4.1
`⫺3.3
`⫺2.3
`⫺1.2
`0
`2.6
`4.5
`4.5
`5.2
`
`grooves and beads for osseointegration with the vertebral end-
`plates, this complex surface was simulated as a flat plane, with
`implant nodes fixed to the endplate nodes, thereby assuming
`perfect bone ingrowth. In addition, the implant was scaled to
`match the spine model’s disc space height, resulting in the disc
`occupying 40% of the segment’s cross-sectional area in the
`transverse plane. Distraction of the segment to restore disc
`height in a clinical restoration was not simulated, because the
`intact model had normal disc height initially. The alumina por-
`tion was assigned appropriate elastic material values (E ⫽ 380
`GPa, ␯ ⫽ 0.26),38 as was the titanium portion (E ⫽ 114 GPa, ␯
`⫽ 0.32).29 Friction was assigned a value of 0.09, which is the
`value reported in literature for wet alumina–alumina sliding
`contact.6 The nodes on the bottom of the L4 vertebra were
`fixed in space, whereas the nodes on the midsagittal plane were
`fixed in the lateral direction (i.e., assuming symmetry across
`that plane).
`This FE investigation included two loading schemes, corre-
`sponding first to model validation and then model investiga-
`tion. Loading conditions for model validation included a small
`compressive force of 10 N (approximately the weight of the
`loading frame used in the in vitro investigation ⫹ superior spine
`segments) and up to 6 N-m flexion or extension moment (as
`was also used in the in vitro study). To simulate compressive
`loads, forces were applied to the nodes of the superior L3 end-
`plate. Model predictions from this set of conditions were com-
`pared to in vitro results. The second set of conditions included
`an axial compressive force of 400 N combined with flexion and
`extension moments up to 10 N-m. The effects of pure axial
`compression also were studied by applying forces of up to 800
`N to L3.
`Changing model parameters facilitated study of some seg-
`
`Figure 4. Finite element model of the artificial disc components.
`Contact surfaces were defined for the entire ball and socket
`interface. Geometry was imported from CAD designs.
`
`ment biomechanics dependencies on surgical technique. The
`artificial disc was placed as posteriorly in the disc space as
`allowed by the remaining anulus (“PD”; Figure 5A). In the
`present study, this meant shifting the prosthesis 8 mm anteri-
`orly, as far as it could be shifted without overhanging the end-
`plate (“AD”; Figure 5B). In another parameter, the smallest
`annular window possible was cut (“A⫹”; Figure 5A), as dic-
`tated by the width of the implant for anterior insertion,
`whereas in a complimentary case, fully one third of the anulus
`was removed (“A⫺”; Figure 5B). After removal, the remaining
`anulus was 42% of the disc space in the A⫹ case and 35% in
`the A⫺ case, the difference being entirely in the anterior third of
`the anulus. Lastly, the anterior longitudinal ligament was re-
`stored to its original form (not removed) to isolate the effects of
`anular and anterior longitudinal ligament resection.
`
`Results
`
`The data shown in Table 2 indicate an FE model valid for
`further analysis. By accounting for both surgical vari-
`ables in the in vitro procedures, FE predictions for rota-
`tions under applied moments either matched or strad-
`dled in vitro results. In flexion, FE model predictions
`were either larger or smaller than in vitro results, de-
`pending on model parameters. In extension, the anteri-
`orly placed disc results closely matched in vitro data. In
`general, predictions for segment rotations fell within one
`standard deviation of the in vitro data.
`In flexion and extension plus 400 N preload, the im-
`planted spine segment model exhibited less rotational
`stiffness than the intact model. In extension, segment
`rotations were sensitive primarily to the amount of re-
`maining anulus. Segments with less anulus remaining
`(AD/PD, and A⫺) rotated approximately 40% more in
`extension (⫺5.12°, ⫺5.23° vs. ⫺3.7°), but segments
`with more anulus remaining (AD/PD, and A⫹) rotated
`approximately 30% more than the intact model
`(⫺4.78°, ⫺4.79° vs. ⫺3.7°). Under flexion moments,
`however, disc placement and amount of anulus affected
`rotation values. Placing the disc anteriorly decreased
`flexion by 19% (4.8° vs. 5.88), but placing it posteriorly
`increased flexion by 44% in the PD/A⫺ and by 36% in
`the PD/A⫹ cases. Under axial compression, AD and
`A⫹/A⫺ cases extended (⫺1.5° at 800 N), and PD and
`A⫹/A⫺ cases flexed (2.4°and 4.4° for PD and A⫹/A⫺,
`
`5 of 9
`
`

`
`E126 Spine • Volume 26 • Number 6 • 2001
`
`Figure 5. Surgical variables as implemented in the finite element model. The artificial disc was placed
`anteriorly and posteriorly, and small and large amounts of anulus were removed. Posterior place-
`ment of the disc was limited by the anulus, and anterior placement was limited by the end plate.
