INTRODUCTION
Spine interbody cages provide anterior column support and correct spinopelvic sagittal alignment while concurrently increasing construct stability and maximizing fusion potential. Failure of fusion (nonunion) may result in persistent pain and implant subsidence, loosening and stress shielding, often necessitating a reoperation [1,2]. Earlier generations of interbody devices possessed native, unprocessed surfaces. Although structurally sound for load carrying, their smooth surfaces offered poor topography for osseointegration [2].
Newer generations of implants improve osseointegration through roughened or porous surfaces that allow bone on-growth, in-growth and through-growth and resist shear through mechanical interlocking of bone and implant surfaces [3-5]. Microscale roughness has been shown to increase osteoblastic differentiation and increase in vivo bone formation. Olivares-Navarrete et al. found the optimal microtextured titanium alloy surface that most promoted osteoblastic differentiation to possess an average roughness between 2–3 μm with moderate peak height, low kurtosis, and slight skewness [5]. In addition, greater roughness increased surface friction coefficient, potentially reducing cage migration [6].
However, raised microscale features that enhance osseointegration may be vulnerable to abrasion or delamination under shear loads experienced during device impaction and positioning [3,7,8]. Although coated devices are at higher risk, compromise of ingrowth surfaces during impaction is relevant to all raised microscale surface designs as osteoblastic differentiation is sensitive to specific morphologic features of implant surface [5]. Surface damage that alters these microscale features may lessen effective osseointegration of the implant. Torstrick et al. [7] found that average surface roughness decreased by 0.45 μm for titanium-coated devices after impaction, which may place surface roughness outside of optimal range for osteoblastic differentiation. Reduced surface roughness may decrease the friction coefficient of cage migration, which Ungurean et al. [9] found to be a cause in 26.2% of cage failures. Further damage may cause collapse of surface structure reducing volume available for ingrowth and loss of the microscale surface potentially leading to pseudoarthrosis. Particulate debris generated by surface abrasion may also cause inflammation, leading to persistent pain, bone resorption and loosening [10-13].
The goal of this study was to investigate impaction durability of 3D-printed titanium interbody fusion cages. We aimed to quantify mass loss and changes in surface roughness due to surgical impaction and identify changes in microscale features and regional changes in roughness. We hypothesized that impacted interbody cages would exhibit mass loss and decreases in roughness parameters; cages impacted under greater preload would experience greater mass loss and decrease in roughness; and the anterior region of cages would experience greater damage.
MATERIALS AND METHODS
1. Custom Lumbar Modular Interbody Fusion Cages for Testing
In preparation, pilot analyses were performed using an U.S. Food and Drug Administration-approved system (SCARLET AL-T, Spineart, Geneva, Switzerland). Roughness and dimensions of custom cages for this study were approximated to those of Spineart cages from pilot testing (roughness of Spineart cages was of the nonporous endplate portion). Custom cages specifications aimed to simulate results generalizable to multiple implants and designs rather than one specific proprietary design.
Four custom modular interbody fusion cages with 8 detachable endplates (n=8) were tested (Fig. 1). Superior and inferior endplates were attached to a body, forming a modular cage construct (Fig. 2). The 30-mm-width, 22-mm-length, 14-mm-height, 10°-lordotic-angle cages were printed by Protolabs (Morrisville, NC, USA) from titanium alloy (Ti6Al4V) using direct metal laser sintering (DMLS) at 20-μm resolution, followed by hot isostatic pressing treatment from Bodycote (Camas, WA, USA). The DMLS printing parameters were as follows: layer thickness: 20-μ minimum; feature height: 0.153 mm; Material: vacuum stress relieved Ti6Al4V with ultimate tensile strength of 1,055 Mpa; yield strength of 951 Mpa; elongation at yield of 15%; and hardness of 35 Hardness Rockwell C.
