Original Article

ISO 12189 standard for the preclinical evaluation of posterior spinal stabilization devices – II: A parametric comparative study

Proc IMechE Part H: J Engineering in Medicine 2016, Vol. 230(2) 134–144 Ó IMechE 2015 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/0954411915621588 pih.sagepub.com

Luigi La Barbera1,2, Francesco Costa3 and Tomaso Villa1,2

Abstract The International Standardization Organization (ISO) 12189 standard was recently introduced to preclinically evaluate and compare the mechanical properties of posterior stabilization devices. This scenario presents some new significant steps ahead over the vertebrectomy model recommended by American Society for Testing and Materials (ASTM) F1717 standard: the modular anterior support allows for describing a closer scenario to the effective clinical use as well as to test very flexible and dynamic posterior stabilization devices. Despite these significant advantages, ISO 12189 received little attention in the literature. Anatomical parameters depending on the spinal level were compared to the published data or original measurements on biplanar stereoradiography on 13 patients. Other mechanical variables, describing the test set-up design, were considered and all parameters were investigated using a numerical parametric finite element model. Stress values were calculated by also considering their worst-case combination. The standard set-up represents quite well the anatomy of an instrumented average thoracolumbar segment. The parametric comparative analysis demonstrates a significant (even beyond + 350%) maximum increase in the stress on the device, compared to the standard currently in use. The anterior support stiffness plays the most detrimental effect (maximum stress increases up to 396%). The initial precompression step has an important role in determining the final stress values achieved at peak load (up to + 76%). Moreover, when combining these two contributions, an even higher stress increase may be achieved (up to 473%). Despite the other anatomical parameters playing a secondary role, their worst-case combination demonstrates that a device could potentially undergo higher stresses than those reached according to standard suggestions (maximum increase of 22.4% at L1). Any user/ designer should be aware of these effects when using ISO 12189 standard for the preclinical evaluation of posterior spinal stabilization devices.

Keywords ISO 12189, ASTM F1717, ISO 10243, standard, fatigue, preclinical evaluation, pedicle screw, spine stabilization, finite element, parametric study, finite element model

Date received: 15 April 2015; accepted: 16 November 2015

Introduction The International Standardization Organization (ISO) introduced ISO 121891 standard, useful to study the mechanical reliability of posterior spinal stabilization implant assemblies. This standard proposes a very simple physiological model, where the well-known American Society for Testing and Materials (ASTM) F17172 was updated with the interposition of a supplemental central vertebral body (VB)-like test block and two synthetic discs. Each one is made up with three calibrated springs

1

Laboratory of Biological Structure Mechanics, Department of Chemistry, Materials and Chemical Engineering ‘Giulio Natta’, Politecnico di Milano, Milano, Italy 2 IRCCS Istituto Ortopedico Galeazzi, Milano, Italy 3 Department of Neurosurgery, Humanitas Clinical and Research Center, Milano, Italy Corresponding author: Luigi La Barbera, Laboratory of Biological Structure Mechanics, Department of Chemistry, Materials and Chemical Engineering ‘Giulio Natta’, Politecnico di Milano, Piazza Leonardo da Vinci 32, 20133 Milano, Italy. Email: [email protected]

La Barbera et al. consistent with ISO 10243,3 which are claimed to catch the overall physiological behaviour of lumbar discs under compression.1 Despite the vertebrectomy (worst-case) scenario implemented by ASTM F1717 guaranteeing a high safety coefficient for the tested implant, ISO 12189 can offer some important advantages. On one hand, such a configuration is more representative of the effective clinical use of rigid stabilization devices, which are usually combined with an anterior support (e.g. intervertebral cages, bone grafts) in order to obtain stability/ fusion of a specific spine segment. On the other hand, semi-rigid, flexible and dynamic stabilization devices were also proposed to allow for a more physiological load sharing with the anterior column and other spinal structures.4–8 Particularly, in this case, the surrounding structures have a significant stiffness to support some load amount: in this light, ISO also allows for testing very flexible posterior stabilization devices, which would deflect excessively when tested in a vertebrectomy scenario. Several studies proposed home-made procedures for the evaluation of rigid and dynamic posterior implants with the introduction of an anterior support.9–15 However, as far as we know, only few authors used ISO 12189 standard to investigate the mechanical performances of spinal implant:16–19 this may suggest that ISO 12189 standard is not so well known or not so much used. Moreover, the relative importance of specific values suggested within the standard in determining the outcome of the experimental test is not so clear and no systematic study has ever been performed: even slight changes in the set-up parameters are expected to influence the outcome of the experimental test. A recent study demonstrated that ASTM F1717 reproduces quite well the overall morphology of an average thoracolumbar segment from a physiological population of patients; however, the stress may potentially increase up to 22% in an anatomical worst-case scenario found at L1 level.20 Other studies compared ASTM2 vertebrectomy and ISO1 anterior support model to more clinical relevant scenarios.16,17 Moreover, La Barbera and colleagues21 proposed a systematic procedure for the validation of the numerical model of a functional spine unit (FSU) construct assembled according to ISO 12189, also catching the preload due to the initial precompression step. This study critically assesses the appropriateness of the complete two-FSU-assembled construct built according to ISO 121891 by means of a parametric finite element (FE) model. Thus, the aims of this study are as follows: (1) the comparison between the set of values suggested by the standard and their value within the physiological range in the thoracolumbar spine and (2) the comparison between the maximum stress level on the device according to standard suggestions and the one obtained varying each parameter within the physiological range.

