**2.1 PEEK cervical total disc replacement device**

The cervical arthroplasty device incorporates a unique, Ti cam blade fixation system that can be described as a rotating shaft with blades for primary fixation (Figure 1). Given that PEEK is a relatively bio-inert material, bone apposition does not readily occur onto the material. Therefore, a plasma-sprayed hydroxyapatite (HA) coating was added to the outer endplates. The design objectives are to achieve secure, long term fixation within the disc space, exhibit the necessary strength and durability for the lifetime of the patient for its intended use, and restore or maintain the range of motion (ROM) at the operative level while simultaneously not adversely affect the biomechanics at the adjacent levels.

Fig. 1. PEEK cervical arthroplasty device.

The Use of PEEK in Spine Arthroplasty 215

devices reached 10 million cycles without experiencing any visible plastic deformation or fracture. For the coated devices, since the shear static loading was significantly higher than the acceptance criteria and the coating was shown to not have a detrimental effect on the axial static properties of the device, only shear fatigue testing was performed. The results were the same as the coated devices, 1600 N of compressive shear to 10 million cycles without experiencing any visible plastic deformation or fracture. Therefore, it was concluded that shear fatigue loads can be withstood far beyond those that would be

Fig. 2. PEEK cervical arthroplasty device after static compressive shear testing to failure.

The fixation strength of the device was assessed by determining its pull out strength using a synthetic vertebral model. The largest (17x17 mm) and smallest footprints (12x14 mm) of 8 mm height were utilized. To avoid the variability produced by human cadaveric tissue, including bone mineral density and endplate geometries, a synthetic fixation medium of ASTM grade 15 polyurethane foam was selected for use. The fixation pull out test model consisted of compressing the device between two polyurethane blocks to 60 lbs, and then performing a uniaxial tensile load to failure test, thereby forcibly removing the implant. The test was conducted on an MTS 858 Mini-Bionix II test frame with appropriate fixturing. The displacement rate was 2.5 mm/sec. The implant was placed between the blocks in accordance with the surgical technique. A minimum of ten pullout tests were performed for each implant size. This data was then compared to eight previously tested cervical disc arthroplasty devices used clinically in cervical reconstruction that utilize press fit, keel or screw fixation using the same test protocol and laboratory [Cunningham, 2010]. Therefore, the results of this testing provides for a standardized testing methodology to quantify the

In all pullout tests for the two implant sizes, there was no incidence of intrinsic device failure. Failure always occurred at the polyurethane block-implant interface. The cam blade fixation indicated the highest pullout strength, which was statistically different from nearly all other designs (Table 1). Prosthetic endplates containing toothed ridges and keel fixation ranked next in terms of fixation strength. Although intuition favored the coronal keel as affording an increased mechanical advantage over the toothed surface, there were no

Left – tested under 10° flexion, right – tested under 10° extension.

strength of fixation and resistance to migration.

expected *in vivo*.

**2.1.3 Fixation strength** 

### **2.1.1 Axial static and fatigue strength**

To determine the static and fatigue strength of the device, the smallest footprint (12x14 mm) and height (5 mm) was utilized. Testing was performed on an MTS 858 Mini-Bionix II test frame with the devices mounted on steel fixtures. A comparison was made between HA coated devices and uncoated devices to determine if the coating had an adverse effect on the static and fatigue strength of the device. A sample size of 6 was used for the static test with a displacement rate of 2 mm/min. The acceptance criteria was an offset yield load of greater than 1200 N for the axial static testing. For the fatigue tests, a sample size of 2 was used with a starting fatigue load of 75% of the static offset yield load of the uncoated devices, with R=10 at 10 Hz.

For the axial static testing, the results showed that the mean load at offset yield was calculated to be 7,631 ± 171 N at a displacement of 0.50 ± 0.01 mm for the uncoated devices and 9640 ± 251 N at a displacement of 0.70 ± 0.02 mm for the coated devices, which was significantly greater (t-test, p<0.001). The primary mode of failure in the tested constructs was fracture of the PEEK. For the axial fatigue testing, the uncoated and coated samples reached 10 million cycles at a load of 5723 N (R=10). No evidence of failure, such as fatigue cracks or gross deformation, was seen during the inspections or after test. When the coated devices were tested at a load of 7230 N, implant fracture occurred at 867,966 cycles.