2. Synthetic Bone Models
Cages were impacted into flexible articulated disc spaces between vertebrae of L4–5 foam cortical shell models (SKU:1352-10, Sawbones, Vashon Island, WA, USA). Anterior longitudinal ligaments were removed, and discectomies were performed to simulate anterior lumbar interbody fusion surgery. Inserted cages contacted only the 17–22 pound-per-cubic-foot (PCF) foam cortical endplate of the model rather than the 7-PCF inner cancellous foam. Compared to 40-PCF foam blocks used in previous studies, this model more accurately simulated anatomy in vivo while approximating shear modulus of human lumbar vertebral endplates [3,7,8]. Shear modulus is 130 MPa for 40-PCF foam, 49 MPa for the cortical foam and 34.5–126.1 MPa for human lumbar endplates. Moreover, Young modulus of the cortical foam (210 MPa) leaned toward values for osteopenic bone (198.3 MPa) [14,15]. The posterior aspect of the spine unit was cast in Smooth-Cast 300 plastic (Smooth-On, Macungie, PA, USA). Potted plastic was split at disc-level to reproduce intervertebral motion with intact posterior longitudinal ligaments. During casting, a temporary spacer placed between the compressed vertebrae kept anterior opening dimensions consistent.
3. Impaction Testing Setup
The testing method in this study was modified from Kienle et al. [3]. Cages were impacted under axial preload. Potted L4–5 levels were secured in the apparatus (Fig. 3A). Using low-friction slide and hanging weights, the apparatus applied a constant axial load of either 200N or 400N to the spinal segment. Axial load was applied through the L4–5 center of rotation at 84.4% of the disc length from anterior edge (5 mm anteriorly from posterior endplate border) [16].
4. Impaction Testing
A 1-lb (0.45 kg) drop weight (LW Measurements LLC, Rohnert Park, CA, USA) representing a standard 1-lb surgical hammer was used to insert cages into disc spaces. The drop weight traveled inside a guide tube aligned in anteroposterior direction of the spinal level. The lower end of the tube was connected to a stainless-steel cage inserter while the upper end housed an initial weight stopper. From release up to contact with the inserter, the drop weight traveled 34.5 cm, with maximum speed at impact of 2.6 m/sec [3,17]. Cages were inserted 2 mm past the anterior border of the vertebral endplates (Fig. 3C).
5. Material Handling
Prior to pre- and postimpaction weighing and roughness measurements, cages were cleaned in ultrasonic bath at 30°C for 10 minutes with E62 enzymatic detergent (Stryker, Kalamazoo, MI, USA), rinsed with distilled water, and dried with silica-gel desiccant for 20 hours.
6. Measurements and Analyses
Mass of each endplate was determined using AS 60/220.R2 Plus analytical balance (Radwag USA, Miami, FL, USA). Endplate surfaces were analyzed using SJ-210 portable surface roughness tester/profilometer (Mitutoyo America Corporation, Aurora, IL, USA). The SJ-210 profilometer stylus had a 60° tip angle, a 2 μm radius, and diamond stylus tracer. Roughnesses were measured at 9 locations on each endplate surface. Measuring trajectories of the profilometer probe are depicted in Fig. 4A. Fig. 4B illustrates how roughness profile (R-profile) was obtained from the measured profile.
Roughness parameters measured in this study included mean roughness (Ra), root mean square roughness (Rq), mean roughness depth (Rz), total height of roughness profile (Rt), skewness (Rsk), and kurtosis (Rku). Skewnesses and kurtosis respectively describe the asymmetry and the peakedness/sharpness (presence of outliers) of the profile [18]. An illustration of R-profiles, R-profile height distributions, and roughness parameters are shown in Fig. 5.
7. Data Analysis
Outlier detection using median absolute deviation method was performed on collected data. Roughness parameters were compared globally and regionally. For global comparison, an overall average roughness parameter of each endplate was taken from all 9 measurements. Pre- and postimpaction roughness parameters were compared for differences. Postimpaction differences were compared between superior and inferior plates and between 200-N and 400-N preloads. For regional comparison, a regional average roughness parameter was taken from 3 measurements for each region (Fig. 5). Statistical significance was determined using the Student t-test with 0.05 level of significance. Data are presented as mean±standard deviation.