135

Materials and methods Reference model A parametric computer-aided design (CAD) model of ISO 12189 set-up describing one-quarter of a complete two-FSU-assembled construct was considered, assuming its symmetry in terms of geometry, boundary and loading conditions with respect to the anatomical planes (Figure 1). The geometry of the polyaxial screw was simplified according to the approach already used in a previous comparative study.20 A reference condition compliant with the nominal dimensions recommended by ISO 12189 standard was defined: details about the set of the values assumed for each parameter in this condition are reported in the second column of Tables 1 and 2. The linear elastic material properties were assumed for the spinal fixator (titanium alloy, E = 110 GPa, n = 0.3) and the polyethylene (PE) block (E = 1.05 GPa, n = 0.4), while the elastic modulus of the spring (E = 195 GPa for the central, E = 97.5 GPa for the posterior one, n = 0.3 for both) was calibrated on the nominal axial stiffness value (375 N/mm according to ISO 121891) according to the procedure already described. Symmetry constraints on the horizontal and vertical planes were defined.16,17 Tie constraints were assumed at screw–rod and screw–block interfaces.7,20 Contacts were defined at spring–block interfaces, using a surface-to-surface approach with ‘hard’ normal behaviour and a friction coefficient of 0.35.16,17,21 In order to simulate the ISO 12189 standard procedure,1 which recommends to precompress the construct before assembly and the subsequent loading, the following steps were considered:21 





Precompression: contacts were defined between the inferior surfaces of the springs and the superior surfaces of the central PE block (hard behaviour with a friction coefficient of 0.3). Since the free length of the springs is 25 mm, while the initial distance between the ultra-high-molecular-weight polyethylene (UHMWPE) blocks is 24 mm according to ISO recommendations,1 a rigid surface moving downwards was used to compress the spring of 1 mm. Release: the rigid plane was moved upwards so that the spring was allowed to gradually come into contact with the inferior surface of the upper PE block, up to equilibrium with the posterior implant. Loading: a vertical force of 300 and 1000 N (600 and 2000 N considering the overall construct) was applied using a spherical rigid surface, which simulates the joint used to experimentally apply the vertical load.

The rod was discretized using eight-node hexahedral elements; the screw head and body were meshed using a hybrid four-node tetrahedral and eight-node hexahedral mesh, respectively (Figure 1(b)). A mesh convergence analysis was performed on the model describing the reference configuration considering the von Mises

136

Proc IMechE Part H: J Engineering in Medicine 230(2)

Figure 1. (a) Definition of the parametric model describing standard ISO 121891 configuration and interpretation of the corresponding parameters in the spine. (b) Overview of the meshed model simulating the reference configuration. The arrows indicate the regions of interest where the von Mises stress values were considered. *Screw insertion point within the pedicles.

stress on the screw head and on the rod; the mesh refinement was judged satisfactory only after a percentage variation less than 2% was reached. The investigated model presented 27,000 elements for the rod, 140,423 for each complete polyaxial screw (16,894 and 125,960 elements for its head and body, respectively) and 151,218 and 61,565 elements for the superior and central PE blocks, respectively; each solid spring was discretized with 37,932 elements. The simulations were run in ABAQUS/Standard 6.10 (Dassault Syste`mes Ri. Simulia, Waltham, MA, USA) assuming geometrical non-linearity.

In order to investigate the value of the parameters considered within ISO 12189 standard and investigate the possibility to include some other important ones, a total of 10 parameters were analysed (Figure 1(a)). According to a previous study which analysed ASTM F1717 standard,20 only those parameters which have high influence on the von Mises stress (sVM) arising in the device were considered in the parametric model. The parameters were identified keeping the same definitions used within ISO 121891 and ASTM F1717,2 when possible. Then they were classified as follows.

La Barbera et al.

137

Table 1. Anatomical parameters: value suggested by ISO 12189,1 physiologic range value, maximum increase in von Mises stress value on the screw (or rod) normalized with respect to the reference configuration (0% corresponds to the reference configuration) and correlation coefficient (R2) for the relationship between the parameters and the stress on the screw. Anatomical parameters Parameter

Suggested valuea

Release

BMA (mm) CoFR (mm) PDIs (°) hAL (mm) ASA (mm) SP1 (mm) SP2 (mm) SP3 (mm)

40 21 15 38 31.67 40 25 16.75

R2

Maximum increase in sVM (%)

Range of variation

15.0–43.0 (Figure 2)b 4.4–25.9 (Figure 3)b 29.0 to 50.9b 13.7–38.4 (Figure 4)b 33.2–41.5 (Figure 5) 39–41 24–26 15.75–17.75

