Length-change patterns of the medial collateral ligament and posterior oblique ligament in relation to their function and surgery

The medial ligament structures of the knee, including the superficial medial collateral ligament (sMCL), the deep medial collateral ligament (dMCL), and the posterior oblique ligament (POL) within the posteromedial capsule (PMC), work together to restrain valgus and internal/external rotatory loads [27, 28, 33]. The MCL is the most commonly injured ligament of the knee and ruptures may occur either in isolation or in combination with damage to other ligamentous or meniscal structures [1]. Non-surgical treatment, principally bracing, rest, and rehabilitation, is the primary treatment for grade I- and II-MCL strains. However, a proportion of grade III injuries, implying a rupture of the three MCL structures [24], knees with persistent valgus instability despite non-surgical treatment [25, 35], and some dMCL lesions with chronic pain [21] require surgery.

Many MCL reconstruction techniques have been proposed [6], which implies that none is ideal. However, due to inconsistent anatomical descriptions, reconstruction techniques differ in the number and the site of attachment points, the number of graft bundles, and graft tensioning protocol [3, 15, 19, 20, 23]. These differences reflect a lack of knowledge of the attachment sites and the resulting length-change patterns of the natural and reconstructed soft tissues. Perhaps, because of this, the present authors find the MCL the most difficult of the knee ligaments to reconstruct reliably. Registry data [31] have found that when an MCL injury in association with ACL rupture is treated conservatively, the likelihood of ACL graft failure is increased. These observations suggest the importance of improving the treatment of medial soft-tissue injuries.

Although injuries of the dMCL contribute significantly to anteromedial rotatory instability (AMRI) [14, 29], established MCL reconstruction techniques do not restore the function of the dMCL. Furthermore, some previous biomechanical tests of MCL reconstruction have been performed, while the dMCL was intact, so they could not have reproduced the grade III injury pattern that is usually required to indicate surgery [5]. A correct understanding of the behaviour of the medial knee structures would profoundly help surgeons to improve surgical techniques for treating medial instability.

The purpose of this study was to define the length-change patterns of the medial knee structures (sMCL, dMCL, POL) at their anterior and posterior borders, across the knee flexion range and with applied anterior and rotational loads.

It was hypothesised that:

Following (Imperial College Healthcare Tissue Bank) ethics approval, ten unpaired fresh-frozen human cadaveric knees (8 male and 2 female) with an average age of 61 (range 51–69) years were used. All specimens were stored at − 20 °C and were thawed 24 h at room temperature before preparation. They were confirmed to have no evidence of previous surgery, nor abnormal laxity, nor misalignment by visual and manual examination by an orthopaedic surgeon (LW). Knees were kept moist with occasional water spray during the entire test.

The preliminary test of five specimens confirmed that the osteotomy of the tibial sMCL attachment had no significant effect on sMCL length-change measurements. It revealed a maximum length-change deviation of 0.4% for the anterior sMCL at 100° and 0.6% for the posterior sMCL at 80° of knee flexion (n.s.) compared to the knees prior to elevating the bone block.

The native length of each structure at 0° knee flexion was: 111 ± 6 mm for the anterior sMCL, 117 ± 6 mm for the posterior sMCL, 40 ± 4 mm for the anterior and posterior dMCL, 46 ± 4 mm for the anterior POL, and 48 ± 6 mm for the posterior POL.

The most important finding of this study was that the ligaments at the medial aspect of the knee differ in their length-change responses to knee flexion, and to external applied loads. The MCL and POL are not isometric—the situation is complex. In particular, structures attaching posterior to the medial epicondyle—the posterior fibres of the sMCL, the posterior fibres of the dMCL, and both the anterior and posterior fibres of the POL—shorten/slacken with increasing knee flexion. Conversely, the anterior fibres of the sMCL lengthen during knee flexion, since they are attached anterior to the axis of knee flexion, which is at the medial epicondyle, as had been hypothesised. The anterior part of the dMCL is slack in early flexion, since its femoral attachment is distal and posterior to the medial epicondyle and lengthens in deeper flexion once its femoral attachment moves relative to the flexion axis. Application of an external rotation torque to the tibia significantly lengthens the anterior fibres of the dMCL throughout the flexion arc. The anterior and posterior fibres of the POL slacken rapidly with knee flexion, while tibial internal rotation lengthens them, but the rotation effect does not cancel the slackening that occurs with knee flexion except near to knee extension. These length changes with tibial external and internal rotation supported the initial hypotheses of the present study.