`PD&A⫺ ⫽ posterior disc with a large amount of anulus removed; PD&A⫹ ⫽ posterior disc with a
`small amount of anulus removed; AD&A⫺ ⫽ anterior disc with a large amount of anulus removed;
`AD&A⫹ ⫽ anterior disc with a small amount of anulus removed.
`
`respectively). In contrast, the intact segment’s rotation
`was negligible under compression.
`Forces across the facet joints increased with artificial
`disc implantation (Figure 6) under most loading condi-
`tions. Only with a posteriorly placed artificial disc at low
`loads (2 N-m or less) were facet loads lower in the im-
`planted FSU. Placing the artificial disc anteriorly in-
`creased the load shift to the facets to more than that in
`the intact case and in the posteriorly placed case (at loads
`less than 10 N-m). The load shift also occurred earlier,
`such that facet loads were generated before any exten-
`sion moment. Total facet unloading with the anteriorly
`placed artificial disc was not achieved until more than 2
`N-m moment was applied in flexion. Under 800 N com-
`pression, the facet joint was unloaded completely with
`the posterior artificial disc placement. With anterior
`placement, however, loads across the facet joint in-
`creased to more than 150% of those in intact case (38 N
`vs. 98 N).
`Changes in the von Mises stresses in the L3 pedicle
`and lamina caused by implantation reflected the rota-
`tional changes. In the PD/A⫺ and PD/A⫹ models under
`10 N-m, L3 pedicle stresses increased by approximately
`94% and 76% from those in the intact model, respec-
`tively, under flexion moment, and by approximately
`50% under extension moment. The anteriorly placed ar-
`tificial disc model predicted less of a change, approxi-
`mately 30% higher pedicle stresses under 10 N-m exten-
`sion moment, but 15% decreased stresses under 10 N-m
`flexion moment. In pure compression (800 N), pedicle
`stresses in the posterior disc cases decreased to half those
`of the intact case, but doubled with anterior artificial disc
`placement, as compared with those in the intact case. In
`the lamina, von Mises stresses in the PD/A⫹ and PD/A⫺
`cases were nearly twice their intact values in flexion and
`50% more in extension. Functional spinal units with
`anteriorly placed implants exhibited a 25% increase in
`lamina stresses in extension, but a 15% decrease in
`flexion.
`
`Anulus fiber stress patterns also changed with implan-
`tation. Generally, the analyses predicted large decreases
`in fiber stresses. Maximum fiber stresses in the anterior
`(A⫹ cases), lateral, and posterior anulus ranged from as
`low as 20% to 40% of the intact model values in 10 N-m
`extension. In flexion, however, maximum fiber stresses
`in the posterior anulus equaled those in the intact case at
`10 N-m. Increasing or decreasing the amount of remain-
`ing anulus (i.e., the size of the annular window) pro-
`duced small changes in maximum fiber stresses. In pure
`compression, the measured stress in the anulus fibers
`with posterior placement of the artificial disc was just
`10 –30% of the stress measured in the intact case, with
`slightly higher stresses observed with anterior placement.
`In the implanted FSU, anular bulge decreased, and the
`disc space did not compress under axial load.
`Stresses in the implanted spine model’s L3 cortical and
`cancellous and L4 cancellous bone elements increased by
`10 –30% under applied flexion and extension moments.