8. Scanning Electron Microscope Imaging
Before and after images of the endplate surface were taken using TM4000Plus scanning electron microscope (Hitachi High-Tech America, Inc., Pleasanton, CA, USA) at × 50, × 150, and × 300 magnifications and 15-kV accelerating voltage for qualitative assessment of topographical changes using similar cleaning and testing protocol.
RESULTS
Prior to testing, modular endplates weighed 7.90±0.03 g and had the following average values of roughness: Ra=9.16±1.60 μm, Rq=11.42±1.86 μm, Rz=46.64±6.95 μm, Rt=74.98±10.07 μm, Rsk =-0.64±0.13, Rku = 3.31±0.37. No outlier was detected using absolute deviation method. For both 200-N and 400-N axial preloads, it took 2–4 impacts to fully insert cages (Table 1).
1. Mass Comparison
Overall, there was a significant decrease in mass after impaction testing for pooled data (-0.10±0.08 mg, p=0.004) (Table 2, panel B). Comparing pre- and postimpaction mass for separate endplate groups showed significant decreases for superior-endplate group, 200-N-preload group, and 400-N-preload group. While the 200-N-preload and the 400-Npreload produced the smallest and largest mass loss, there were no significant differences between the groups (Fig. 6). Scanning electron images showed evidence of both mass loss and postimpaction topographic changes (Fig. 7).
2. Overall Roughness Comparison
For pooled data comparison, significant postimpaction decreases were observed for Ra, Rq, Rz, and Rt while no significant differences between before and after observed for Rsk and Rku (Table 2 panel B) (Fig. 8). The superior-endplate group showed significant decreases in Ra, Rt, and Rsk, while the inferior-endplate group had significant decreases in Ra, Rq, and Rz. For the 200-N-preload group, Ra and Rq reduced significantly and for the 400-N-preload group, decreases in Rz and Rsk were significant. There were no significant differences when comparing postimpaction changes between superior and inferior groups as well as between 200-N-preload and 400-N-preload groups.
3. Regional Roughness Comparison
There were significant changes in all measured roughness parameters, except for Rt in the anterior region after impaction (Table 3) (Fig. 9). In the middle region, there were significant decreases in Rz and Rt. In the posterior region, there were significant decreases in Ra, and Rq. In the center region, there were significant decreases in Rz, and Rt.
DISCUSSION
There are several additive manufacturing techniques used in implant fabrication that differ in material deposition and energy sources (Fig. 10) [19,20], with selective laser sintering and DMLS (as used in this study) being the most common to fabricate titanium implants [21,22]. Bulk and surface porosity is often introduced to 3D-printed titanium implants to increase osseointegration and prevent stress shielding [21]. However, in this study, using dense rather than porous 3D-printed cages allowed analyzing endplate surfaces by means of surface roughness, a property that is not present in porous surfaces. The initial surface finish of our cages was grainy and skewed negatively, with sharp valleys and round peaks. This was confirmed by a leptokurtic baseline surface and Rq> Ra, indicating a surface with high peaks and deep valleys [23].
These raised microscale features on cage surface may be vulnerable to abrasion or delamination under shear loads experienced during impaction [3,7,8]. This may place implant surfaces outside of the optimal range for osteoblastic differentiation, reducing subsequent ingrowth and decreasing the effective osseointegration. Olivares-Navarrete et al. [5] found the optimal microtextured titanium alloy surface that promoted the highest osteoblastic differentiation to possess an average roughness between 2 and 3 μm, and with specific morphologic features of moderate peak height, low kurtosis, and slight skewness. The pooled date from this study showed a difference in the arithmetic mean roughness of 0.71±0.42 µm. This difference was higher for the inferior plate specifically as well as impaction under 400-N preload. Thus, impaction could place the average roughness of the implant outside of the optimal 2 to 3 μm range that most promotes osteoblastic differentiation. Cage migration may subsequently occur via 2 mechanisms; in the acute period after implantation lower surface roughness may reduce the friction coefficient causing downstream cage migration, and in the long term a lack of fusion predisposes to migration and subsidence. For transforaminal lumbar interbody fusion, Park et al. [24] found a fusion rate of 97.1% for no cage migration, a fusion rate of 55.0% for cage migration without subsidence, and 41.7% cage migration with subsidence. In addition, reduced surface volume available for ingrowth may potentially lead to pseudoarthrosis. Lastly, particulate debris generated by surface abrasion may lead to inflammation, possible metallosis-related osteolysis and lymphocytic infiltrates, similar to metal-on-metal total joint arthroplasty [10-13].
