Decreased stress shielding with a PEEK femoral total knee prosthesis measured in validated computational models

1 Due to their high stiffness, metal femoral implants in total knee arthroplasty may cause stress shielding of the peri-prosthetic bone, which can lead to loss of bone stock. Using a polymer (PEEK) femoral implant 3 reduces the stiffness mismatch between implant and bone, and therefore has the potential to decrease strain 4 shielding. The goal of the current study was to evaluate this potential benefit of PEEK femoral components 5 in cadaveric experiments. Cadaveric femurs were loaded in a materials testing device, while a 3-D digital 6 image correlation set-up captured strains on the surface of the intact femurs and femurs implanted with 7 PEEK and CoCr components. These experimental results were used to validate specimen-specific finite 8 element models, which subsequently were used to assess the effect of metal and PEEK femoral components 9 on the bone strain energy density. The finite element models showed strain maps that were highly 10 comparable to the experimental measurements. The PEEK implant increased strain energy density, relative 11 to the preoperative bone and compared to CoCr. This was most pronounced in the regions directly under the 12 implant and near load contact sites. These data confirm the hypothesis that a PEEK femoral implant can 13 reduce peri-prosthetic stress shielding.


15
Total knee arthroplasty is one of the most successful interventions in orthopaedic surgery for treatment of 16 patients suffering from degenerative knee joints, with impressive survival rates [AOANJRR, 2019]. Despite its 17 success, there is still a small number of cases that require revision for aseptic loosening, which is why there 18 is an ongoing drive for development of new implant systems and materials. Before clinical introduction, these 19 new systems should be subjected to extensive pre-clinical evaluation to determine its potential benefits and 20 risks.

28
In TKA, change in mechanical loading of the bone is the main stimulus driving remodelling and 29 periprosthetic bone density changes, which is influenced by the relatively stiff implant materials being used, 30 such as Cobalt-Chromium (CoCr) (Carter et al., 1987;Frost, 1964;Huiskes et al., 1987;Kummer, 1972

37
Therefore, alternative, more compliant materials that have the potential to reduce stress shielding have been 38 considered for femoral TKA components, such as polyacetal implants that were evaluated in a clinical trial 39 (Moore et al., 1998  Before testing, specimens were thawed at room temperature for at least 12 hours. After drying in ambient 91 air, the surface was coated with matt white spray paint (Plastikote Ltd, UK). Once dry, a matt black speckle 92 pattern was applied onto the white surface by an experienced DIC operator. Then, the specimen was placed 6 in the unidirectional servo-hydraulic loading apparatus with a 15kN load cell (MTS45820, MTS Systems Corp., 94 USA) (Figure 1). The specimens-specific polyurethane mould was positioned onto the femoral cartilage, positioned to obtain optimal lighting of the specimen (Figure 1). The uniaxial compression load for the 98 specimens was not predefined as a fixed value, to avoid femoral fractures. Alternatively, the specimen-99 specific load was determined using the left femur by first measuring the unloaded baseline noise in the DIC 100 setup, after which the femoral load was increased incrementally until a 10:1 signal-to-noise ratio (SNR) was 101 achieved. A 10:1 SNR provides good accuracy of the captured data, accounts for potential noise increase and 102 reduces any measurement error to an acceptable level. The resulting loads were 3.5 kN, 2.5 kN and 3.5 kN 103 for specimens 1, 2 and 3, respectively. These loads were then applied to both the left and right femurs. The 104 load applicator allowed for free translations in the horizontal plane and free varus/valgus rotations, while 105 other degrees of freedom were constrained. Six cycles of loading/unloading were executed. Of each loaded 106 and unloaded state six DIC image arrays were captured.

107
Following the intact femur measurements, either a CoCr (Maxx Freedom Knee) or PEEK (adopted from 108 Maxx Freedom Knee) implant was cemented (Palacos R, Palamix system, Heraeus Medical GmbH, Germany) 109 onto the femur according to surgical guidelines. After implantation, one hour was allowed for cement curing, 110 after which the specimen was placed back into the testing rig and the experimental procedure as described 111 above was repeated.

113
Digital image correlation 114 A dual-camera DIC setup using Sigma 105mm lenses was used (Limess GmbH, 2 megapixel) to capture 3-D 115 femur strain data (Figure 1). Prior to a measurement series, the corresponding rigid DIC-calibration tool (12 116 x 9 grid of 5 mm targets) was used to calibrate the position of the cameras relative to one another via 117 triangulation to define the 3D coordinate system for the bone surface. The camera setup was placed at 118 approximately one meter from the specimens for optimal focal depth, with a relative pan angle of 10 degrees 119 for 3-D capturing. As the region of interest (ROI - Figure 2) must be visible for both cameras, higher angles 7 could lead to loss of field of view. Images were captured using Vic3D software (Correlated Solutions Inc., Irmo, SC, USA). Lighting was arranged such that maximal contrast was reached while avoiding pixel

209
Once implanted, the epicondylar surface strains generally decreased, suggesting stress shielding in both 210 reconstructions. The strain decrease was larger with a CoCr implant than with a PEEK component, which was 211 particularly obvious in specimens 1 and 2, and more subtly in specimen 3. Frequency plots of the surface 212 microstrains showed a higher peak in the low-strain region for the implanted specimens, which in case for 213 specimens 1 and 2 were more pronounced for the CoCr reconstructions (Figure 4). For the larger strain 214 regions, typically, the curves for the intact specimens were slightly above the implanted specimens, indicating 215 strain-shielding for both implant types.

216
In general, the FE simulations showed good agreement with the experimental strain patterns (Figure 3) and 217 magnitudes (Figure 4), and thus provided a satisfactory validation of the models. Exceptions were the intact 218 right (PEEK) FE models of specimens 1 and 2, which showed a similar distribution but generally lower strain 219 values, which was also reflected in the strain frequency plots (Figure 4).

222
For all specimens PEEK led to an overall increase in SED compared to the intact cases, whereas for CoCr a 223 decrease was seen in specimen 1 and slight increases in specimens 2 and 3 ( Figure 5). For specimens 1 and 224 2, the increase in SED in the PEEK reconstructions was most pronounced distally (ROI 5), while in specimen 3 225 this increase was also seen more proximally (ROI 3). Stress shielding was observed in the PEEK 226 reconstructions of specimens 2 and 3 in the anterodistal area (ROI 1), but to a lesser extent than CoCr. This cadaveric study demonstrates the potential for PEEK femoral components to reduce periprosthetic bone 296 strains to more physiological levels when compared to CoCr.   The region is chosen such that the speckle patterns are visible in both trials.