Peak load

Screw

Rod

Screw

Rod

0.9 0.1 0.4 3.9 0.1 0.1 1.1 0.0

0.4 0.4 0.7 1.9 1.6 1.6 0.7 2.9

11.8 0.6 1.3 5.7 4.4 4.4 1.4 0.3

9.9 2.2 -0.3 2.2 7.7 7.7 2.6 1.5

0.98 0.98 – 1.0 1.0 1.0 1.0 –

BMA: block moment arm; CoFR: centre of fixation to rotation; PDI: pedicular inclination; hAL: half of the active length; ASA: anterior support arm. a The set of parameters reported in this column was assumed in the reference configuration. b Range of variation from La Barbera and Villa.21

Anatomical parameters The anatomical parameters describe the biomechanics of the FSU, the position and the stiffness of a physiologic lumbar intervertebral disc (IVD), the orientation and the position of the pedicles with respect to the anatomical planes, assuming that the principal axes of the screw and the pedicle coincide (Figure 1(a)). The anatomical parameters are as follows (Table 1): 





Block moment arm (BMA): it is the lever arm of the applied load according to the standard set-up, roughly equivalent to the distance between screw insertion point (IP) in the pedicle and the followerload (FL) line path in the spine (Figure 1(a)). The FL can be used to model the overall contribution due to upper body weight as well as muscle forces.22 The FL line path can be assumed passing through the centre of each VB.23–26 BMA varies over a wide range of values within the physiological range according to the data previously published20 (Table 1). Centre of fixation to rotation (CoFR): it is the vertical distance between screw IP within the PE block and the centre of rotation of the cylindrical pin used to apply the load in the test set-up, whereas in the spine it may represent the distance between screw IP and the instantaneous centre of rotation (ICR) of the adjacent FSU, which is located close to the centre of the IVD or slightly anterior in flexion for an intact and not instrumented FSU.24 This parameter could be defined both on the superior (CoFRsup) and on the inferior (CoFRinf) IVD adjacent to the instrumented segment (Figure 1(a)). CoFR varies over a wide range of values within the physiological range according to the data previously published20 (Table 1). Pedicular inclination (PDI) with respect to the sagittal planes: according to Panjabi et al.’s27,28







definition. It varies over a wide range (Table 1): in order to reduce the computational cost of the parametric investigation, PDIs were set to the values which lead to the highest percentage stress increase within the physiological range (0° according to La Barbera et al.20). Half of the active length (hAL): it is the distance between the axis of the screws in the test set-up (neglecting spine curvature); it can be interpreted as the distance between the ideal IPs of the pedicular screws belonging to a single FSU in the craniocaudal direction. This distance can be taken parallel to the line which connects the centres of the vertebral bodies and approximates the curvature of the FSU. Given the geometrical (i.e. practical) constrains due to spring presence, the active length (hAL) was varied 61 mm around the position suggested by ISO 121891 standard. Spring position (SP), in particular, three distances are considered (Figure 1(a)): the distance between the axis of the anterior springs and screws IP in the antero-posterior direction (SP1), the distance between the axis of the anterior springs and the posterior one’s in the antero-posterior direction (SP2) and the distance between the axis of the anterior springs and the sagittal plane (SP3). The dimensions for these parameters were proposed by Mosnier19 on the basis of the anatomical measurements from Semaan et al.29 dealing with endplates’ (EPs) width and depth. The positions of the springs in the antero-posterior direction (SP1 and SP2) as well as in the medial-lateral one (SP3) were varied conventionally 61 mm with respect to standard ISO 12189 suggestions; Anterior support arm (ASA): it is the distance in the test set-up between the centre of gravity of the spring system and the screw IP in the PE blocks

138

Proc IMechE Part H: J Engineering in Medicine 230(2)

Table 2. Mechanical parameters: value recommended by ISO 12189,1 investigated range of values, maximum increase in von Mises stress on the screw (or rod) normalized with respect to the reference configuration (0% corresponds to the reference configuration). Mechanical parameters Parameter

Reference valuea

Maximum increase in sVM (%)

Investigated range

Release

k (N/mm) Pre (mm) d0 (mm)

375 1 2b

100, 147, 375, 459 0–1 0–3

Peak load

Screw

Rod

Screw

Rod

2.3 0.0 0.0

6.2 0.0 2.9

395.6 51.2 6.0

359.1 75.7 8.0

a

The set of parameters reported in this column was assumed in the reference configuration. The current version of ISO 12189 was assumed to have a d0 value equal to 0 mm.

b

(Figure 1(a)); it can be interpreted in the spine as the distance between the screw IP in the pedicle and the centre of gravity of the anterior column (IVD and ligaments). Notice that if the three springs have the same stiffness and are loaded in parallel, ASA is equal to 3 3 SP1 2 SP2 (about 31.67 mm in the reference configuration), otherwise it falls between SP1 and SP1 2 SP2 (as the load increases, it moves anteriorly towards SP1, but it moves posteriorly upon release). The ASA was varied consequently of 61 mm in the antero-posterior direction. In order to have some quantitative data describing ASA, a total of 13 patients, 6 males (average age: 70 years, range: 59–81 years) and 7 female (average age: 66 years, range: 48–74 years), were collected from the database of Neurosurgery Department of IRCCS Istituto Ortopedico Humanitas (Rozzano, Milano, Italy). All the selected patients signed the consent for the processing of their personal data and received during their hospitalization a standing position X-ray performed with EOS System (EOS Imaging, Paris, France) for clinical reasons. The measurements (maximum intraobserver difference of 1.2 mm) were expressed as a mean value 6 standard deviation for each spinal level within the thoracolumbar spine and then compared to the value suggested by ISO 121891 standard.