A novel finding is that when the sMCL is intact and the knee is near extension, the dMCL and adjacent joint capsule are slack in neutral tibial rotation; releasing the sMCL leads to approximately 3 mm joint line opening at 0° and 5 mm 30° knee flexion with a valgus load applied. In addition, whilst most descriptions of dMCL anatomy suggest a vertical alignment in the sagittal plane, the present study describes, for the first time, that the dMCL passes distally and anteriorly to create obliquity in its alignment which is ideal for resisting tibial external rotation.

These descriptive findings can be applied to the clinical focus on how best to restore normal stability and function to the injured knee, in particular with AMRI, which may also be associated with ACL injury.

As the knee flexes, the femoral attachment of the sMCL rotates around the medial epicondyle, resulting in a lengthening of the anterior fibres and a shortening of the posterior fibres. The strain patterns of the sMCL fibres were, therefore, primarily dependent on the femoral attachment site with regard to the centre of rotation at the medial epicondyle. An anterior translation (draw) force did not affect the sMCL length-change patterns, while the ACL, the primary restraint, was intact. However, the anterior sMCL fibres were lengthened by tibial external rotation, while the posterior fibres were lengthened by internal rotation. These findings agree with the previous studies that have similarly described different length-change patterns depending on the femoral sMCL attachment sites [2, 9, 13, 22, 32, 33]. Arms et al. [2], using a strain transducer, found the highest strain at the proximal anterior sMCL fibres as the knee flexed and a decrease of strain at the posterior fibres at the same time. A similar strain pattern was reported by Gardiner et al. [10] who used optical strain measurement. Furthermore, fluoroscopic studies in vivo agreed with these findings and described an elongation of the anterior part of the sMCL [13, 22]. In contrast with earlier works and the present study, Feeley et al. [8] reported a slackening with flexion for all MCL graft attachment points studied. They also reported the sMCL attachment to be posterior to the medial epicondyle, as did Laprade et al. [18], which differs from the present observations (Figs. 1 and 2). Victor et al. [32] found that an MCL fibre attached proximal and posterior to the medial epicondyle was near-isometric, whereas the reciprocal length-change patterns of the anterior and posterior fibres of the sMCL in this present study indicate that the epicondyle will be isometric.

The results of the present study are very relevant to clinical practice. Regarding sMCL reconstruction techniques, the present data suggest that a femoral attachment at the centre of the anatomical attachment, enveloping the medial epicondyle, would lead to the most isometric behaviour (Fig. 1). Graft positioning posterior to the medial epicondyle, as recommended by other authors and commonly practiced [10, 19], would lead to a slackening of the graft with knee flexion, and thereby be likely to compromise clinical results.

The femoral attachment of the dMCL is just distal and slightly posterior to the sMCL attachment and its fibres fan out to its wider tibial attachment. This present work emphasizes that the anterior fibres of the dMCL run anterodistally from the femur to the anteromedial tibia (Fig. 2). This oblique orientation means that the dMCL is well oriented to withstand tibial external rotation, and the fibres can be clearly identified under tension, whilst tibial external rotation is applied. The shape of the dMCL and the obliquity of the anterior portion have not been described previously. Earlier anatomical studies, which have usually shown the dMCL as having a purely proximal–distal orientation in the sagittal plane may have described the posterior portion—the oblique anterior fibres become clear when the tibia is in external rotation. The present results showed a large lengthening strain of the anterior dMCL when the tibia was externally rotated. It was also observed that in neutral tibial rotation, the dMCL was loose, while the sMCL was intact, with an average of 5.0 mm slack at 30° knee flexion. We hypothesise that this dMCL laxity might reduce the risk of dMCL rupture, given that its short fibres can withstand less elongation to failure than those of the longer sMCL [26], and accounts for the dMCL’s main function of resisting external rotation [4, 11, 27] and not valgus stress, against which the sMCL provides the most important restraint [12]. It also explains the minimal contribution of the dMCL to restraint of valgus load, while the sMCL and PMC are intact [27]. These findings suggest that dMCL injuries may occur through a mechanism involving tibial external rotation [14], which also corresponds clinically to dMCL injury [21]. Presumably, since the dMCL and sMCL are commonly injured together a combination of external rotation plus valgus can be the injurious mechanism as well as major pure valgus. In the sporting scenario, the latter mechanism is common in rugby and American football but the former in football (soccer) [21] and alpine skiing.

To date, acute dMCL injuries are solely treated non-surgically and no reconstruction technique has been described. However, isolated injuries of the dMCL lead to a significant increase of tibial external rotation near knee extension, and could, therefore, possibly result in AMRI of the knee [4, 14, 29], particularly in footballers who impose tibial external rotation loads with certain kicking techniques and sudden change of direction. In footballers, dMCL injuries can lead to failure of healing and disabling pain, which does require surgery at times [21]. Typically, in this scenario, the original injury is relatively minor allowing rapid return to activity such as running, but perhaps at the cost of inhibiting healing, and development of an inflamed region of dMCL—usually close to the femoral attachment. The footballers concerned only have pain when the dMCL is stressed. Hence usually they ‘can do everything except- “bend” a ball’. To do that, the player’s foot addresses the ball in a way that jerks the knee into external rotation. It is a manoeuvre that a player uses so frequently that the pain associated with it is disabling. If the dMCL injury is noted on MRI, it is, therefore, good to advise the player and medical team to rehabilitate more slowly avoiding external rotational load (for example, walking for 4 weeks with the foot straight ahead rather than the natural 10 degrees or so of external rotation during stance phase of gait), and most do heal. However, persistence of symptoms at 12 weeks may be a reason to offer surgery, which involves exposure of the dMCL lesion (including splitting what appears to be a normal sMCL), its suture repair and a layer-by-layer closure of the sMCL (layer 2 [34], and deep fascia (layer 1 [34]).

The posteromedial capsule (PMC) includes fibres that are often described as the posterior oblique ligament (POL). The anterior POL fibres slacken immediately with knee flexion, and the posterior fibres slacken even more, demonstrating that the more-posterior the femoral attachments, the greater the slackening with knee flexion. This accords with the previous findings that the main contribution of the POL is to resist valgus loading at and near full knee extension [27]. The present work has found that the POL becomes taut when tibial internal rotation is applied at flexion angles up to 40°, beyond which the slackening with knee flexion causes it to be ineffective. Hence, the POL inhibits tibial internal rotation of the extended knee [27]. Although a knee trauma sustained close to extension including an internal rotation torque could injure the PMC/POL, it is a rare scenario where the POL is injured in isolation rather than with the PCL, for example.

The results of this study suggest that a POL reconstruction should be tensioned and fixed in full knee extension and neutral rotation to reproduce its anatomical length, and prevent over-tightening. Previous reconstruction techniques [15, 20] tensioned the POL graft at 30° and 60° knee flexion, respectively, but that would not recreate POL anatomical function. It would risk inability to fully extend the knee, and a fixed flexion deformity is a cause of great morbidity.

This study has limitations. The effect of applying a valgus moment to the knee was not studied despite resistance to valgus angulation being the primary restraining action of the MCL, leading to direct MCL elongation [12]. The applied muscle loads were only a small portion of physiological loads; they were used to maintain joint contact for the length-change measurements, and not to simulate a specific load-bearing activity. Although the sMCL only showed a flimsy attachment to the proximal tibia, disrupting it to access the dMCL could have led to small alterations in biomechanical behaviour, but it had no effect on the length patterns of the fibres of the sMCL. Similarly, the osteotomy of the distal tibial attachment of the sMCL might have altered its position when replaced for subsequent testing. This study was the first that investigated in vitro length-change patterns of the dMCL, while the sMCL and the PMC were intact and, hence, simulated an intact uninjured condition. The methods for length-change measurements and loading were similar to well-established test protocols [16, 30]. Length changes depend critically on the bone attachment locations, so a study of inter-and intra-observer errors of locating the femoral medial epicondyle was conducted by three of the authors across five knees: the mean inter-observer error was 1.2 ± 0.8 mm in anterior–posterior and 1.9 ± 1.3 mm proximal–distal with an overall ICC 0.851 (95% CI 0.644–0.956).

The anterior and posterior fibres of the sMCL act reciprocally either side of the femoral medial epicondyle: as the knee flexes, the anterior sMCL lengthens/tightens, whilst the posterior fibres shorten/slacken. The anterior fibres of the dMCL are oriented in an oblique antero-distal direction, so they are predominantly lengthened/tensioned by tibial external rotation. The dMCL is significantly loose in neutral rotation when the sMCL is intact. The POL is tight in knee extension, and further lengthened/tightened by tibial internal rotation, but it immediately shortens/slackens with knee flexion. This knowledge is critical to optimize reconstructive surgery and to understand the aetiology of injury patterns to the medial ligamentous structures.