The literature suggests that the maximum compressive load on the cervical spine during the performance of physical tasks is 1.2 kN [Moroney, 1988] and that the elastic limit of single motion segments could be as high as 1.23 kN, with a predicted ultimate compressive strength of 2.40 kN [Przybyla, 2007]. The results showed that the loads reached were significantly above the expected *in vivo* loads of the cervical spine, and that the HA coating did not negatively affect the axial compressive static or fatigue performance.

### **2.1.2 Shear static and fatigue strength**

To determine the shear static and fatigue strength of the device, the smallest footprint (12x14 mm) and height (5 mm) was utilized. Testing was performed on an MTS 858 Mini-Bionix II test frame with the devices mounted on steel fixtures. According to the literature, the total ROM for flexion/extension of the cervical spine is greatest at the C4-C5 level, and appears to be slightly more than 20° when measured *in viv*o using computer assisted tracking software [Reitman, 2004]. The literature also shows that the amount of mean flexion is only slightly higher than the mean extension [Panjabi, 2001]. Therefore, the shear test was conducted at 10° for both flexion and extension. A comparison was made between HA coated devices and uncoated devices to determine if the coating had an adverse effect on the shear static and fatigue strength of the device. A sample size of 6 was used for the static tests with a displacement rate of 2 mm/min. The acceptance criteria was an offset yield load of greater than 135 N [Moroney, 1988] for the shear static testing and for the fatigue tests, a sample size of 2 was used with a run out load greater than 135 N [Moroney, 1988] with R=10 at 10 Hz. For the shear static testing at 10° flexion, the mean load at offset yield was calculated to be

4774 ± 372 N at a displacement of 0.44 ± 0.01 mm. For the shear static testing at 10° extension static shear, the mean load at offset yield was calculated to be 6788 ± 412 N at a displacement of 0.53 ± 0.02 mm. The primary mode of failure was fracture of the PEEK (Figure 2). Since the results of the static compressive shear strength resulted in a lower offset yield load in flexion (4744 N) versus extension (6788 N), the fatigue test was run at 10° flexion as a worst case scenario. The results of the fatigue testing at 1600 N showed the

To determine the static and fatigue strength of the device, the smallest footprint (12x14 mm) and height (5 mm) was utilized. Testing was performed on an MTS 858 Mini-Bionix II test frame with the devices mounted on steel fixtures. A comparison was made between HA coated devices and uncoated devices to determine if the coating had an adverse effect on the static and fatigue strength of the device. A sample size of 6 was used for the static test with a displacement rate of 2 mm/min. The acceptance criteria was an offset yield load of greater than 1200 N for the axial static testing. For the fatigue tests, a sample size of 2 was used with a starting fatigue load of 75% of the static offset yield load of the uncoated devices, with

For the axial static testing, the results showed that the mean load at offset yield was calculated to be 7,631 ± 171 N at a displacement of 0.50 ± 0.01 mm for the uncoated devices and 9640 ± 251 N at a displacement of 0.70 ± 0.02 mm for the coated devices, which was significantly greater (t-test, p<0.001). The primary mode of failure in the tested constructs was fracture of the PEEK. For the axial fatigue testing, the uncoated and coated samples reached 10 million cycles at a load of 5723 N (R=10). No evidence of failure, such as fatigue cracks or gross deformation, was seen during the inspections or after test. When the coated

The literature suggests that the maximum compressive load on the cervical spine during the performance of physical tasks is 1.2 kN [Moroney, 1988] and that the elastic limit of single motion segments could be as high as 1.23 kN, with a predicted ultimate compressive strength of 2.40 kN [Przybyla, 2007]. The results showed that the loads reached were significantly above the expected *in vivo* loads of the cervical spine, and that the HA coating