Mechanical parameters The mechanical parameters describe the test set-up configuration and some of them may not be explicitly considered in the current ISO 121891 standard (Table 2): 

Spring stiffness, (k): it is the axial compressive stiffness of each calibrated spring compliant with ISO 102431 standard. According to ISO standard,1 only few discrete values of k are available, in particular a colour code is adopted. Yellow, red, blue and green correspond to springs having a nominal calibrated axial stiffness of 459, 375, 147 and 100 N/ mm, respectively: this code will be used herein, while black will be used for the vertebrectomy case





(no springs). To be consistent with the compressive behaviour of lumbar FSUs and IVDs, three springs in the same plane have to be used with the specific layout described in standard ISO 12189.1 ISO 12189 standard allows for using different combinations of springs in order to tune the ratio of internal loads on the implant: this aspect has always been neglected in the literature and only Mosnier19 partially investigated it with simplified FE models and experimental tests. Therefore, to compare the effect of using different springs’ stiffness, all possible combinations are analysed in this study (Table 2). Precompression (Pre): it describes the amount of initial compression applied on the springs, in order to make them fit between the test blocks. As already explained, the parametric model was drawn so that it was possible to investigate the effect of a variable precompression in the range between 0 mm (meaning no precompression) and 1 mm (according to ISO 12189 standard). While investigating the contribution of the initial precompression, the initial distance between the test blocks was set to 24.5 and 25 mm; therefore, it was also possible to investigate the contribution of a slight precompression of 0.5 mm (Pre = 0.5 mm) and no precompression at all (Pre = 0 mm). Unsupported screw length (d0): it is the portion of the screw which is left outside the block (Figure 1(a)), as introduced in a recent study.20 This allows the polyaxial screw head free to rotate following rod curvature. This parameter was assumed equal to 2 mm in the reference configuration, as already suggested in ASTM F270630 standard was varied between 0 and 3 mm.

The other constant.20

remaining

parameters

were

kept

Parametric FE analysis A systematic approach was adopted, so that each parameter was set to its minimum or maximum value maintaining all the other parameters fixed according to the

La Barbera et al.

Figure 2. Block moment arm (BMA) as a function of the spinal level: comparison between the values suggested by standard ISO 121891 and anatomical data from La Barbera and Villa.21 For the sake of comparison, the values recommended by ASTM F17172 are also reported. The lower horizontal axis shows the corresponding percentage increase in the von Mises stress on the screw compared to the reference configuration.

reference condition. The contribution of each parameter on the load on the device (screw and rod) was quantified in terms of von Mises stress increase normalized on the reference configuration (Figure 1(b)). A percentage variation of 2% was assumed to be significant,20 so that only those parameters overcoming this threshold are discussed here. The anatomical and overall worst-case conditions, respectively, combining the most influent anatomical parameters and also the mechanical ones were then investigated considering the worst-case condition previously found at L1.20

Results Anatomical parameters The values used in the standard for the anatomical parameters are generally within the physiological range of the thoracolumbar spine (Figure 3) or they are chosen on the upper (Figures 2 and 4) or beyond the lower limit (Figure 5). Since ISO 121891 was proposed as a development of ASTM F1717,2 it is based almost on the same anatomical parameters but CoFR and those which deal with the anterior support (ASA, SP1, SP2).

139

Figure 3. Centre of fixation to rotation (CoFR): comparison between the value suggested by standard ISO 121891 and anatomical data taken referring to the superior/inferior FSU (CoFRsup/CoFRinf, respectively) from La Barbera and Villa.21 For the sake of comparison, the values recommended by ASTM F17172 are also reported. The lower horizontal axis shows the corresponding percentage increase in the von Mises stress on the screw compared to the reference configuration.

The anatomical parameters playing a role in determining a significant percentage stress increase on the screw head are BMA, hAL and ASA (or SP1). Conversely, for the spinal rod BMA, ASA (or SP1, SP2), hAL and CoFR are the ones that play an important role in increasing the stress (Table 1). Among the anatomical parameters, the lever arm of the applied force (BMA) plays the most important role, with a maximum percentage stress increase of about 10% both on the screw head and on the rod (Figure 2). Despite the relationship between the stress and BMA found to be linear for the vertebrectomy model,20 here the relationship appears non-linear reaching the highest value on the superior limit of the variability range measured at L1. CoFR is the only anatomical parameter which differs from ASTM F17172 standard (Figure 3): its value is increased to 21 mm, in good agreement with our average measurements on the thoracolumbar spine obtained considering the inferior FSU (CoFRinf in Figure 3). Considering this shift towards higher values, the contribution of CoFR results to be significant only for the spinal rod (Table 1).