`L4 cortical bone stresses, however, did not follow the
`
`Figure 6. Facet contact forces as a function of applied moment,
`disc placement (AD and PD), and amount of remaining anulus (A⫹,
`A⫺). Disc placement affected posterior loads more than anulus
`resection. Facet contact forces in implanted cases were higher
`than those in intact conditions in segments with moments greater
`than 4 N-m; facet contact forces for moments less than 4 N-m
`were lowest in segments with posteriorly placed discs.
`
`6 of 9
`
`

`
`Load-Sharing in an Implanted Lumbar Motion Segment • Dooris et al
`
`E127
`
`Table 3. Effect of Restoring the Anterior Longitudinal
`Ligament. Restoration Significantly Reduced Segmental
`Rotation in the AD&Aⴙ/ⴚ Cases and Reduced Both
`Facet Loads and Rotation in the PD&Aⴙ/ⴚ Cases Nearly
`Back to Intact Levels for 10 N-m Extension with 400 N
`Compressive Preload
`
`Condition
`
`Facet Load (N)
`
`Rotation (°)
`
`Intact
`AD&A⫹/⫺
`AD&A⫹/⫺ & ALL
`PD&A⫹/⫺
`PD&A⫹/⫺ & ALL
`
`166
`221/222
`220/261
`211/230
`163/169
`
`⫺3.70
`⫺4.8/⫺5.1
`⫺3.9/⫺3.7
`⫺4.8/⫺5.2
`⫺3.8/⫺3.9
`
`same general increase: L4 cortical bone stresses continu-
`ously decreased with extension. Whereas the PD/A⫺ L4
`maximum cortical stresses were more than twice those of
`the intact model under 10 N-m applied flexion moment,
`they decreased to 7% less than those in the intact model
`in 10 N-m extension. Similarly, in the PD/A⫹ case, max-
`imum L4 cortical stresses decreased from 62% more
`than those in the intact model at full flexion to just 5%
`more than those in the intact model at full extension. In
`compression, the average von Mises L4 cortical bone
`stress decreased from 4.9 MPa in the intact model to 1.76
`MPa in the PD/A⫹ case, but increased to 5.4 MPa in the
`AD/A⫹ case. Similar trends were seen for the L4 cancel-
`lous bone in compression.
`Restoring the anterior longitudinal ligament signifi-
`cantly affected segment mechanics in extension, as is
`shown in Table 3. The ligament had no effect on the
`segment under flexion because of its inability to transmit
`compressive loads. It did, however, marginally reduce
`rotation under compression for the AD/A⫾ cases be-
`cause of the segment’s tendency to extend under com-
`pressive load. Rotation, L3 pedicle stresses, and facet
`contact forces all returned to nearly intact values under
`extension with posterior artificial disc placement. Lam-
`ina stresses decreased from 50% more than intact values
`to just 20% more. L4 cortical stresses increased in exten-
`sion between the anterior longitudinal ligament-restored
`cases and their implanted counterparts.
`
`Discussion
`
`For artificial disc replacement to be successful, biomate-
`rial, biomechanical, and clinical requirements all must be
`met. Biomechanically, extreme alterations in the load-
`sharing among the different spinal structural elements
`(such as the disc anulus and the facet joints) caused by
`artificial disc replacement could lead to instability, soft
`tissue destruction, abnormal bone resorption/growth, or
`fracture. Therefore, analysis of changes in load-sharing
`relations may aid in evaluating the indications and con-
`traindications for disc replacement. Because of the diffi-
`culties in addressing these issues through in vitro exper-
`iments, complementary approaches are mandated.
`In the present study, a computational approach (with
`experimental validation) was used to assist in evaluation
`
`of the biomechanics of an artificial disc replacement. An
`FE model of a spinal motion segment incorporating a
`ball-cup–type artificial disc design was created.
`The experimental protocol and the corresponding FE
`model had several differences worth enumerating here. It
`was not practical to quantify the degree of degeneration
`in each specimen used for in vitro testing, nor was it
`practical to simulate those parameters in the FE model.
`During the experiments, the disc implant size was se-
`lected based on the disc dimensions of a specimen and the
`size of the implants available to the authors. Also, during
`insertion, it was not practical to control precisely the
`location of the implant with respect to the disc morphol-
`ogy. The FE model assumed perfect osseointegration,
`also a departure from the in vitro test condition. (This
`may, however, compensate for the distraction induced
`across the