Pre-clinical assessment of novel orthopaedic joint procedures generally includes material characterization and biomechanical testing in the lab setting, biocompatibility evaluation and in vivo studies in large animal models prior to the commencement of clinical trials. Biomaterial characterization provides insight into the properties and performance of single or combined materials using physiologically relevant parameters but is limited by sample geometry and controlled environments. In vivo implantation in large animal models such as goats or sheep (Zur et al., 2011; Patel et al., 2015; Vrancken et al., 2017) provides a thorough evaluation of the design, safety, performance and biocompatibility of the novel procedure in vivo, thus derisking the procedure before clinical trials. However, such studies are costly and time-consuming, and researchers are ethically bound to reduce such studies and limit sample sizes as far as possible while satisfying regulatory requirements.

The ovine stifle joint is a well-established knee joint model for pre-clinical research in reconstruction or replacement of the cruciate and collateral ligaments, partial and total meniscus replacements, cartilage lesion repair and osteoarthritis treatment (Radford et al., 1996; Tapper et al., 2006; Taylor et al., 2006; Zur et al., 2011; Gregory et al., 2012; Patel et al., 2015)

A clear gap exists in the pre-clinical assessment of novel orthopaedic joint procedures, for dynamic cadaveric testing of the ovine knee joint under simulated physiological gait, as a screening method and at a reduced cost (Maher et al., 2011), before advancing to in vivo animal studies. Research studies have developed techniques for dynamic testing of meniscus replacements in cadaveric ovine stifles but were limited to low axial loads and were only run for 10 cycles (Cottrell et al., 2008; Brophy et al., 2010; Holloway, 2012). This paper describes such a simulator, its verification in terms of loading and kinematics, then a proof of concept study in which a meniscus replacement is tested in a cadaveric ovine knee joint under simulated gait.

Meniscal surgery is one of the most common orthopaedic surgical interventions (Englund and Lohmander, 2006; McDermott et al., 2008; Abram et al., 2018). Although meniscectomy is the current standard of care after irreparable meniscal injury, and may alleviate symptoms in the short-term, it increases the risk of the onset of osteoarthritis (Englund and Lohmander, 2006; McDermott et al., 2008). In such cases, high tibial osteotomy is often performed to shift joint contact loads to the unaffected compartment in order to reduce pain and help slow OA progression in the meniscectomized compartment (Lee and Byun, 2012). Meniscal allograft transplants (Novaretti et al., 2019; Verdonk et al., 2013) and partial replacement scaffolds (Stone et al., 1997; Verdonk et al., 2011) have been used to replace the injured meniscus, with limited success in long-term function and survivorship (Hutchinson et al., 2013; Winkler et al., 2020). Novel total meniscus replacement devices aim to fill this treatment gap (Zur et al., 2011; Patel et al., 2015; Vrancken et al., 2017). No proven long-term joint-preserving treatment options exist for patients with irreparable meniscal damage.

The aim of this study was to develop an ovine knee joint simulator and to assess the kinematics and contact pressures of the ovine stifle joint under simulated gait. The use of the simulator will be demonstrated in a pilot study when the knee has been implanted with a novel fiber-matrix reinforced polyvinyl alcohol-polyethylene glycol (PVA-PEG) hydrogel meniscus and to compare it against intact stifles. It was hypothesized that the gait kinematics and contact pressures post meniscal replacement would not differ significantly from those of intact stifles.

Two single-station ovine gait simulators were used (Figure 1). The joint simulator controlled femoral flexion-extension and applied a cyclic axial contact force to the distal end of the tibia, while allowing articular movement in five degrees of freedom (DOF). During knee flexion-extension, tibial movement was guided by the natural ligaments of the stifle joint, namely, the anterior (cranial) cruciate ligament, the posterior (caudal) cruciate ligament and the medial and lateral collateral ligaments, which were left intact together with surrounding soft tissue structures. This method was expected to result in stifle kinematics that were not significantly different from in vivo conditions because motion of the natural knee is restricted and stabilized by passive soft tissues (Houtem et al., 2006; Taylor et al., 2006). The simulator design was based on a force-controlled joint simulator that has been shown to allow physiological movement of cadaveric knee joints under gait loading conditions (Houtem et al., 2006).

The gait simulator ran at a frequency of 1.55 Hz for 500,000 gait cycles (3.73 days), equivalent to approximately a year of normal use by adult sheep in vivo (Ruckstuhl, 1998). The simulator was mounted inside a refrigerator to maintain the stifle at 4°C–8°C to delay tissue necrosis for the duration of the test. An in-line pressure syringe pump (Graseby 3200; Smiths Medical International, MN) was used to deliver diluted sterile-filtered Bovine Calf Serum (BCS) (product 12133C; Merck Group, Germany) from a mounted 50 mL Luer Lock syringe at a rate of 2 mL per hour. The lubricant was delivered directly into the joint space throughout the test by placing an 18G × 38 mm cannula in the femoral intercondylar notch. The syringe pump was replenished with fresh lubricant every 24 h.