140

Figure 4. Half of the active length (hAL) as a function of the spinal level (top): comparison between the values suggested by standard ISO 121891 and anatomical data from La Barbera and Villa.21 For the sake of comparison, the values recommended by ASTM F17172 are also reported. The lower horizontal axis shows the corresponding percentage increase in the von Mises stress on the screw compared to the reference configuration.

hAL was found to be linearly related to the stress on the implant: reducing the active length of only 1 mm, the stress increases to 5.7% and 2.2%, respectively, on the screw and on the rod when compared to the reference configuration (Figure 4). SP on the sagittal plane (ASA or SP1 and SP2) has a significant effect on the stress arising in the implant (Figure 5). The maximum stress increase is achieved when the anterior support is shifted posteriorly, that is, when ASA or SP1 is reduced and SP2 is increased (Table 1).

Mechanical parameters The spring stiffness (k) lead to the highest percentage stress increase with respect to the reference configuration (Figure 6, Table 2). The stress after preload increases only slightly with the anterior support stiffness, however the values reached at peak load decrease of a much higher amount (Figure 6). Please note that the overall stiffness of the anterior support (or the stiffness of the unassembled construct KU according to ISO 12189)21 depends only on the stiffness of the anterior springs, which support the greatest amount of the applied vertical load.

Proc IMechE Part H: J Engineering in Medicine 230(2)

Figure 5. Anterior support arm (ASA) as a function of the spinal level (top): comparison between the values suggested by standard ISO 121891 and anatomical data from our patients’ database. The lower horizontal axis shows the corresponding percentage increase in the von Mises stress on the screw compared to the reference configuration.

Using the stiffest spring set (k = 459 N/mm), the preload stress increases to 2.3% on the screw with respect to the reference condition (rod: + 6.2%), while at peak load it decreases to 31.1% on the screw (rod: 224.5%). Conversely, using the softest spring set (k = 100 N/ mm), the preload stress decreases to 34% at the screw head (rod: 217.2%), while the stress achieved at peak load increases beyond 350%. The load shared by the posterior implant, defined as the axial force on the rod normalized by the applied force, depends on the vertical applied load, as well as on the spring stiffness. The initial high negative values predicted at low loads are due to precompression; then, as the vertical load increases, the shared load reaches an equilibrium value. When using the stiffest axial spring (k = 459 N/mm), the vertical applied load is unable to compensate the tensile preload and the load shared is 211.2%; conversely, using the softest one (k = 100 N/ mm), the axial load is reversed to compression and the load shared is 13.1%. As concern the precompression step (Pre), it significantly affects the initial stress on the spinal fixator and the stress values reached at peak load (Table 2). Avoiding precompression (Pre = 0 mm), the stress

La Barbera et al.

141

Figure 6. Percentage increase in the von Mises stress on the screw and on the rod for each possible combination of spring available from ISO 12189.1 Each colour corresponds to a specific spring stiffness (k), with green, blue, red and yellow representing 100, 147, 375 and 459 N/mm, respectively.2 Data are normalized to the values reached at 2000 N in the reference configuration (red springs). (Please refer to the on-line version of the document for a correct interpretation of colours).

value reached at 2000 N is beyond 50% higher than assuming a 1 mm precompression. The unsupported screw length (d0) plays a significant role, leading to a maximum increase of 6.0% on the screw when it is reduced to 0 mm (Table 2). It is important to note that, according to our reference configuration, d0 was set by default to 2 mm, as already suggested in ASTM F270630 standard: this assumption produces a percentage stress decrease of 6.0% on the screw (rod: 28.0%), when compared to the current version of ISO 121891 which may assume d0 equal to 0 mm.

Discussion Preclinical evaluation of any implant is a decisive step to ensure its mechanical reliability, to guarantee enough safety for any patient and to finally obtain the approval for clinical use. Some international bodies, such as the ISO and the ASTM, defined through the years some test methods useful to assess, evaluate and compare the mechanical behaviour of different posterior spinal stabilization devices. Obviously, the way an implant is tested (e.g. geometrical and mechanical set-up parameters, load level) detrimentally affects not only its mechanical behaviour but also the clinical outcome. For this reason, it is very important to critically assess, improve and update the current standards. In a previous comparative parametric study, the well-known vertebrectomy scenario implemented by ASTM F17172 has already been discussed by comparison with anatomical data, as well as by considering several mechanical parameters.20 However, such a worst-case scenario may not be indicated to test very flexible or dynamic implants and a more physiological anterior support model1 may be used instead. As far as we know, only few authors used such a standard,16–19 potentially suggesting that ISO 12189 standard is not so well known or not so consolidated.

In particular, La Barbera and colleagues16,17 compared the standard set-ups to more clinically relevant scenarios and proposed a procedure to validate the FE model of a spinal stabilization device assembled according to ISO standards.21 Therefore, the present parametric study critically assessed set-up parameters recommended by ISO 121891 standard, investigating their contribution on the stress arising on a posterior stabilization device.

