This review was conducted in accordance with the PRISMA guidelines (Moher et al., 2009). A comprehensive and systematic review of the literature was performed to identify studies investigating the meniscal influence on the tibiofemoral contact mechanics in human knee joints. For the literature research, the Cochrane, PubMed and Web of Science databases were used. The literature search strategy was developed using a combination of keywords and medical subject heading (MeSH; Table 1), which were extended to maximize the inclusion of potentially relevant studies.

The results of the database searches were transferred into Endnote (Clarivate Analytics, PA, United States), where the references were automatically updated and duplicates removed. All titles and abstracts of the identified publications were screened. Papers containing in vitro tibiofemoral CP data and those investigating healthy, non-degenerated menisci, meniscal interventions, meniscal tears, repair or replacement or meniscectomy and their influence on the CP were included. Per requirement, the CP must also be clearly assignable to the respectively used knee flexion angles (0, 30, and 60°). Studies with computational/simulation approaches, studies in which the focus was other than on human knee joints, gender impact studies and those investigating knees with pathological alterations were excluded. Additionally, the reference lists of the selected publications were screened to include relevant studies that were missed during the previously described selection process.

All publications that included CPs were evaluated graphically, indicating axial loads, meniscal states, interventions, knee joint compartment and the respective peak CP. Only studies directly reporting CP values were investigated and analyzed. The detailed selection process of the study selection is given in Figure 2. The full text of the remaining articles was then analyzed and data from eligible studies were extracted.

From a biomechanical point of view, a reduced tibiofemoral contact area caused by meniscal tears (Figure 8) or partial or total meniscectomy (Figure 9) can potentially lead to premature gonarthrosis. Therefore, we investigated the effects of meniscal injuries and their surgical treatments on the tibiofemoral CP in this chapter.

The different test setups clearly indicate that the used hardware and test strategies greatly influence the level and changes of CP values (Figure 10). A non-physiological alignment and embedding of knee specimens can lead to considerably changed and misleading values.

During natural movements, the knee joint offers six degrees of freedom (DOF) (Hirschmann and Muller, 2015). In the case of a limitation of these six DOF, the kinematics of the knee joint is altered, leading also to an altered CP. In addition, depending on the horizontal or vertical alignment of the knee joint during the experiments, the peak CP might also be altered. While the horizontal knee alignment is more likely to eliminate gravitational effects in the flexion-extension plane (Amis et al., 2006), because of its anatomically correct knee joint alignment, the vertical alignment reflects a more physiological load bearing situation. To obtain realistic kinematical parameters that are comparable to the in vivo situation, conventional or modified material testing machines, Oxford-rigs, horizontal rigs and wear simulators are used (Supplementary Material). Using conventional material testing machines (Agneskirchner et al., 2007; Marzo and Gurske-Deperio, 2009; Van Thiel et al., 2011; Forkel et al., 2014; Zhang et al., 2015; Brown et al., 2016), the application of an axial load on a vertically aligned knee joint can be readily implemented. Modified material testing machines with customized loading setups are normally able to provide all six DOF during in vitro experiments (Becker et al., 2005; Rodner et al., 2006; Schillhammer et al., 2012; Kim et al., 2013; Bryant et al., 2014; Goyal et al., 2014; Lee, 2014). The main components of custom-made vertical knee joint simulators, so-called Oxford-rig simulators (Mcculloch et al., 2016; Steinbruck et al., 2016), are ankle and hip assemblies that also enable an unconstrained movement of the knee joint in all six DOF (Zavatsky, 1997). In Oxford-rigs, a horizontal actuator normally simulates the flexion-extension movements without simulating an according axial or ground reaction force. Another type of the experimental setup is a horizontal rig (Stukenborg-Colsman et al., 2002; Ostermeier et al., 2006; Hofer et al., 2012), where the specimens are mounted with the femur or the tibia fixed horizontally and the patella facing up- or downwards (Stukenborg-Colsman et al., 2002). Movement of the tibia or femur is achieved either passively by an actuator or actively by means of active flexor and extensor muscle simulation, such as quadriceps or hamstring muscle. Additional muscle forces can be also simulated with dead weights. Lastly, CP can be investigated using a knee joint wear simulator (Bedi et al., 2010), which is normally used to examine different knee replacement designs. Such commercial simulators focus on long-term gait simulations, providing six DOF and flexion angles of up to 60°, while the flexion angle, axial force, anterior-posterior shear force or internal-external torque can be varied (Walker et al., 2000; van Houtem et al., 2006) accordingly.

Of course, the kinematical configuration of a test setup plays a crucial role when investigating the peak CP, where it is either possible to simulate dynamic (e.g., gait cycles, drop jump, etc.), static (e.g., fixed knee flexion angle) or so-called quasi-static (e.g., very slow simulation of predefined flexion-extension cycles) loading situations. Several groups used fixed knee flexion angles to simplify the testing apparatus (Marzo and Gurske-Deperio, 2009; Van Thiel et al., 2011; Forkel et al., 2014; Brown et al., 2016). The fixed angles and pre-defined axial loads allow groups to investigate research questions and their effects on knee kinematics without additional expense or new machine investment. Complex setups like Oxford-rigs, by contrast, offer greater dynamics and the simulation of muscle forces and motion patterns (Bryant et al., 2014; Mcculloch et al., 2016; Steinbruck et al., 2016; Schall et al., 2019).

An adequate load application is essential when attempting to simulate in vivo knee joint loading conditions and, therefore, CP values at a realistic pathological or physiological level. Loads can be applied either directly via the bony structure (isolated axial load) by the loading apparatus, indirectly by muscle force simulation, or using a combination of both.

The results of our systematic literature review indicated, that an injured meniscus, for example, meniscal tears, leads to significantly increased CPs within the knee joint. The peak CP values increase up to 235% in total meniscectomized knees compared to their intact meniscal state (Baratz et al., 1986). The analysis of the published peak CPs showed a wide range from 0.1 MPa (van Egmond et al., 2017) to 12.06 MPa (Poh et al., 2012). Surgical treatments like sutures or meniscus replacements are able to restore the CP almost to the initial intact state. Degenerated or pathologically altered menisci lead to increased peak CPs and their repair is necessary to ideally restore the native CP distribution. This is essential because it is widely accepted that meniscal alteration, and thus an increased peak CP, is one crucial factor initiating the early onset of PTOA (Badlani et al., 2013; Thomas et al., 2017). Therefore, one primary treatment goal of a meniscus pathology is to restore the CP close to the intact state and ideally rebalance the load distribution between the lateral and medial compartments.

In summary, post-traumatic changes of the meniscus significantly increase the CP in the knee. Tears and partial or total meniscectomy treatments lead to increased CP in the knee joint, whereas repairing techniques, like meniscus sutures, are able to restore the CP almost similar to the intact meniscal state. The publications evaluated in this review widely coincide in their outcomes, particularly after a meniscal injury when the peak CP tendentially increases, whereas it can be restored by surgical treatment. Moreover, the published CP values showed clear trends after an injury or its treatment.

However, it is also clear from the present study that the absolute values of the reported CPs cannot be compared. The different approaches of the research groups make it difficult to determine the “truth” of the CP. Therefore, it more practical to compare the relative values, while real values (in MPa) are more likely to lead to non-comparable values, particularly because of the large number of different test setups and simulations. What has become clear, however, is the fact that untreated meniscal injuries inevitably lead to a significant increase in CP and in the long term are highly prone to progress to PTOA.

As indicated above, large differences were found in the application of loads and the alignment of the knee joint specimens in different test setups. It could be demonstrated that greater applied axial loads lead to a higher peak CP and muscle force simulation resulted in a more physiological knee joint loading condition with an accordingly higher peak CP. Muscle force simulation leads to a more realistic environment of the knee joint in in vitro testing. On the one hand, the knee is loaded multidimensionally, and on the other hand, the knee is held and guided more stably in the test environment. This results in a more physiological outcome compared to pure axial loads. The test setups and their respective load simulations in in vitro peak CP measurements indicated a remarkable axial loading range from 0 to 2,700 N. It appears clear that greater axial loads and applied muscle force simulations lead to higher CP, which makes it difficult for an objective comparison of the so gathered CP values. Values obtained through the different research approaches cannot be compared with each other. This can be seen, for example, in the comparison of the two studies with similar test setups by Agneskirchner et al. (isolated axial load 1,000 N) and Inaba et al. (isolated axial load 2,700 N), with peak CPs of 1.72 MPa (Agneskirchner et al., 2007) vs. 4.01 MPa (Inaba et al., 1990) shown in the intact knee state with normal alignment. Furthermore, Huang et al. demonstrated that pressure differences cannot be scaled up from lower load scenarios to higher ones (Huang et al., 2003). Rather, the trend of the peak CP outcome in each study needs to be evaluated, as well as its general research question. This can be used, for example, to determine the success of a treatment strategy, but not to investigate questions in which the magnitude of the peak CP is important. This could be the case, for example, when investigating the stability of load-critical replacements or implants.

In addition, lower limb varus malalignment leads to an increased medial CP, which is further considered as a promoter of premature OA (Inaba et al., 1990; Bruns et al., 1993; Agneskirchner et al., 2007; Willinger et al., 2020). The limb alignment must be considered in each test, because this elementary parameter can have a major effect on the result by increasing or decreasing CPs unintentionally. Therefore, care must be taken to ensure that the clamping is performed in a native manner without placing the knee unintentionally in a malalignment and thus influencing the peak CP.

Therefore, we propose recommendations for future in vitro testing of knee joint contact mechanics. This was done to obtain the most realistic in vivo data possible from testing in addition to the already gathered significant trends. This is also intended to make biomechanical studies more comparable, objective and consistent in approach and findings.

A biomechanically based recommendation regarding the experimental procedures for in vitro investigations on CP of the knee joint focusing on meniscus modifications is presented in Table 2, including the requirements for the test setup, static or dynamic testing methods, magnitude and direction of the applied loads and muscle force simulation. Here, the physiological vertical knee alignment is the key factor. The test setup should be extended by the knee-stabilizing quadriceps and hamstrings muscles and an axial load adapted to the specimen (in vivo weight simulation). For the kinematic examination, normal daily exercises with flexion and extension (e.g., gait cycle) should be selected (Huang et al., 2003).

Due to the time-dependent responses of meniscal tissue and its dynamic environment within the knee joint, motion of the knee joint may be better suited to investigate CP with meniscal tears or after meniscus resection (Goyal et al., 2014). The range of motion and the velocity of the motion should be as close as possible to the physiological movement of a human knee joint seen during activities of daily life. Ideally, the axial load should be either combined with a sufficient muscle force simulation or alternatively tibiofemoral contact forces should result from such an appropriate muscle force simulation. The major knee spanning muscles - quadriceps and the hamstrings - are particularly important for the stabilization of the knee joint and should be simulated. The preferred test setups are horizontal rigs, Oxford-rigs and modified material testing machines, in particular Oxford-rigs and material testing machines, because they may have advantages due to their upright knee joint position. The axial load should be related to the biometric data of the donor of the knee joint specimen to allow reasonable in vivo-like weight bearing simulation. Static knee joint loading should ideally be avoided, because the highly anisotropic and inhomogeneous viscoelastic menisci show a time- and localization-depending behavior. The latter is mainly determined by the interaction of the menisci with their surrounding soft tissues within the knee joint. Therefore, we recommend to rather simulate dynamic loading situations with varying knee joint flexion angles as seen during activities of daily life, like e.g. normal gait or stair climbing. However, dynamic testing is not possible in every research institution. If dynamic testing, as suggested by us, is not possible, static tests with different axial load profiles and different flexion angles should be performed. Additionally, the axial load and limb alignment should be adapted to the donor as a minimum requirement.

Considering this, it may lead to greater transparency and, therefore, an increased knowledge in the understanding of knee joint contact mechanics. When consistent test setups and test procedures are introduced and implemented by research institutions, different fields can thereby be investigated and a wide-ranging field of interest with a high level of knowledge exchange could be obtained.

At the very least, the description of the experimental procedure should be presented in a reproducible and thus fully transparent way. The description of the test setup, test procedure, load application, and muscle force simulation should be made very precisely to be able to reproduce tests and their results. Such a description should at least incorporate the following terms: Description of test setup and included modifications, DOF at all bearings, fixation of femur, and tibia, actuators, sensors, and the used control mode; for static testing all tested flexion angles should be mentioned; for motion cycles, both, the range and velocity of the motion should be indicated; direction and magnitude of the applied loads must be known. Information about muscle simulation should include the simulated muscle groups, simulation type with weights or actuators (control mode), magnitude and direction of the simulated muscles and a description of the change of magnitude over time or flexion angle.