The most important findings of this study were that the novel ovine gait simulator led to peak contact pressures and kinematics of intact stifle joints that did not differ significantly from those reported in literature. The fiber-matrix reinforced PVA-PEG hydrogel medial meniscal replacement restored the medial peak contact pressures, but not the lateral contact pressures, thus demonstrating an example of the potential use of the simulator.

Previous research studies that performed fatigue testing of cadaveric ovine stifles for meniscus evaluation were limited to low axial contact loads, thus their results could not be compared to this study (Cottrell et al., 2008; Brophy et al., 2010; Maher et al., 2011; Holloway, 2012). To the authors’ knowledge, this study is the first to perform fatigue testing of ovine cadaveric joints under simulated physiologic gait conditions. The absence of macroscopic damage or degeneration on native menisci and their attachments, and comparable cartilage condition following 500,000 gait cycles, confirmed native tissue survivability during a test that was designed to replicate normal joint function. Future studies should assess the articular surfaces for damage after novel surgical procedures.

The peak contact pressures were of primary investigative interest for this study to confirm physiologic cartilage loading on each of the condyles. The static medial and lateral peak pressures in the intact group did not differ significantly from published literature (Heckelsmiller et al., 2017; Fischenich et al., 2018), which also reported higher lateral peak pressures. Contradictory results from Lee-Shee et al. (2007) show higher peak contact pressures on the medial condyle with a magnitude of 8 ± 1 MPa. The large variability in peak contact pressures recorded in literature can most likely be attributed to the varying contours of the stifle joint plateaux (Lee-Shee et al., 2007). However, the contrasting results from Lee-Shee et al. could be a result of the constraint of the varus-valgus and internal-external rotations of the joints during their testing. The results of this study may also provide a closer representation of the peak contact pressures in the joint given that Tekscan pressure mats, such as those used by Fischenich et al. and Heckelsmiller et al., are relatively stiffer making them less suitable and less repeatable in small and substantially curved surfaces such as those of the ovine stifle joint (Brand, 2005; Herregodts et al., 2015). Furthermore, Tekscan pressure mats have a lower spatial resolution than pressure sensitive films and require averaging across sensor nodes due to artefactual recordings (Brand, 2005). It remains possible that the peak pressures measured were inaccurate due to a combination of the contact area not being measured around the cruciate ligaments, and saturation of the pressure film in part due to shear loading effects from the sloping articular surface in that area.

Research studies have evaluated the peak contact pressures of meniscus replacement or scaffold implanted ovine stifle joints in comparison to the intact, partially or fully meniscectomized conditions (Brophy et al., 2010; Maher et al., 2011; Holloway, 2012; Fischenich et al., 2018). The implanted results from the present study were directly compared to those reported by Fischenich et al. for the intact, meniscectomized, allograft and TPE hydrogel implanted conditions performed in the medial compartment at 60° flexion with an axial load of 893N (91 kg) (Fischenich et al., 2018). All other studies were limited to the lateral compartment and low axial loads. The medial and lateral peak pressures were the primary interest for this study, to provide insight on the load bearing and load transmission capabilities of the implant in comparison to the intact condition during loading that simulated normal ovine gait and, thus, showed the usefulness of the simulator as a screening tool for total meniscus replacements.

Medial peak pressures were not significantly different between the implanted and intact groups tested in this work (p > 0.4). The medial peak pressures of the implanted group were also not significantly different from the Fischenich et al. intact group (p > 0.1) and their TPE hydrogel implanted group (p > 0.08) (Figure 6). The medial peak pressures of the implanted group in the present study were significantly lower than the Fischenich et al. meniscectomized group (p < 0.03) and not significantly different from the Fischenich et al. allograft group (p > 0.09). These findings show that the fiber-matrix reinforced PVA-PEG hydrogel meniscal replacement restored the medial peak contact pressures.

Similarly to the intact group, the peak contact pressures of the implanted group were significantly higher in the lateral compartment (p < 0.01) following implantation in the medial compartment. This may have resulted from a change of limb alignment or joint congruity caused by the implantation procedure. In spite of this, the peak lateral contact pressures of the PVA-PEG hydrogel implanted group were not significantly different from the Fischenich et al. allograft group (p > 0.1) and TPE hydrogel implanted group (p > 0.08).

