Study approaches that do not include muscles may not adequately predict injury mechanisms and risk factors of non-contact ACL injury. Omitting muscles leads to inaccuracies since the forces transmitted to the ligaments and bones are dominated by the muscle forces [38]. It appears that ligaments can only play a limited role in maintaining the integrity and stability of the knee joint. In fact, it is likely that ligament rupture occurs only when muscle action fails to protect them. Muscle recruitment strategies may also be one indicator why females suffer more non-contact ACL injuries than men [39-42], and strongly points to the importance of musculo-tendon contributions in maintaining knee joint stability and integrity. Hence, if for any reason the firing pattern and the magnitude of the load by which the muscle pulls on the bones of the knee are improper, then the ACL will then have to take the brunt of the load and in some instances will rupture.

In-vivo studies [43] have concluded that ACL strain increases if there is an increase in quadriceps activity relative to hamstring activity. McConkey [44] are the first to describe eccentric quadriceps contraction as the intrinsic force responsible for ACL injury. Numerous in-vitro studies using defect free cadavers have also focused on the effects of increased quadriceps forces on ACL strain [45, 46] and concluded that high unopposed quadriceps forces induce ATT, and consequently increase the strain in the ACL over a specific range of knee flexion angles. One in-vivo study with healthy men and women (using 11 subjects) aged 21-42 years and with normal ACL [47] showed similar results. However, one in-vitro study using six pairs of human cadaveric knees [48] found that the quadriceps muscle force protected the ACL from injury. As well, one study demonstrated that under non-physiological loading without any ground reaction forces (GRFs), the knee locked in one position, and static quadriceps loads of 4500 N applied, ACL injury did occur [49]. The authors suggested quadriceps drawer as an injury mechanism. Hamstring muscle force applies a posterior directed force component and has been reported to strain shield the ACL by reducing ATT at large flexion angles [50]. An early study by Solomonow et al. [51] suggests that strength training of the hamstring muscles can help prevent damage to the intact ACL. On the other hand, a later study by Shelburne [52] demonstrated that the hamstring cannot apply large enough posterior forces to unload the ACL especially at low knee flexion angles- a position where many non-contact ACL injuries occurs. It has also been demonstrated by Simonsen et al. [53] that the ability of the hamstring muscles to reduce the ACL load is marginal.

The results of an in-vivo study by Fleming et al. [54] demonstrated that the gastrocnemius muscle is an antagonist to the ACL. This is in agreement with an analytical study by Pflum et al. [55]. On the other hand, a cadaveric study by Durselen et al. [56] and a theoretical investigation by Shelburne [57], both demonstrated that the contraction of the gastrocnemius muscles did not strain the ACL over the entire range of knee flexion.

The discrepancies among these studies highlight the need to determine the contribution of muscle activation and loading on ACL loading. Nonetheless, many studies clearly demonstrate that ACL loading mechanism is affected by muscle activity and that the muscles must be included in these investigations. Moreover, considering the limited strength of the ACL, active control of the knee joint depends on the balance of the resultant muscle forces.

The majority of non-contact ACL injury studies do not address the effects of hip and ankle kinematics and kinetics on injury. In addition, the majority of studies in the literature do not account for the effects of whole body movement on ACL loading. The inclusion of the ankle is important in non-contact ACL injury studies as it has been shown that the interaction of shoe surface interface with the ground is an important risk factor in non-contact ACL injury [24]. An increase in the coefficient of friction between the shoe and playing surface may increase the traction which can cause the foot to catch or stop inadvertently during play. Such types of events are associated with higher risk of ACL injury. By one study conducted using soccer players with varying cleat lengths, it was also shown that foot contact with the ground is an important risk factor in non-contact ACL injury [58]. Lambson et al. [58] also demonstrated that shoes with more cleats results in higher torsional resistance at the foot-surface interface which can result in increased risk of ACL injury. As well, muscle activity across the ankle controls the position of the foot at landing, which most likely influence the loading at the ankle [59]. Moreover, muscular activity at the ankle may influence the loads that are transferred through the ankle to the knee [59, 60]. Pflum et al. [55] found that the forces applied by the ground to the foot have a major impact on peak forces seen at the ACL. Another study found that when the foot was not flat, the ground reaction forces (GRFs) cannot be transmitted effectively through the bones to the ground without the actions of muscles [61]. In addition, excessive foot pronation was shown to increase the likelihood of non-contact ACL injury [62].

Inclusion of the hip articulation is also important to the understanding of how and why the ACL gets injured since hip movement is known to affect the loading on the knee [40]. Because the upper body contains over half of the total body mass, the trunk and pelvis position will have coupled effects on knee angles and resultant ACL strain. The position of the leg at the time of non-contact ACL injury displays tibial rotation, apparent knee valgus, foot pronation, and a relatively extended knee and hip [4, 14]. Muscle contraction at the hip can also affect knee loading because the hip transfers the upper body loads to the leg [6]. In addition, muscle contraction at the hip can also affect energy absorption [63, 64]. Moreover, a cadaveric study [19] and an analytical study [65] both demonstrated that if the hip movement is restricted, ATT will result and ACL injury can easily occur. Some other studies also suggested that the knee is one part of the kinetic chain and that the torso, hip, and ankle may also contribute to ACL injury [40, 66]. The importance of including the hip and ankle joints in any non-contact ACL injury study is supported by many other studies which are captured in a review article by Hewett et al. [30]. So, one can conclude that non-contact ACL injury is a whole body event that requires the study approach to capture the contributing effects of ankle and hip articulations on ACL loading.

Specific contours of the knee joint geometry, for example, the femoral notch [67] has been implicated as a risk factor for non-contact ACL injury. The complex geometry of the knee joint aids in stability. The femoral groove that the patella slide on, the femoral condyles resting on the tibial eminence [68] and the posteriorly sloped tibial plateau [69] are all factors that may affect loading seen at the ACL during various non-contact ACL injury mechanisms. There are many computational studies that overlook the contribution of these factors to ACL injury. In fact, Liu [70] found that articular surface geometry, ligament locations, and ligament insertion sites had a pronounced effect on the modeling outputs. Irrespective of muscular activity, Rentrom et al. [71] demonstrated that the ACL is subjected to an inherent increase in strain as the knee extends owing to the geometry of the articular surfaces of the knee. Despite this, many studies have not considered accurate three dimensional (3D) hard and soft tissue geometries and few have investigated the effect of geometry changes on loads seen at the ACL [72-74]. The inclusion of accurate 3D tissue geometries appears to be a crucial non-contact ACL injury study approach prerequisite. Without an objective view of these parameters in problem definition, a clear and comprehensive understanding of injury mechanisms and risk factors of non-contact ACL injury may remain elusive.

It can be argued that historically research on the human musculoskeletal system was focused on empirical methods. One form of in-vivo testing entails the direct implantation of a transducer to subjects’ tissues. These are termed contact methods and have the advantage of simultaneously including many risk factors, forces, the effects of active muscles, as well as weight-bearing loads. A significant challenge with this form of in-vivo testing is that it is invasive. Ethical policies controlling testing with humans make in-vivo testing extremely difficult. Moreover, because of the size of the sensor employed, size of the ACL, intraarticular location of the ACL and intercondylar notch, in-vivo studies to-date can only measure strain on the anteromedial bundle of the ACL for movements confined to the sagittal plane [76]. Measurements are restricted to linear displacement at discrete locations, knee flexion angles approximately greater than 15 degrees, and movements only in the sagittal plane [47]. In addition, many of these studies look at movements that are non-ballistic. Woo et al. [77] have also argued that since strain gauges and other similar type transducers are implanted into the ligament, they can alter the ligament length, and subsequently ligament force. Another complicating factor with in-vivo techniques is the fact that specific bands of the ligament are tensioned at different portions of the loading cycle. Another form of in-vivo testing includes non-contact methods. These methods do not require the implantation of a sensor into the subject’s ACL. Non-contact in-vivo methods include, but are not limited to, the use of dynamic magnetic resonance imaging (MRI) [78], bi-plane fluoroscopy [50, 79-82] and video dimension analyzer [83] to capture tissue kinematics at various knee flexion angles. These techniques do not provide any details of the forces seen at the ACL during movement. Another, non contact method include optical techniques such as Roentgen Stereophotogrammetry [84, 85]. Various contact techniques have been used in the past to measure in-vivo ACL strain: implantable extensometers [47, 71, 86-91] and buckle transducers [92-97]. It may be argued that none of these techniques are truly in-vivo given they alter in one form or the other the way the subject moves during testing.

In-vitro testing is conducted outside of the body typically with human subjects or post mortem human subjects (PMHS)/ cadavers. The vast majority of studies investigating non-contact ACL injuries are in-vitro [19, 77, 98-100]. The major advantage of in-vitro testing is its utilization of biological tissue for testing. In-vitro testing also has the capability to simulate knee kinematics and muscle loads. However, in-vitro studies using cadavers may not accurately describe ligament function in-vivo since loading applied to the cadavers during experiment is different from that applied by the muscles during activity. Other challenges with in-vitro studies using cadavers are the inability to simulate realistic muscle activation and the difficulty in obtaining repeatable results. Furthermore, replicating an isolated ACL injury in-vitro is difficult, and has been achieved with only limited success [19]. This shortcoming may exist because cadaveric knee studies lack the stability and control provided by active knee musculature [101]. It is also difficult to apply large muscle forces seen during normal physical activities such as landing from a jump to in-vitro experiments using cadavers. This is partly due to the fact that many cadavers are from the elderly population. Another prevalent in-vitro study method is gait analysis that employs skin markers. The use of skin markers with gait studies have been shown to induce significant errors in predicting in-vivo ligament behavior [102-104]. Since some gait studies are conducted in the laboratory setting, there is also some level of uncertainty with respect to the ability of these tools in capturing true human response. Given the subject is aware that the camera is focused on them or their feet have to strike a force plate, it can be argued that their body response may be somewhat altered to fit the event. As a result, the data collected may not be a true representation of human motion. Despite its shortcomings, in-vitro studies have the capability to provide much freedom to investigate function and behavior of the ACL. From this standpoint, it can be argued that gait analysis is the only way available today to determine the kinematics and kinetics during activity to cause non-contact ACL injury.

There are mainly three approaches to computational modeling in the current literature: mathematical, finite element (FE), and musculoskeletal rigid body (RB) modeling. A computational model of the knee joint is a graphical representation of the joint anatomy that can be manipulated. Computational modeling have become popular partly because technological restrictions and ethical considerations have prevented the direct measurement of the ACL strain in-vivo. Many mathematical models of the human knee focus on the intact ACL. With many of these models, if injury is studied, it is done primarily by removing the effect of the ACL in the model predictions [105]. Even though mathematical models have assisted us in gaining a better understanding of the mechanics of the knee, the gross approximations and assumptions made by these models can readily be addressed with technology and information available today. Two such technologies are FE and RB modeling. Mathematical models have become less popular over recent years perhaps due to: (1) Easy access and lower cost of FE and RB modeling software; and (2) Inherent complexity and nonlinearity in the governing equations related to ACL biomechanics, and subsequently the challenges entailed in developing and verifying these models. Hefzy [106] provided an excellent review of mathematical models of the human knee joint found in the scholarly literature.

The first application of finite element analysis (FEA) to biomechanics was in orthopedics [107]. The most extensive application of FEA to knee biomechanics has been in artificial joint design and fixation. Two studies [108, 109] provided a brief review of the application of FEA to biomechanics. Andriachi et al. [110] was the first to develop a knee joint model for the determination of the forces using FE methods. Once verified and validated, a FE model can provide greater capabilities to answer many what if questions over mathematical models. Finite element models also have the advantage of being able to calculate parameters that are difficult to measure experimentally. The literature indicates that only a small number of FE models are used to study ACL mechanics, and none has focused on predicting mechanisms and risk factors to non-contact ACL injury (see for instance [72, 73, 111-128]).

Many of these FE studies do not report data on model verification and validation, thus limiting their use in clinical applications. Other FE studies investigating ligament mechanics in the literature modeled only the ACL [112, 129, 130]. None of these studies investigated non-contact ACL injury. Nonetheless, the challenge with FE modeling of biological tissues include the complexity of modeling material responses under loading, the level of discretization required to capture complex and intricate anatomical geometries, long computational time to converge [131], and dependence on empirical data for validation among others. Increased and affordable computing capacity and more sophisticated (nonlinear and 3D) FE software packages has allowed more realistic modeling and the application of iterative procedures to describe time-dependent biomechanical behavior. Nonetheless, this may not be enough to fully address the challenges posed by the complexity of predicting non-contact ACL injury mechanisms and risk factors.

Musculoskeletal RB modeling of the knee is based on research aimed at developing a dynamic rigid body model for balancing internal forces with externally applied forces to produce motion [132-134]. Three main approaches may be used to balance internal forces with externally applied forces for a specific motor task: inverse dynamics; forward dynamics; and optimum control theory. Erdemir et al. [88] provided a thorough review of these three approaches. The main advantage of musculoskeletal RB modeling is that it enables us to determine the forces in the muscles during activities implicated to cause non-contact ACL injuries. However, musculoskeletal models cannot provide details of the loads and stresses in the hard and soft tissues, like FE models can. In addition, RB musculoskeletal models can require extremely long computational time to converge to a solution and in some cases requires parallel computing [135].

Based on this investigation, it is understood that the type and complexity of the research method employed depends on the question(s) posed. Many studies have pointed out that several intrinsic and extrinsic factors are responsible for non-contact ACL injuries [30, 138, 139]. Other studies have categorized risk factors into four areas: environmental, biomechanical, neuromuscular, and hormonal [40]. However, many studies focused on investigating the effects of a single risk factor by comparing intrinsic or extrinsic differences between males and females, such as Q-angle [140], intercondylar notch width [141], and hormones [142] without considering the combined effects of these and other factors implicated to cause injury. It seems that the multitude of intrinsic and extrinsic risk factors involved during sports make focusing on one factor difficult. In addition, some studies address the effects of only one muscle group on ACL rupture. It seems quite unlikely that a single risk factor or only one muscle group will be responsible for ACL injury. There are few studies, to the best of our knowledge, where a tool or methodology is employed that lends itself to simultaneously determining the effects of numerous parameters on injury mechanisms and risk factors of non-contact ACL injury. Finally, the low annual incident rate for ACL injury in the general population of 1 per 3000 people [143] –and even smaller in the athletic population– has compelled researchers aiming at pinpointing factors contributing to risk of ACL injury to adapt a reductionist approach by focusing on a single factor or few factors due to small sample size. Given this, studies to date are narrowly focused due to statistical power limitations.

Combining and comparing results from separate studies that use similar research methods can be valuable, but differences in, for example, specimen type and data reporting may prevent drawing solid conclusions about the results. The challenge is these differences may exist due to the lack of guidelines on study method, instrumentation to employ, and procedures for data analysis. The dearth of standards and specifications in the field of biomechanics is believed to be one reason why dialogue among research groups and comparisons in experimental studies remains challenging, debatable, and sometimes with no solid outcome. A few attempts at standardization that have led to considerable benefits in the research community include the Visible Human Project [144], VAKHUM [145], and the standardized femur [146]. These endeavors have aided in simplifying cross validation of research results. Due to the lack of standards and specifications in study approaches, accumulating knowledge on injury mechanisms and risk factors of non-contact ACL injury is believed to be slowed. As well, many of these studies are too highly focused and are not powered to handle the contribution of multiple factors on non-contact ACL injury.