To determine the shear static and fatigue strength of the device, the smallest footprint (12x14 mm) and height (5 mm) was utilized. Testing was performed on an MTS 858 Mini-Bionix II test frame with the devices mounted on steel fixtures. According to the literature, the total ROM for flexion/extension of the cervical spine is greatest at the C4-C5 level, and appears to be slightly more than 20° when measured *in viv*o using computer assisted tracking software [Reitman, 2004]. The literature also shows that the amount of mean flexion is only slightly higher than the mean extension [Panjabi, 2001]. Therefore, the shear test was conducted at 10° for both flexion and extension. A comparison was made between HA coated devices and uncoated devices to determine if the coating had an adverse effect on the shear static and fatigue strength of the device. A sample size of 6 was used for the static tests with a displacement rate of 2 mm/min. The acceptance criteria was an offset yield load of greater than 135 N [Moroney, 1988] for the shear static testing and for the fatigue tests, a sample size of 2 was used with a run out load greater than 135 N [Moroney, 1988] with R=10 at 10 Hz. For the shear static testing at 10° flexion, the mean load at offset yield was calculated to be 4774 ± 372 N at a displacement of 0.44 ± 0.01 mm. For the shear static testing at 10° extension static shear, the mean load at offset yield was calculated to be 6788 ± 412 N at a displacement of 0.53 ± 0.02 mm. The primary mode of failure was fracture of the PEEK (Figure 2). Since the results of the static compressive shear strength resulted in a lower offset yield load in flexion (4744 N) versus extension (6788 N), the fatigue test was run at 10° flexion as a worst case scenario. The results of the fatigue testing at 1600 N showed the

devices were tested at a load of 7230 N, implant fracture occurred at 867,966 cycles.

did not negatively affect the axial compressive static or fatigue performance.

**2.1.1 Axial static and fatigue strength** 

**2.1.2 Shear static and fatigue strength** 

R=10 at 10 Hz.

devices reached 10 million cycles without experiencing any visible plastic deformation or fracture. For the coated devices, since the shear static loading was significantly higher than the acceptance criteria and the coating was shown to not have a detrimental effect on the axial static properties of the device, only shear fatigue testing was performed. The results were the same as the coated devices, 1600 N of compressive shear to 10 million cycles without experiencing any visible plastic deformation or fracture. Therefore, it was concluded that shear fatigue loads can be withstood far beyond those that would be expected *in vivo*.

Fig. 2. PEEK cervical arthroplasty device after static compressive shear testing to failure. Left – tested under 10° flexion, right – tested under 10° extension.

### **2.1.3 Fixation strength**

The fixation strength of the device was assessed by determining its pull out strength using a synthetic vertebral model. The largest (17x17 mm) and smallest footprints (12x14 mm) of 8 mm height were utilized. To avoid the variability produced by human cadaveric tissue, including bone mineral density and endplate geometries, a synthetic fixation medium of ASTM grade 15 polyurethane foam was selected for use. The fixation pull out test model consisted of compressing the device between two polyurethane blocks to 60 lbs, and then performing a uniaxial tensile load to failure test, thereby forcibly removing the implant. The test was conducted on an MTS 858 Mini-Bionix II test frame with appropriate fixturing. The displacement rate was 2.5 mm/sec. The implant was placed between the blocks in accordance with the surgical technique. A minimum of ten pullout tests were performed for each implant size. This data was then compared to eight previously tested cervical disc arthroplasty devices used clinically in cervical reconstruction that utilize press fit, keel or screw fixation using the same test protocol and laboratory [Cunningham, 2010]. Therefore, the results of this testing provides for a standardized testing methodology to quantify the strength of fixation and resistance to migration.

In all pullout tests for the two implant sizes, there was no incidence of intrinsic device failure. Failure always occurred at the polyurethane block-implant interface. The cam blade fixation indicated the highest pullout strength, which was statistically different from nearly all other designs (Table 1). Prosthetic endplates containing toothed ridges and keel fixation ranked next in terms of fixation strength. Although intuition favored the coronal keel as affording an increased mechanical advantage over the toothed surface, there were no

The Use of PEEK in Spine Arthroplasty 217

elevated beyond 1.0mm did not significantly improve fixation strength (p>0.05). Prosthetic endplates containing serrated edges (0.2mm) alone indicated the lowest fixation strength. These results indicate not only a high fixation strength but the ability of the PEEK housing for the cam blades to withstand the expected *in vivo* shear forces placed upon the device.