Spring/anterior support stiffness The choice of the spring stiffness (k), that is, the anterior support stiffness, decisively affects the stress level of the tested implant (Figure 6) from a 225% to a + 350% with respect to the loading conditions described by ISO standard.1 It is unclear how the user should select the proper set of springs to use for the evaluation of a specific stabilization device according to its clinical indication. Despite ISO 12189 standard1 explicitly referring to a physiological lumbar disc,21 spinal stabilization is not always meant to treat a physiological FSU. In this light, it would be more meaningful to reproduce an ‘unphysiological’ anterior support, that is, degenerated or damaged (denucleated), but published data about the compressive behaviour of FSUs and IVDs are widespread over a huge range without consistent differences between healthy and degenerated specimens due to the high anatomical variability among specimens, the limited number of tests for each level, the diversity in the testing protocol (i.e. preload, strain rate, displacement/force control and level) as well as in the definition of stiffness (i.e. the compressive load–displacement curve is nonlinear).21,31,32 Moreover, since k is the most decisive parameter in determining the stress on the spinal implant, it is very important to guarantee that its values are as much as possible repeatable. In this light, the 10% tolerance in

142 the nominal axial stiffness allowed by ISO 102433 may lead to a significant variation in the stress values applied on the device ( + 15.2% and 211.6% at the screw head and between + 18.2% and 215.7% on the spinal rod with respect to the nominal value). Concerning the mechanical parameters other than the spring stiffness, the unsupported screw length (d0) plays a less significant role in determining the stress on the implant. The trend found for ISO 12189 configuration is different from the one already reported for ASTM F1717 model.20 Since the assumptions made to build up the FE model are essentially the same (i.e. same boundary conditions and constraints between components), it may be inferred that the stress acting at screw head may be influenced by a mechanism other than the lever arm of the applied force. For instance, the load shared by the implant may decrease with increasing d0 simply due to a decrease in the overall stiffness of the posterior implant: in a vertebrectomy model, the load is supported only by the posterior instrumentation, so the stress basically depends only on the lever arm.

Precompression and initial preload It has already been discussed that the initial precompression (Pre) induces a significant preload on the spinal rod of the posterior implant,16,21 consequently affecting the achieved final stress level. In the reference condition, we observed that neglecting the effect of precompression may lead to a decisive overestimation of the stress on the implant. In this study, we investigated all the possible combinations of the springs available from ISO 10243,3 demonstrating that the initial preload is related to spring stiffness. Using the stiffest spring set, the preload stress found on the implant increases with respect to the reference condition, while using the softest one it decreases much more significantly. The maximum stress increase at peak load can be found when the spring stiffness is minimized and the precompression step is avoided (Pre = 0 mm) and it is beyond 420%. Given this detrimental combined effect, it is reasonable to think that any uncertainty on the nominal characteristics of the springs (i.e. 10% variation on the axial stiffness, 61 mm on the free length), even if allowed by ISO 10243,3 may have a very significant influence on the effective stress value applied to the device. This problem may be faced on one hand by manufacturers, in reducing the tolerance in the characteristics of their springs, and on the other one by implant designers, who may perform some preliminary testing on each spring and discard those which exhibit geometrical features, as well as a mechanical behaviour far from the nominal value.

Anatomical parameters This study critically assesses the effect of several set-up parameters on the stress arising on a posterior spinal stabilization device. Among the anatomical parameters,

Proc IMechE Part H: J Engineering in Medicine 230(2)

Figure 7. ISO 12189 standard configuration seen as a lever system: depending on the value of each geometrical parameter, the scheme can represent (a) a type I and (b) a type III lever system. The red spring is used here to represent the centre of gravity of the spring system.

the relative position between the applied force (BMA) and the anterior support (ASA or SP1 and SP2) significantly contributes to the stress arising in the implant (Table 1). The non-linear relationship between BMA and the stress can be qualitatively explained considering that the loading configuration can be assimilated to a lever system (Figure 7). When 35.3 mm \ BMA 4 43 mm (type I lever), the fulcrum produced by the spring system (whose centre is ideally located at 31.7 mm anteriorly with respect to the screw insertion type) is set in between the applied load and the resistance (spinal rod); in this condition, the relation between BMA and the stress on the implant is linear with a relatively high slope (screw head: 13.4 MPa/mm; rod: 6.2 MPa/mm). When 15 mm \ BMA \ 35.3 mm (type III lever), the applied force is set in between the resistance and the fulcrum; in this unfavourable condition, the slope is about two times smaller. Increasing ASA, as well as SP1 or decreasing SP2, has the same effect: the anterior support is shifted anteriorly, thus changing the relative position of the fulcrum with respect to the applied load (shift from a type I lever towards a type III), finally increasing the load supported by the springs and reducing the stress on the implant. Similar observations were made by Polly et al.14 while investigating the effect of the position of an anterior cage in the antero-posterior direction. Given the measurements for BMA and ASA as a function of the spinal level (Figure 2 and 5), which were obtained from standing position X-rays, it is difficult to determine whether the load axis should pass anteriorly with respect to the position of the fulcrum of the anterior support. Nevertheless, reproducing simple standing would not be so meaningful during preclinical evaluation. The loading condition implemented in ISO 12189 standard1 assumes ASA \ BMA which is compatible with an anterior shift of the centre of mass of the upper body: this may mimic some more meaningful everyday life activity, such as anterior load carrying,

