Functional biomechanical performance of a novel anatomically shaped polycarbonate urethane total meniscus replacement

Implant geometry was established by segmenting the medial menisci from eight routine knee MR scans of healthy male subjects (aged 21–67) using Mimics (v14.0, Materialise, Leuven, Belgium), and 3D models were created. Meniscal length and width were determined for each 3D model and the model that approached mean dimensions closest was chosen to serve as the implant geometry. The meniscal horns were extended 5 mm to allow for suture fixation to the tibia plateau. Bionate^® PCU (grade II 80A, DSM Biomedical, Geleen, Netherlands) implants were subsequently produced through injection moulding. The implants were hydrated in 37 °C saline for 2 weeks prior to testing.

Fresh frozen, left human cadaveric knees were obtained from our institutional Department of Anatomy. Calibrated anteroposterior and lateral knee radiographs were taken to exclude specimens that showed joint space narrowing, osteophytes or chondrocalcinosis. In addition, the images were used to estimate medial meniscus length and width from tibial plateau size measurements [21]. As the prototype implant was available in one size only, knees were excluded from this study when their estimated medial meniscus length or width deviated more than ten per cent from the implant dimensions. In total, ten knees were excluded based on radiographic signs of joint degeneration and seven additional knee joints because of size mismatching. Five male knees (aged 70–88) were included for testing.

To fit the specimens in the loading rig, the femur and the tibia were transected approximately 15 cm above and below the joint line, while taking care to leave the knee capsule intact. All soft tissues surrounding the knee capsule were left intact, including the full quadriceps tendon. The specimens were then transferred to the knee loading rig, which allowed the joints to move within their natural six degrees of freedom (Fig. 1a) [28]. To align the knee joint, the femur was positioned with its intercondylar notch in the intersection of the flexion/extension axis and the varus/valgus axis, and the tibial shaft was aligned with the internal/external rotation axis of the apparatus. Upon alignment, the specimens were permanently fixed to the holders of the loading rig with polymethyl methacrylate. These holders allowed repositioning of the knee joint without affecting its alignment.

Subsequently, the joint capsule was opened through a medial parapatellar incision. To access the full medial knee compartment, the femoral attachment of the medial collateral ligament (MCL) was released by isolation of a bone cap that could be repositioned with a screw (Fig. 1b). The bone cap technique was chosen to allow easy and reproducible repositioning of the MCL, without changing its kinematics [34]. Three tantalum beads (0.8 mm diameter) were inserted in the periphery of the anterior and posterior horn, and the mid-region of the medial meniscus and the implant (Fig. 1c). One bead (1 mm diameter) was inserted in the tibial origin of the anterior cruciate ligament, to define a reproducible origin of the coordinate system. Four additional tantalum markers (1 mm diameter) were inserted into both the proximal tibia and the distal femur. A metal ring was sutured to the distal end of the rectus femoris tendon, to allow for patellofemoral and tibial load transfer from the loading apparatus.

The loading protocol (Table 1) was designed such that we could analyse meniscal displacement and rotational laxity during a simulated squat movement (between 0° and 90° flexion), knee AP laxity in an unloaded knee joint (in 30° and 90° flexion) and cartilage contact pressures under the maximum compressive load (in 90° flexion). Compressive loads in each flexion angle were derived by solving the equilibrium equations belonging to a free body diagram of a squat [31]. Pilot tests demonstrated that the cadaveric tissue could not withstand physiological loads. To prevent tissue damage, the compressive loads, as derived by solving the equilibrium equations, were linearly down scaled to a maximum of 1000 N. For the same reason, loading of the rectus femoris tendon was restricted to 250 N, which was kept constant during all tests. Since the quadriceps loads resulting from the equilibrium equations were larger than 190 N, the quadriceps loads applied in this study should mainly be considered as joint stabilizers.

In each flexion angle, the knee was tested without tibial torque (neutral), and with an internal and external tibial torque of 3.4 Nm [15]. In addition, in 30° and 90° flexion, AP stability of the knee joint was tested through anterior drawer tests. Anterior loads of 67 N were applied to the proximal tibia, to mimic the load that is applied in clinical knee laxity testers [7]. To assess whether repetitive loading of the cadaveric tissue influenced the outcomes of our tests due to tissue degeneration, a second set of measurements in 0° flexion was performed (Table 1).

Testing of the native meniscus (condition 1, Fig. 2a) was followed by the implant (condition 2, Fig. 2b), a total meniscectomy (condition 3, Fig. 2c) and a meniscal allograft (condition 4, Fig. 2d). To start condition 2, a complete medial meniscectomy was performed. The excised meniscus was wrapped in saline-soaked gauze and frozen at −80 °C until its use as allograft meniscus. Tibial bone tunnels (2.5 mm diameter) were drilled from the anteromedial side of the tibia to the anterior and posterior attachment site of the meniscus on the tibial plateau. Two suture threads (FiberWire #2, Arthrex, Naples, FL, USA) were fixed to each horn of the implant and guided through the tunnels. The implant was positioned and fixed by knotting the sutures together. For testing of condition 3, the implant was removed, and the knee was retested for the meniscectomy condition. Lastly, for condition 4, the tests were repeated for an allograft meniscus, which was obtained from the preceding experiment. As all knees were size matched to the implant, their menisci were regarded to be properly sized allografts as well. The horns of the allograft were fixed in a similar fashion as those of the implant. To comply with the surgical standards that are clinically applied, in addition, the allograft was circumferentially fixed to the joint capsule using four vertical mattress sutures.