Wear testing was performed on six devices of the smallest footprint (12x14 mm) and 5 mm height using an MTS Spine Wear Simulator. The test parameters consisted of ASTM F2423-05 recommended multidirectional motion and static load profiles from 0-10 million cycles followed by ISO 18192-1 recommended motion and dynamic load profiles from 10- 20 million cycles. The test frequency was 2 Hz for motion and 4 Hz for load. The testing fluid consisted of newborn calf serum diluted with phosphate buffered saline (PBS) to a final protein content of 20 g/L. Ethylene-diamine tetraacetic acid (EDTA) was added to the serum at a concentration of 20 mM to bind the calcium ions present in the serum. EDTA is a known preservative, and together with the low protein content and PBS, the addition of sodium azide or other anti-bacterial agent was not used. The test fluid temperature was kept at 37 ± 3° C. Unloaded soak controls were used to account for moisture uptake. After 0.5 million cycles, the tests were stopped at 1 million cycles and at million-cycle intervals thereafter to clean, dry and gravimetrically assess the wear rates. The test fluid was subsequently collected and stored at -20° C. The average wear rates were determined using linear regression analysis with one-way-analysis-of-variance (ANOVA) used to determine if significant differences (p < 0.05) in the wear rates between methodologies were present. Complimentary to the wear testing, particle analysis was performed using light scanning electron microscopy (SEM). The proteinacious test serum containing PEEK wear debris underwent enzymatic digestion (5x Trypsin digestion and 1.5 mg/mL of Proteinase K per sample at 37° C for 24 hours) followed by a mild acid treatment (10% HCl at 37° C for 24 hours), meeting or exceeding similar protocols previously established to be equivalent for acid and base digestion (Niedzwiecki, 2001). The particles were then analyzed by SEM analysis and characterized according to their equivalent circle diameter (ECD), aspect ration (AR), roundness (R) and form factor (FF)

All implants maintained full functionality throughout each test duration. The results showed only a slight, non-significant variation in the wear rates over the course of 20 million cycles. The wear rate for the first 10 million cycles was approximately 0.26 ± 0.01 mm3/million cycles. For the interval of 10 to 20 million cycles, the wear rate increased slightly, 0.32 ± 0.02 mm3/million cycles. These wear rates are similar to other arthroplasty devices in clinical use (Figure 3). Although the wear rates were relatively consistent, the particle size was sensitive to the test method. The particle analysis revealed that the size (ECD) of the particulate was larger for the ISO testing methodology (2.56 µm) as compared to the ASTM methodology (1.67µm) (Figure 4). However the particle morphology, an equally important parameter influencing biological activity, was similar for both methods. The ASTM testing revealed a smaller particulate; however it may be unlikely that a static compressive load occurs in the cervical spine *in vivo*, since the literature suggests that a dynamic compressive load occurs due to muscle contraction during kinematic activities

**2.1.4 Wear testing and wear particulate characterization** 

in accordance with ASTM F 1877-05.

[Snijders, 1991].


observed statistical differences in pullout strength between the two endplate designs (p>0.05). In fact, the toothed surfaces indicated a slightly higher failure load over the keel device, although not statistically different. Moreover, the utilization of toothed ridges

Table 1. Fixation strength of the PEEK cervical arthroplasty device as compared to other forms of fixation. \* Fixation strength was significantly greater than all others with the exception of PCM Fixed Flange and Anterior Cervical Plate with Interbody Cage. \*\*Fixation strength was significantly greater than all others with the exception of PCM Fixed and Modular Flange, Prestige LP and Anterior Cervical Plate with Interbody Cage. There were no statistical differences between endplate surface areas.

observed statistical differences in pullout strength between the two endplate designs (p>0.05). In fact, the toothed surfaces indicated a slightly higher failure load over the keel device, although not statistically different. Moreover, the utilization of toothed ridges

> **Endplate Surface Area (mm2)**

Cam blade 168 386.3 ± 44.1\*

Cam blade 289 566.2 ± 72.0\*\*

**Failure Load (N)** 

300 257.4 ±28.5

300 308.8 ± 15.3

300 496.4 ± 40.0

300 528.0 ± 127.8

275 306.4 ± 31.3

250 286.9 ± 18.4

180 635.5 ± 112.6

**Type** 

serrations 

serrations with 1.0 mm toothed ridges 

mm x 15 mm screws

screws

center keel with teeth

serrations with 2.5 mm height center keel

Serration with 4 screws

Iliac Crest Autograft N/A 235 161.6 ± 16.6

Table 1. Fixation strength of the PEEK cervical arthroplasty device as compared to other forms of fixation. \* Fixation strength was significantly greater than all others with the exception of PCM Fixed Flange and Anterior Cervical Plate with Interbody Cage. \*\*Fixation strength was significantly greater than all others with the exception of PCM Fixed and Modular Flange, Prestige LP and Anterior Cervical Plate with Interbody Cage. There were

**Experimental Group Device Fixation** 

PCM Low Profile 0.2 mm endplate

PCM V-Teeth 0.2 mm endplate

PCM Modular Flange Serration with (4) 4

PMC Fixed Flange (4) 4 mm x 15 mm

Prestige LP 2.3 mm height

Kineflex C 0.2 mm endplate

no statistical differences between endplate surface areas.