La Barbera et al. forward bending of the upper body or walking, which determines more severe loads on the fixator.33–35 Of course, the scenario is complicated by the initial preload due to spring precompression. This contribution can be seen as an eccentric upward load (FPre) acting approximately in the centre of the spring system, thus inducing a tensional load and an extension on the spinal fixator upon release. During testing, the applied eccentric vertical force produces a compressive axial load and a flexion on the implant, so that the final configuration of internal loads/stress field within the spinal implant is determined by the relative length of these lever arms (BMA vs ASA or SP1, SP2), as well as by the intensity of these contributions (applied vertical load F vs Pre). The anatomical parameters play a secondary role with respect to the mechanical ones, even when they are combined together. In fact, even considering the worst-case values for the most important anatomical parameters (BMA = 43 mm, CoFRinf = 25.9 mm, hAL = 37 mm), a maximum stress increase of 22.4% at the screw head and 8.6% on the spinal rod can be reached.

Limitations The main limitations of this study are the simplified geometry design used for the spinal stabilization device, the assumed linear elastic material properties, the tieconstraints at screw–block/screw–rod interfaces as well as the systematic approach used for the sensitivity analysis. The implications of these approaches have already been discussed.20 Despite the comparative aim of this study, the numerical approach used here gave satisfactory results when compared with the experimental data in terms of stiffness of the unassembled and assembled constructs and strains on the implant.21 Obviously, an experimental investigation would be crucial in confirming whether a variation in the set-up parameters may lead to an effective reduction in the fatigue life of a commercial device.

Conclusion Standard ISO 12189 was developed on the basis of ASTM F1717; therefore, the set of values that it proposes represents quite well an average two-level instrumented construct. The specific set-up implemented by ISO standard represents a step forward towards the development of a test method which is closer to the effective clinical use of spinal stabilization devices. The great improvement over ASTM vertebrectomy model is represented by the addition of a modular anterior support based on calibrated springs. Our study sheds light in better understanding which are the most important parameters affecting the stress arising on a posterior implant tested according to standard ISO 12189. Any user/designer should be aware of

143 their effect when using the current standard for the preclinical evaluation of posterior spinal stabilization devices. The parametric comparative analysis demonstrates a significant maximum increase in the stress on the device, compared to the standard currently in use. The anterior support stiffness plays the most detrimental effect ( . 359%). Second, the initial precompression step was confirmed to have an important role in determining the final stress values achieved at peak load ( . 51%). Moreover, when combining the effect of spring stiffness with the initial precompression, stress variations well beyond 427% may be reached. The other anatomical parameters play a secondary role even when they are combined together in the anatomical worst case obtained assuming BMA = 43 mm, CoFRinf = 25.9 mm and hAL = 37 mm (screw: + 22.4%, rod: + 8.6%). Acknowledgements The authors gratefully acknowledge Fabio Galbusera (IRCCS Istituto Ortopedico Galeazzi (Italy)) and Hans-Joachim Wilke (Institute of Orthopaedic Research and Biomechanics, Centre of Musculoskeletal Research Ulm, Ulm University, Germany) for their interest in discussing the implications of this work. Declaration of conflicting interests The author(s) declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article. Funding The author(s) received no financial support for the research, authorship and/or publication of this article. References 1. ISO 12189:2008. Implants for surgery – mechanical testing of implantable spinal devices: fatigue test method for spinal implant assemblies using an anterior support. 2. ASTM F1717:2015. Standard test methods for spinal implant constructs in a vertebrectomy model. 3. ISO 10243:2010. Tools for pressing – compression springs with rectangular section: housing dimensions and colour coding. 4. Jahng T, Kim Y and Moon K. Comparison of the biomechanical effect of pedicle-based dynamic stabilization: a study using finite element analysis. Spine J 2013; 13: 85–94. 5. Gornet M, Chan F, Coleman J, et al. Biomechanical assessment of a PEEK rod system for semi-rigid fixation of lumbar fusion constructs. J Biomech Eng 2011; 133(8): 081009. 6. Ponnappan RK, Serhan H, Zarda B, et al. Biomechanical evaluation and comparison of polyetheretherketone rod system to traditional titanium rod fixation. Spine J 2009; 9(3): 263–267.