Comparing variations in preload, 400N force tended to cause a more significant alteration in surface features. The values of 200N and 400N were chosen based on previous consensus in the literature. Previous studies utilized 200N, as this lies within the range for physiological preload in vivo in a relaxed lying position (approximately 140–240N) [3,7,17]. As anterior lumbar interbody cages are commonly impacted in the supine position, this value mirrors the intraoperative force experienced during impaction. Similarly, 390N lies between physiological preload values and the American Society for Testing and Materials standard for lumbar spine implants of 500N [3,8]. In brief, 400N is the axial preload condition for worst-case testing, while 200N replicates more physiologic conditions.
Mass loss and microscale parameter changes were not equivalent for superior and inferior endplates. As the implant is designed with flat modular superior and inferior endplates, this variance originates from vertebral endplates’ morphologic asymmetry. Natively in the lumbar spine, superior endplates tend to be more concave with an average depth of concavity of 1.5 mm, while inferior endplates are flatter [25]. Thus, the implant’s inferior endplate will experience more distributed wear patterns, resulting in consistent reduction in mean roughness parameters Ra, Rz, and Rq. In comparison, the superior endplate will experience a more concentrated wear and stress pattern, resulting in large reductions in measures such as mass, and Rt, and variance measurements such as Rsk. In addition, the superior vertebral endplate is approximately 40% weaker than the inferior endplate, and subsequent subsidence of implants occurs more often at the superior endplate [25,26]. Thus effective osseointegration is especially critical at the superior endplate and future implant design may compensate for this concentrated wear pattern.
Analysis of regional postimpaction changes in roughness parameters showed the anterior aspect underwent significant decreases in all parameters, except for total roughness depth (not statistically significant) and kurtosis (increased after impact). The anterior surface experienced the highest change in parameters during impaction. Several aspects of the anterior surface are crucial for successfully cage implantation and osseointegration. Foremost it is the widest part of the cage in a lordotic implant and largely responsible for load bearing. This is likely responsible for the increased wear seen at this aspect of the cage, and a more neutral implant without lordosis would likely experience a more uniform loss of features and mass. However, the corrective ability of an anterior interbody fusion stems from this ability to implant a cage with a higher degree of lordosis than that of other approaches. In addition, the anterior aspect of the cage imparts significant stability in flexion, extension, bending, and rotation. Furthermore, the anterior region of the endplate is weaker than the posterior region, highlighting the need for a robust implant at this aspect. The posterior aspect experienced less significant decreases in Ra and Rq. Middle and center aspects experienced significantly decreased Rz and total roughness depth, but not a significant change in Ra. Having more domed endplates would likely result in more material and feature loss along the periphery, where the contact with endplates is greater.
Mass loss and wear on insertion in this model are of ‘best-case scenario’: only 2–4 strikes without repositioning to fully insert solid (instead of porous) cages. Multiple factors suggest that in vivo wear could potentially be significantly higher than in a benchtop study. In vivo, soft tissue anatomy may preclude optimal impaction angle; cages may undergo multiple hits to achieve satisfactory placement, and potentially be repositioned. In the technique video guide by Duan et al. [27] over twenty strikes are used to position the cage, and although likely of a softer impaction force, this is far more than are used in this study. Subtle repositioning is also seen in the technique video, which can apply a lateral or rotational shear force in addition to the anteroposterior shear force applied in this study. As seen in the technique video the endplates are commonly prepared using a combination of curettes and rasps, which can produce incomplete or uneven endplate preparations that can potentially aggravate wear at the affected areas. Furthermore, although in this study cages were inserted 2 mm past the anterior border of the vertebral endplates, cages in vivo can vary in their placement. Those placed farther than 2 mm likely require more impactions for placement resulting in increased wear. Lastly, biolubricants may induce higher wear rate and additional wear mechanism compared to dry sliding. Wang et al. [28] found dry sliding to have a lower coefficient of friction than human simulated body fluids in a bovine cortical bone model. By evaluating mass loss under the most optimal conditions, we can postulate that mass loss could be significantly greater in vivo and therefore of importance to consider in future research and design for commercial spine cages.
There are significant differences in the postimpaction analysis of titanium implants in this study and titanium-coated polyether ether ketone (PEEK) cages in prior literature. Both Torstrick et al. [7] and Kienle et al. [3] demonstrated mass loss and changes in roughness parameters of titanium-coated PEEK cages following impaction. Titanium cages in this study lost less mass compared to titanium-coated PEEK cages. Torstrick et al. [7] showed titanium-coated PEEK interbody cages lost 4.6 mg after impaction under 200N preload. In comparison, mass loss in this study at 200N was only 0.06 mg, indicating less implant surface loss and decreased particle dispersion. Less particulate debris should minimize inflammatory effects on surrounding tissue responsible for bone resorption and aseptic loosening [10-13]. An analysis by Torstrick et al. [7] reported an approximately 0.45% change in average surface roughness between pre- and postimpaction. In our results, we observed a 0.06% change in average surface roughness for the titanium implants under 200N preload. 3D-printed titanium interbody fusion cages appear superior in resisting shear forces of impaction as compared to titanium-coated PEEK interbody cages as reported in the literature.
Future implant design may benefit from redistribution and reinforcement of surface topography. Quantified knowledge of mass loss and surface parameter changes allows for cages to be manufactured to account for this future loss. The pooled date from this study showed a difference in the arithmetic mean roughness of 0.71±0.42 µm at the anterior aspect of the cage. Knowing this the original roughness of the manufactured cage could be bolstered by this amount so that after implantation the remaining value falls between the optimal 2 to 3 μm range. This could be applied to all of the surface parameters known to promote fusion, such as specific morphologic features of moderate peak height, low kurtosis, and slight skewness. Future implants would benefit from greater roughness at the posterior aspect, and increased roughness depth at the center to account for additional wear during impaction. In addition, as compared to the literature, titanium-coated cages appear weaker to the forces of impaction than titanium 3D-printed cages. By extension, mass loss and change in surface topography may be significant to consider in any implant where surface coating or topography is of particular importance, such as cell-based or drug-eluting surfaces, or surfaces with anti-infective coatings. Ultimately, we recommend in vivo cadaver testing of mass loss and change in surface topography prior to any regulatory approval of interbody cages.
Our study has limitations. First, different wear patterns may be possible with various insertion techniques (using a “squidtype” inserter or skids) or implant designs (e.g., hyperlordotic cages) [3]. Second, to keep a consistent testing environment, human cadaveric vertebral bodies were not used, as their material properties are not uniform, and vary depending on host factors and stages of degeneration. The consistent shape and density offered by synthetic vertebrae allowed for consistent wear and greater generalizability of results. Third, our cages were created using DMLS and results may not be generalizable to other 3D-printing processes. Lastly, roughness was measured using conventional stylus roughness gauge method, rather than laser scanning. Adjustment of laser wavelength may be required, as reflectivity of Ti6Al4V depends on surface roughness and laser wavelength making laser scanning more difficult to implement. Measuring surface roughness at different locations compensated for the lack of planar roughness data.
CONCLUSION
Titanium 3D-printed cages lose mass and surface roughness with impaction. Loss of roughness parameters was not uniform between the implant endplates, and across the endplate surface regions, which may require additional emphasis in future implant designs. Mass and roughness changes are less compared to those reported for titanium-coated PEEK indicating increased resistance to impaction. The testing of impaction durability and resistance to surgical stresses of 3D-printed implants should be standardized and incorporated into pre-approval testing.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Print