Anterior Cervical Plate with Interbody Cage

PEEK cervical device

PEEK cervical device

(small)

(large)

elevated beyond 1.0mm did not significantly improve fixation strength (p>0.05). Prosthetic endplates containing serrated edges (0.2mm) alone indicated the lowest fixation strength. These results indicate not only a high fixation strength but the ability of the PEEK housing for the cam blades to withstand the expected *in vivo* shear forces placed upon the device.

### **2.1.4 Wear testing and wear particulate characterization**

Wear testing was performed on six devices of the smallest footprint (12x14 mm) and 5 mm height using an MTS Spine Wear Simulator. The test parameters consisted of ASTM F2423-05 recommended multidirectional motion and static load profiles from 0-10 million cycles followed by ISO 18192-1 recommended motion and dynamic load profiles from 10- 20 million cycles. The test frequency was 2 Hz for motion and 4 Hz for load. The testing fluid consisted of newborn calf serum diluted with phosphate buffered saline (PBS) to a final protein content of 20 g/L. Ethylene-diamine tetraacetic acid (EDTA) was added to the serum at a concentration of 20 mM to bind the calcium ions present in the serum. EDTA is a known preservative, and together with the low protein content and PBS, the addition of sodium azide or other anti-bacterial agent was not used. The test fluid temperature was kept at 37 ± 3° C. Unloaded soak controls were used to account for moisture uptake. After 0.5 million cycles, the tests were stopped at 1 million cycles and at million-cycle intervals thereafter to clean, dry and gravimetrically assess the wear rates. The test fluid was subsequently collected and stored at -20° C. The average wear rates were determined using linear regression analysis with one-way-analysis-of-variance (ANOVA) used to determine if significant differences (p < 0.05) in the wear rates between methodologies were present. Complimentary to the wear testing, particle analysis was performed using light scanning electron microscopy (SEM). The proteinacious test serum containing PEEK wear debris underwent enzymatic digestion (5x Trypsin digestion and 1.5 mg/mL of Proteinase K per sample at 37° C for 24 hours) followed by a mild acid treatment (10% HCl at 37° C for 24 hours), meeting or exceeding similar protocols previously established to be equivalent for acid and base digestion (Niedzwiecki, 2001). The particles were then analyzed by SEM analysis and characterized according to their equivalent circle diameter (ECD), aspect ration (AR), roundness (R) and form factor (FF) in accordance with ASTM F 1877-05.

All implants maintained full functionality throughout each test duration. The results showed only a slight, non-significant variation in the wear rates over the course of 20 million cycles. The wear rate for the first 10 million cycles was approximately 0.26 ± 0.01 mm3/million cycles. For the interval of 10 to 20 million cycles, the wear rate increased slightly, 0.32 ± 0.02 mm3/million cycles. These wear rates are similar to other arthroplasty devices in clinical use (Figure 3). Although the wear rates were relatively consistent, the particle size was sensitive to the test method. The particle analysis revealed that the size (ECD) of the particulate was larger for the ISO testing methodology (2.56 µm) as compared to the ASTM methodology (1.67µm) (Figure 4). However the particle morphology, an equally important parameter influencing biological activity, was similar for both methods. The ASTM testing revealed a smaller particulate; however it may be unlikely that a static compressive load occurs in the cervical spine *in vivo*, since the literature suggests that a dynamic compressive load occurs due to muscle contraction during kinematic activities [Snijders, 1991].