144 7. Schmidt H, Heuer F and Wilke H. Which axial and bending stiffnesses of posterior implants are required to design a flexible lumbar stabilization system? J Biomech 2009; 42(1): 48–54. 8. Qi L, Li M, Zhang S, et al. Comparative effectiveness of PEEK rods versus titanium alloy rods in lumbar fusion: a preliminary report. Acta Neurochir 2013; 155: 1187– 1193. 9. Campbell J, Imsdahl S and Ching R. Evaluation of a synthetic L2-L5 spine model for biomechanical testing. In: Proceedings of the 17th meeting of the Canadian society for biomechanics/Socie´te´ Canadienne de Biome´canique (CSB/SCB), Vancouver, BC, 6–9 June 2012. Available at: http://ocs.sfu.ca/csb-scb/index.php/csb-scb/2012/ paper/viewFile/306/222 10. Gioia G, Scotti C, Mandelli D, et al. Posterior spinal instrumentation: biomechanical study on the role of rods on hardware response to axial load. Eur Spine J 2011; 20(Suppl. 1): S3–S7. 11. Braithwaite G and McCormack C. Towards an artificial functional spinal unit for long-term in vitro testing. In: Proceedings of the 56th annual meeting of the orthopaedic research society, New Orleans, LA, 6–9 March 2010, paper no. 1538. Available at: http://www.ors.org/Transactions/56/1538.pdf 12. Quaglini V and Villa T. Biomechanical evaluation of spinal fixators. In: Proceedings of the 19th European conference on biomaterials, Sorrento, Italy, IT, 11–15 September 2005. 13. Haher T, Yeung A, Ottaviano D, et al. The inverse effects of load transfer and load sharing on axial compressive stiffness. Spine J 2001; 1(5): 324–329. 14. Polly D, Klemme W, Cunningham B, et al. The biomechanical significance of anterior column support in a simulated single-level spinal fusion. J Spinal Disord 2002; 13(1): 58–62. 15. Wilke H, Russo G, Schmitt H, et al. A mechanical model of human spinal motion segments. Biomed Tech 1997; 42(11): 327–331. 16. La Barbera L, Ottardi C and Villa T. Comparative analysis of international standards for the fatigue testing of posterior spinal fixation systems: the importance of preload in ISO 12189. Spine J 2015; 15(10): 2290–2296. 17. Villa T, La Barbera L and Galbusera F. Comparative analysis of international standards for the fatigue testing of posterior spinal fixation systems. Spine J 2014; 14(4): 695–704. 18. Hocquard L. E´valuation clinique, caracte´risation me´canique et mode´lisation pour l’e´volution de la conception d’un implant rachidien dynamique. PhD Thesis, Department of Sciences and Technologies, Universite´ Bordeaux 1, Bordeaux, 2012. 19. Mosnier T. Contribution a` l’analyse biome´canique et a` l’e´valuation des implants rachidiens. PhD Thesis, E´cole Nationale Supe´rieure d’Arts et Me´tiers, Paris, 2008. 20. La Barbera L, Galbusera F, Villa T, et al. ASTM F1717 standard for the preclinical evaluation of posterior spinal

Proc IMechE Part H: J Engineering in Medicine 230(2)

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

fixators: can we improve it? Proc IMechE, Part H: J Engineering in Medicine 2014; 228(10): 1014–1026. La Barbera L and Villa T. ISO 12189 standard for the preclinical evaluation of posterior spinal stabilization devices - I: Assembly procedure and validation. Proc IMechE, Part H: J Engineering in Medicine 2015; 230(2): 122–133. Han K. Biomechanical roles of spinal muscles in stabilizing the lumbar spine via follower load mechanism. PhD Thesis, The University of Iowa, Iowa City, IA, 2008. Rohlmann A, Zander T, Rao M, et al. Applying a follower load delivers realistic results for simulating standing. J Biomech 2009; 42(10): 1520–1526. Schmidt H, Heuer F, Claes L, et al. The relation between the instantaneous center of rotation and facet joint forces – a finite element analysis. Clin Biomech 2008; 23(3): 270–278. Rohlmann A, Bauer L, Zander T, et al. Determination of trunk muscle forces for flexion and extension by using a validated finite element model of the lumbar spine and measured in vivo data. J Biomech 2006; 39(6): 981–989. Patwardhan A, Havey R, Meade K, et al. A follower load increases the load-carrying capacity of the lumbar spine in compression. Spine 1999; 24(10): 1003–1009. Panjabi MM, Takata K, Goel V, et al. Thoracic human vertebrae. Quantitative three-dimensional anatomy. Spine 1991; 16(8): 888–901. Panjabi MM, Goel V, Oxland T, et al. Human lumbar vertebrae. Quantitative three-dimensional anatomy. Spine 1992; 17(3): 299–306. Semaan I, Skalli W, Veron S, et al. Quantitative 3D anatomy of the lumbar spine. Rev Chir Orthop Reparatrice Appar Mot 2001; 87: 340–353. ASTM F2706:2008. Standard test methods for occipitalcervical and occipital-cervical-thoracic spinal implant constructs in a vertebrectomy model. Kemper A, McNally C, Manoogian S, et al. Stiffness properties of human lumbar intervertebral discs in compression and the influence of strain rate. Washington, DC: National Highway Traffic Safety Administration (NHTSA), 2013. Gardner-Morse M and Stokes I. Physiological axial compressive preloads increase motion segment stiffness, linearity and hysteresis in all six degrees of freedom for small displacements about the neutral posture. J Orthop Res 2003; 21: 547–552. Rohlmann A, Claes L, Graichen F, et al. Comparison of intradiscal pressures and spinal fixator loads for different body positions and exercises. Ergonomics 2001; 44(8): 781–794. Rohlmann A, Graichen F and Bergmann G. Influence of load carrying on loads in internal spinal fixators. J Biomech 2000; 33(23): 1099–1104. Rohlmann A, Claes L, Graichen F, et al. Loads on an internal spinal fixation device during walking. J Biomech 1997; 30(1): 41–47.