RSA was used to determine meniscal kinematics, knee AP laxity and tibial rotational laxity [26]. For each separate loading step from Table 1, a bi-planar roentgen image was taken. The 3D locations of the tantalum beads inside the tibia, native meniscus, implant and allograft were obtained from each image by relating their 3D positions to the distances known from a calibration image. All tibial beads were superimposed, allowing the determination of the different locations of the meniscus on the tibia for each condition. The meniscal locations were converted to meniscal translations along the AP and ML axes for the flexion intervals 30°–0°, 60°–0° and 90°–0° for the anterior, mid and posterior regions of the meniscus or implant. To determine AP laxity (in 30° and 90° flexion), the centroid of the tantalum beads inside the tibia was determined in unloaded and loaded state. AP laxity was defined as the translation of the centroid along the AP axis. Rotational laxity was defined as the angular rotation of the tibia around its longitudinal axis, as a result of the application of the internal or external torque. The set-up used in our laboratory allowed to determine bead location with an accuracy of 50 μm and repeated measurements can be performed with an accuracy of 200 µm [6].

In 90° flexion, tibial contact mechanics were evaluated using piezo-electric pressure sensors (Fig. 1d, e, K-scan 4010 N, Tekscan, South Boston, MA, USA). To prevent fluid penetration into the sensor, the sensor was placed between two layers of Tegaderm film dressing (3M, St. Paul, MN, USA). After sealing, all sensors were conditioned, equilibrated and calibrated using a materials testing system (MMED, Materials Technology Corporation, La Canada, CA, USA) and a custom calibration tool consisting of two Teflon plates. To optimize the accuracy of the pressure sensor, the calibration curve was derived from a ten point polynomial fit through the calibration points that were equally distributed between 0.9 and 9.0 MPa [8]. Consequently, cartilage contact pressure could be determined with a precision of 0.035 MPa. To allow sensor insertion, the knee was distracted by the weight of the tibia and a 2-cm incision was made to the anterior meniscotibial ligament (Fig. 1b). A suture with a 10″ straight needle (PDS II 2-0, Ethicon, Somerville, NJ, USA) was knotted to the most posteromedial corner of the sensor and guided along the cruciate ligaments. This suture allowed to pull the sensor exactly underneath the meniscus and was re-used to place the sensors for the implant, meniscectomy and allograft conditions in the same position. Additionally, the anteromedial edge of the pressure sensor was fixed by a suture through the knee capsule. A new sensor tab was used for each meniscal condition. Peak pressure, mean pressure and contact area were determined from the pressure maps for each meniscal condition.

A sample size calculation was based on a previous study that evaluated the difference in knee contact mechanics after total medial meniscectomy and subsequent meniscal transplantation [25]. Five knees were required to detect a difference of 1.1 MPa (standard deviation 0.5 MPa) in mean pressure, using a significance level of 0.05 and a power of 80 %.

For all three meniscal conditions, the translation patterns along the AP axis were similar; translation occurred in posterior direction and increased with flexion angle, and the anterior horn was more mobile than the mid-region (p = 0.002) and posterior horn (p < 0.001) (Fig. 3a). A significant increase in the posterior translation was observed when replacing the native meniscus with the implant (p < 0.001) or an allograft (p < 0.001). When comparing the outcomes for each location–flexion interval combination, the differences in posterior translation between the implant and allograft were never statistically significant, whereas those between the meniscal replacements and the native meniscus were always significant (Fig. 3a).

Translations along the ML axis were less regular, particularly for the anterior horn. With increasing flexion angle, the implant and allograft anterior horn both moved medially, whereas the native meniscus moved laterally (Fig. 3b). Also along the ML axis, meniscal mobility increased when the native meniscus was replaced by the implant (p < 0.001) or an allograft (p = 0.001). ML displacement of the implant and allograft was not significantly different. When comparing the outcomes for each location–flexion interval combination, the differences in ML translation between the implant and allograft were always non-significant, whereas the differences between the implant and native meniscus were always significant (Fig. 3b).

The effect of changing the meniscal condition on AP laxity was larger in 30° flexion than in 90° flexion (Fig. 4). In 30° flexion, AP laxity was significantly increased for the implant, meniscectomy and allograft conditions (all p < 0.001). However, replacing the native meniscus by the implant did not significantly increase AP laxity in 90° flexion, whereas the increases in AP laxity found after meniscectomy and allograft transplantation were significant (p = 0.012 and p = 0.016). The changes in internal and external rotational laxity observed between the four meniscal conditions at each separate flexion angle were minor. The differences in rotational laxity between the first and second evaluation in 0° flexion for each meniscal condition were small and not significant (Fig. 5).

A representative example of the pressure distribution on the tibia articular cartilage for the four meniscal conditions is shown in Fig. 6. The mean pressure for the native meniscus condition was significantly increased for the implanted knee (p < 0.001), after meniscectomy (p < 0.001) and after allograft transplantation (p = 0.019, Fig. 7). Both the implant condition and allograft condition did significantly reduce the mean contact pressure compared to the meniscectomy condition (p = 0.018 and p < 0.001, Fig. 7). Also the peak pressure was significantly elevated for the implant, meniscectomy and allograft conditions with respect to the native meniscus (all p < 0.001). The peak pressures for the implant and allograft were highly similar. The contact area on the tibia articular cartilage was largest for the native meniscus and was significantly reduced for the implant, allograft and meniscectomy conditions (all p < 0.001). Whereas the implant only slightly increased the contact area with respect to the meniscectomized condition, meniscal allograft transplantation resulted in a significant increase (p < 0.001).