The Use of PEEK in Spine Arthroplasty 219

**Methodology Group 1- Normal Load Group 2 - Artificially Aged Group 3 – High Load** 

200 kGy in air

225…1024 N 225…1024 N 200…2000 N

for 40 days

ISO - 6°/3° flexion/extension ±2° lateral bending and axial rotation to 10 million cycles

The HA coating is applied to the outer endplates of the device via the well known plasma spray process. The coating was characterized in accordance with FDA guidelines. Given that the device' material (PEEK) is being exposed to what is considered a high temperature process, a potential change in the materials biocompatibility profile and/or material properties could occur. Therefore, several tests were performed to ensure that the thermal and chemical signature of the PEEK substrate was not adversely affected. This consisted of Differential Scanning Calorimetry (DSC), Thermal Gravimetric Analysis (TGA) and Fourier Transform Infrared Spectroscopy (FTIR), the most common tools used to characterize polymers. This was then compared with the 20-year historical measurements of PEEK provided by the

For the DSC, TGA and FTIR analysis, six machined samples of HA coated PEEK that subsequently had the coating removed by a 10% phosphoric acid treatment were compared to machined PEEK samples that had not been coated. The DSC analysis showed similarity between all samples and to machined PEEK samples. Likewise, ATR-FTIR showed no appreciable difference in the tested samples. TGA analysis resulted in primary degradation of the samples starting at approximately 550ºC and continuing until about 625ºC, showing an initial degradation loss of 35%. There was a dramatic decrease in sample mass (65% of the initial mass down to 0% of the initial mass) of the samples between 625º C and 800º C. There was no appreciable difference in the samples, and the degradation response looked very similar to the machined PEEK samples. All observed values (mean, maximum and minimums) were within the 20 year historical ranges of PEEK as verified by Invibio. More importantly, from a biocompatibility standpoint, since there were no additional absorption peaks found in the regions that formerly had an HA coating, and the HA coating was

manufacturer (Invibio). Analyses were performed on bulk and surface samples.

Exposure to O2 at 70° C

Compressive Load 4 Hz Motion 2 Hz

30 kGy in air

ISO - 6°/3° flexion/extension ±2° lateral bending and axial rotation to 10 million cycles

Compressive Load 4 Hz Motion 2 Hz

N/A

Environmental Condition Sterilization Aging

Dynamic Compressive

Load

30 kGy in air

flexion/extension ±2° lateral bending and axial rotation to 10 million cycles

Motion 2 Hz

**2.1.5 Hydroxyapatite coating and substrate effects** 

Table 2. Summary of Testing Methodology for All Groups.

Test Frequency Compressive Load 4 Hz

N/A

Motion ISO - 6°/3°

Fig. 3. Wear rate comparison of the PEEK cervical arthroplasty device to other devices in clinical use.

Fig. 4. SEM analysis of the PEEK particulate generated from the wear tests.

Fig. 3. Wear rate comparison of the PEEK cervical arthroplasty device to other devices in

Fig. 4. SEM analysis of the PEEK particulate generated from the wear tests.

clinical use.


Table 2. Summary of Testing Methodology for All Groups.

### **2.1.5 Hydroxyapatite coating and substrate effects**

The HA coating is applied to the outer endplates of the device via the well known plasma spray process. The coating was characterized in accordance with FDA guidelines. Given that the device' material (PEEK) is being exposed to what is considered a high temperature process, a potential change in the materials biocompatibility profile and/or material properties could occur. Therefore, several tests were performed to ensure that the thermal and chemical signature of the PEEK substrate was not adversely affected. This consisted of Differential Scanning Calorimetry (DSC), Thermal Gravimetric Analysis (TGA) and Fourier Transform Infrared Spectroscopy (FTIR), the most common tools used to characterize polymers. This was then compared with the 20-year historical measurements of PEEK provided by the manufacturer (Invibio). Analyses were performed on bulk and surface samples.

For the DSC, TGA and FTIR analysis, six machined samples of HA coated PEEK that subsequently had the coating removed by a 10% phosphoric acid treatment were compared to machined PEEK samples that had not been coated. The DSC analysis showed similarity between all samples and to machined PEEK samples. Likewise, ATR-FTIR showed no appreciable difference in the tested samples. TGA analysis resulted in primary degradation of the samples starting at approximately 550ºC and continuing until about 625ºC, showing an initial degradation loss of 35%. There was a dramatic decrease in sample mass (65% of the initial mass down to 0% of the initial mass) of the samples between 625º C and 800º C. There was no appreciable difference in the samples, and the degradation response looked very similar to the machined PEEK samples. All observed values (mean, maximum and minimums) were within the 20 year historical ranges of PEEK as verified by Invibio. More importantly, from a biocompatibility standpoint, since there were no additional absorption peaks found in the regions that formerly had an HA coating, and the HA coating was