At 55° flexion, representing the stance phase of the gait cycle, peak contact pressures of the intact group were generally located on the posterior aspect of the medial tibial plateau, whereas the lateral peak contact pressures were located along the midline of the joint. These results matched those of previous studies (Heckelsmiller et al., 2017; Fischenich et al., 2018). However, Lee-Shee et al. found that peak contact pressures were located along the midline of the joint in both the medial and lateral compartments of the stifle, which could be a result of the applied constraints during testing (Lee-Shee et al., 2007). For the implanted group, peak contact pressures had shifted to the central aspect of the medial condyle at the cartilage-cartilage contact area and were distributed across the central aspect of the lateral condyle at 55° flexion. It could be speculated that the medial pressure distribution shift resulted from the absence of peripheral attachments on the ovine implant such as the meniscotibial ligaments and the deep fibers of the MCL (Allen et al., 1998) respectively, that constrain its anterior-posterior movement during knee flexion. Similarly, the effects of the surgical procedure and differences in conformity between the implant and each stifle joint on the contact pressure distributions cannot be ruled out and are not well understood. Nevertheless, the contact pressure distributions of the implanted group of this study were in agreement with those reported by Fischenich et al. (2018) for both the allograft and TPE hydrogel implanted groups in both compartments of the knee.

The kinematics of the ovine stifle in this study did not differ significantly from those reported in vivo (Tapper et al., 2006) in any of the degrees of freedom of motion. The general patterns of motion were the same, with coupled internal rotation and anterior translation of the tibia as the stifle flexed, and minimal movements in the remaining degrees of freedom. Thus the relatively simplified loading method via the distal tibia was adequate to replicate physiological gait and loading parameters reported previously (Taylor et al., 2006; Taylor et al., 2011). It would have been possible to record the joint kinematics at the beginning as well as the end of the test, but the lack of visible joint surface erosion meant that we did not suspect any resulting kinematic changes. Higher variability in vivo could be due to normal inter-subject gait variations, which did not occur in this controlled in vitro study. To the authors’ knowledge, this is the first study to assess 3D in vitro kinematics of intact ovine stifle joints in a physiological gait simulator, and to show that the passive ligaments and articular contacts guided the joint kinematics such that the motion was a close simulation of normal gait.

This study has limitations: firstly, tests were performed in a short timescale, under a consistent pattern of gait and with continuous motion, and did not fully simulate in vivo conditions, including dynamic muscle function and natural lubrication of a synovial joint. Joint simulators normally apply a consistent loading/kinematic profile to simulate normal gait to test wear of joint replacements. Variations in the pattern are not usually imposed by the control system. However, the novel ovine gait simulator uses native joints rather than prostheses, and they have inherent natural variability between specimens. The simulator enables the natural kinematics of the stifle joint because the kinematics are partly controlled by the ligaments, introducing tibial rotation and AP translations, so that the articulating path is not a linear one but also includes shear. Thus, there are inter-specimen variations in the gait pattern, but not simulation of different activities. Secondly, the ovine simulator operated at refrigeration temperatures of 4°C–8°C, which are significantly lower than body temperature (39°C) and could possibly affect friction and wear during joint articulation (Bortel et al., 2015). Lower testing temperatures were required to delay tissue necrosis across each 4-day long test, in order to obtain 500,000 cycles at physiological rates. Increasing the loading frequency would be unphysiological and would likely alter the lubrication regime, so the reduction of testing temperature was taken to be the least-bad compromise necessary to attain 500,000 cycles, equivalent to approximately a year of normal use by adult sheep in vivo (Ruckstuhl, 1998). Thirdly, contact pressures were limited to static measurements within the pressure film range (2.5–10 MPa), which allowed for the measurement of peak pressures but resulted in an incomplete contact pressure map because pressures below 2.5 MPa were not recorded. Peak contact pressures were prioritized in this study as they are a primary variable affecting chondroprotection and joint preservation. Furthermore, pressure film measurements are sensitive to temperature, humidity and shear forces, similarly to pressure sensors (Lee-Shee et al., 2007; Herregodts et al., 2015). The addition of plastic wrap, which was used to prevent its exposure to joint fluid, increased the pressure film thickness (240 µm thick) which could affect stifle joint contact mechanics (Herregodts et al., 2015).

The current feasibility study of the use of the ovine gait simulator to evaluate meniscus replacements was limited by a relatively small sample size (n = 5/6). Variability in the stifle joint anatomy, surgical procedure and stifle joint alignment in the ovine gait simulator (Houtem et al., 2006; DesJardins et al., 2007) warrant a larger sample size and a repeated-measures analysis in future work to obtain more definitive results when using the ovine gait simulator. Nonetheless, the results from this study had better repeatability than comparable ovine cadaveric studies reported in literature (Fischenich et al., 2018). Cartilage damage assessment across intact, meniscectomised and implanted conditions, could also be of interest and would require sourcing stifle joints from young sheep with pristine cartilage, which we leave for future work. Furthermore, a power analysis was not performed because this is a feasibility study to demonstrate the novel ovine gait simulator, rather than a study of the surgical procedures.

This study found that peak contact pressures and kinematics of intact stifle joints in the ovine gait simulator were comparable to literature, thus providing support for its efficacy as a tool to evaluate novel meniscus or other surgical procedures. The pilot study with the novel fiber-matrix reinforced meniscal replacement implant found that it restored medial peak contact pressures and native joint kinematics but did not restore lateral peak contact pressures. These findings support the use of an ovine gait simulator as a screening tool prior to evaluation of novel procedures in live animals.