The Use of PEEK in Spine Arthroplasty 221

Using an *in-vitro* human cadaveric model, the range of motion (ROM) of the device was evaluated at a single level, with data analysis based on the resulting operative and adjacent level multidirectional flexibility properties. A total of twelve fresh-frozen human cadaveric cervical spines (C1T3) were harvested en-bloc and utilized. Prior to biomechanical analysis, standard anteroposterior and lateral plain films were obtained to exclude specimens demonstrating intervertebral disc or osseous pathology. In preparation for biomechanical testing, the specimens were thawed to room temperature and cleaned of all residual musculature, with care taken to preserve all ligamentous attachments and facet joint capsules. The proximal (C1C3) and distal (T1T3) ends of the specimen were stabilized in metal tubing containers using four, four point compression screws. An optoelectronic motion measurement system (3020 Optotrak System) was used for kinematic analysis under a pure moment loading to evaluate the flexion/extension, lateral bending and axial rotational response after reconstruction of the C4C5 cervical level. A comparision was made

Flexion/extension testing demonstrated the diskectomy condition as producing a five degree increase (10.5 to 15.6°) in operative level range of motion compared to the intact spine and was statistically greater than intact and anterior plate-cage groups (p<0.05). Implantation of the PEEK cervical artificial disc restored motion to near intact level (11.9°) which was not statistically different from the intact spine (10.5°) (p > 0.05). The final reconstruction using the anterior plate-cage decreased the mean range of motion to 8.71° at the operative level, which resulted in significantly less range of motion than the PEEK artificial disc (11.9°) (p<0.05) but not intact condition (10.5 dg) (p>0.05). The elevated mean range of motion value for this reconstruction is attributable to the segmental instability created by the diskectomy condition (ANOVA F=3.72, p=0.018). Biomechanical testing in lateral bending loading did not reveal any significant differences in range of motion when comparing the intact spine and three treatment groups (p>0.05). Diskectomy (7.63°), PEEK artificial disc (6.90°) or anterior plate-cage construct (6.77°) afforded a statistical change in flexibility compared to the native intact condition (7.10°). The high standard deviation in the range of motion account for the lack of statistical significance under this loading modality (ROM: ANOVA F=0.58, p=0.633). Biomechanical testing in axial rotation indicated no statistical differences in ROM between the intact spine, diskectomy, PEEK total disc replacement or anterior plate-cage group (p>0.05). The cervical disc replacement procedure (7.21°) resulted in a slight increase in operative level flexibility relative to the intact (6.05°) and diskectomy conditions (6.44°) (p>0.05). The trends under this biomechanical loading mode failed to reach statistical significance secondary to specimen variability and high standard deviations. (ROM: ANOVA F=0.22, P=0.882). Despite the use of the Panjabi Hybrid testing protocol, there were very minimal changes in adjacent level ROM when comparing the intact spine, diskectomy condition and two reconstruction groups (p>0.05). The only significant change was observed at the adjacent level in flexion extension loading condition. The normalized-to-intact ROM at C6C7 level in these two reconstruction conditions was significantly higher than that was detected in diskectomy condition (p<0.05). Similar to the operative level ROM, the trends under this biomechanical loading mode failed to reach statistical significance secondary to

between the intact, destabilized, reconstructed and fusion procedures.

specimen variability and high standard deviations.

**2.1.6 Biomechanical analysis** 

readily removed with a mild solvent, it is highly unlikely that any covalent bonding occurred between the HA and the PEEK material. Furthermore, there was no measurable change in the molecular structure of the PEEK induced by adding the HA layer. Therefore, no chemical bonding was noted to occur, the surface properties and subsequently the biocompatibility profile of the PEEK substrate have not changed, and it can be concluded that only a mechanical bond exists between the PEEK and HA coating.

Biocompatibility tests in accordance with ISO 10993 parts 5, 10 and 11 were conducted at NAMSA. These tests were carried out on coated and an equal number of substrates having the coating removed with a mild solvent (phosphoric acid) in order to expose the underlying substrate. The control consisted of virgin PEEK:
