The bone attachments of the medial collateral and posterior oblique ligaments are defined anatomically and radiographically

The superficial medial collateral ligament (sMCL), the deep medial collateral ligament (dMCL) and the posterior oblique ligament (POL)—a part of the posteromedial capsule (PMC)—are the medial ligamentous stabilisers of the knee against valgus and rotatory loads [2, 8, 10, 24, 32]. The MCL is the most frequently injured ligament of the knee [1] and can mostly be treated non-surgically with good clinical results [6, 12, 14, 23]. However, surgery is indicated in high grade MCL injuries and also when valgus instability persists in spite of conservative treatment [7, 11, 16, 19, 22] and with lesser degrees of laxity in combined ligament injury.

Orthopedic surgeons rely on the definition of the ligament attachment points to be able to perform and evaluate operations accurately. Inaccurate MCL reinsertion or graft tunnel placement would either cause ligament over-tension and over-constraint of the knee, or an insufficient and loose reconstruction, across the arc of knee motion [3, 35]. There are significant differences of isometry, or length-change behaviour, across the widths of each of the medial ligaments [34], so it is apparent that accurate positioning is important if the normal patterns of restraint are to be reproduced.

Medial knee anatomy has been described using several different methods. Besides anatomical dissection [20, 21, 25, 31], lateral radiographs [33] and computed tomography [26] have also been used to define the soft-tissue attachment points. Despite many previous publications, conflicting anatomical descriptions still exist. While many anatomical studies have found the femoral sMCL attachment on the medial epicondyle (ME) [13, 21, 25, 36], which would seem logical given that bony prominences usually correspond to soft tissue attachment sites, it has been described as completely separate, proximal, and posterior [20]. Definitions of the attachment points in relation to different bony landmarks given as absolute dimensions in mm do not account for the large range of sizes of the femoral condyle and proximal tibia across the population, so it is desirable to have normalised data. Furthermore, and surprisingly, despite its important function to restrain tibial external rotation [2, 4, 24], the attachment points of the dMCL have only ever been reported by Liu et al. [21] and Robinson et al. [25]. These differences and limitations demonstrate the need for more definitive guidance for the surgeon, such as where to position tunnels for anatomical MCL grafts, or to aid critical evaluation of those positions achieved in images post-surgery as quality control.

The purpose of this study was to provide measurements to define the medial ligament complex (that is: sMCL, dMCL and POL) bony attachments (‘footprints’) on the femur and tibia using both optical navigation and radiographic methods. These objective anatomical and radiographical data are described for the first time both in relation to useful surgical bony landmarks, and also normalised in relation to the overall sizes of the bones. These data will facilitate anatomical repairs and reconstructions that can restore physiological laxity and stability patterns across the arc of knee flexion.

Following approval from the Imperial College Healthcare Tissue Bank, Human Tissues Authority licence 12275, application R18027, 22 non-paired fresh-frozen human cadaveric knees (15 male and 7 female) with an average age of 47 (range 24–69) years were used. All specimens were stored at − 20 °C and thawed for 24–36 h before use. All knees were free of osteoarthritis and ligaments and menisci were intact; this was confirmed by inspection during the dissection process. Knees were kept moist with intermittent water spray during the entire test.

The femur and tibia were cut 200 mm from the joint line in the first 12 knees, then at 200 and 150 mm respectively in the remaining 10 knees. The fibula was cut and secured to the tibia in its anatomical position by a transcortical bone screw. Skin, subcutaneous fat, muscles, and the anterior capsule with the patella were removed, keeping the cruciate and collateral ligaments and the remaining capsule intact. Intramedullary (IM) rods were cemented into the tibia and femur to allow subsequent mounting in adjustable clamps.

In the first 12 knees a metal pin 1 mm diameter was inserted into the bone at the point judged visually and by palpation to be the femoral ME to allow radiographic measurement of its position.

In the remaining 10 knees the sartorius fascia (layer 1 of Warren and Marshall [31]) along with the semitendinosus and gracilis tendons were removed from their tibial attachments to visualise the sMCL and POL within the second medial soft tissue layer. The connecting fibres between the POL and the semimembranosus tendon were dissected to uncover the distal POL on the tibia proximal to the semimembranosus groove. The behaviour of the sMCL and POL fibres was observed through the range of knee motion to identify the border between the sMCL and POL. The femoral and distal tibial attachments of the sMCL were identified and each was defined by two radiopaque metal staples (6 mm long × 1 mm wide); these markers were inserted into the bone at the most anterior and posterior edges of each attachment (Figs. 1, 2). The attachments of the POL were likewise marked with staples at the most anterior and posterior edges of the femoral and tibial attachments. A tibial bone block (average 45 mm length × 28 mm width × 14 mm depth) including the tibial sMCL attachment was created, elevated and reflected en bloc proximally by sharp dissection along the margins of the overlying sMCL to expose the dMCL. Staples were inserted at the anterior and posterior edges of the femoral and tibial attachments of the dMCL. The tibial bone block was then replaced and fixed with two bi-cortical bone screws placed in pre-drilled holes over a 1 mm thick spacer to compensate for the saw cut bone loss. This held the distal sMCL in its anatomical relationship for digitisation and radiography. Using a linear displacement transducer it was confirmed by measurement that this procedure did not alter the length of the sMCL fibres significantly (≤ 0.5%) [34].

The test–retest reliability of the optical digitisation technique had an ICC of 0.994 (excellent agreement). When 3 examiners digitised the ME the mean intra-observer 95% CI of the mean was ± 0.6 mm in A-P and ± 0.5 mm in P-D directions, while the mean inter-observer differences were ± 1.8 mm in A-P and ± 0.9 mm in P-D directions. The mean intra-observer difference of radiographic measurements was 0.8%, equivalent to 0.5 mm across the width of the femoral condyle. The inter-observer reliability test of the radiographic analysis between two blinded examiners gave ICC 0.922—excellent agreement.

The most important outcome of this study is a quantitative anatomical description of the attachments of the medial knee ligaments in relation to the femoral ME and the tibial plateau, and radiological positions with respect to the ‘Schöttle point’ [27]. This knowledge is clinically very useful and practical to use during surgery. The normalised data allow us to relate the results to any knee size: the range of sizes of knees limits the value of descriptions of attachment sites with only mean dimensions. The accuracy and repeatability of the methods are demonstrated to be high. It provides a method for intraoperative identification and postoperative evaluation of the anatomical attachments and MCL graft tunnel positions. The isometric patterns of the ligaments, their repairs and reconstructions, depend on the exact femoral attachments, with significant differences even across the widths of each structure [34], so the data in this paper are a foundation for surgery aiming to restore normal knee behaviour.

Previous anatomical descriptions have sometimes been contradictory and do not allow for definite conclusions [13, 20, 21, 25, 26, 31]. Furthermore, whilst accurate intraoperative location of anatomical landmarks on the medial aspect of the knee is important it is often difficult by palpation alone. Therefore, in the operating room landmark identification is often best achieved with a combination of visualisation, palpation and radiographic localisation. This paper used the radiographic method of Schöttle et al. [27] to define the MCL attachments because of the lack of visibility of the ME in lateral view radiographs [33].

The study confirms previous reports [13, 21, 25] that the femoral sMCL attachment covers the ME. It seems logical for the sMCL to attach directly to the ME since soft tissues often attach to bony prominences. Saigo et al. [26] defined the femoral sMCL attachment similarly and found it at 47% A-P and 48% P-D distance. However, they defined a 0% datum at the anterior femoral shaft and not at the anterior border of the MFC, so the results of the two studies are very much the same. In contrast, one study [20] described the centre of the femoral sMCL attachment 5 mm posterior and 3 mm proximal to the ME, which is outside the mean limit of the attachment area in the present study. The exact femoral attachment site is important: if a graft is placed posteriorly it will slacken with knee flexion [34], or could be too tight in extension. The centre of the femoral attachment of the sMCL is at the same distance from the distal and posterior surfaces of the medial femoral condyle, which suggests that a reconstruction placed on the ME will be isometric. Even small alterations in position can make a big difference to the performance of a construct so it is imperative during surgery to check for isometry of a proposed construct by connecting guide pins placed at the proposed attachment sites on femur and tibia with a suture and taking the knee from full extension to high flexion.

The ME is on a relatively flat area rather than being a single localised prominence and so locating it is not easy. Studies using surgical navigation systems have found much variability in the identification of the ME [15, 17, 29]: Jerosch et al. [15] reported that it was found across a range of 22 mm in one knee, with mean inter-observer error of 10 mm. Thus, a radiograph of a guidewire positioned by palpation may be very useful. The radiographic method of Schöttle et al. [27] was used by Wijdicks et al. [33] to obtain A-P measurements similar to those reported in the present study, but with P-D measurements differing markedly by 6–9 mm. This may reflect variations in identifying the most posterior/proximal point of Blumensaat’s line.

This study found mean inter-observer differences of ± 1.8 mm in A-P and ± 0.9 mm in P-D directions, so there is little point in describing an anatomical measurement for surgical guidance to less than the nearest whole mm. While previous quantitative reports of the medial knee anatomy have not performed a repeatability analysis for detecting the ME [20, 21, 26], future studies should ascertain and report their precision. This study, and others [33], found that radiographic measurements are accurately repeatable by different examiners and therefore could be a valuable tool for clinical practice. It is critical to obtain a perfect lateral radiographic projection to avoid error, and surgeons need discipline to not accept radiographic projections that are ‘close to acceptable’!

An important novel finding is that the dMCL is fan-shaped, oriented antero-distally towards a 22 mm wide tibial attachment. Analogous to the oblique fibres of the “anterolateral ligament”, the dMCL acts as an important restraint against tibial rotation [2, 4, 24]. Tibial external rotation tightens the dMCL rapidly [25, 34]. Thus, it is ideally oriented to resist external rotation and anterior translation of the medial tibial plateau—one could argue like an ‘anteromedial ligament’! Robinson et al. [25] described the dMCL tibial attachment as 10–13 mm wide and 2–3 mm distal to the articular cartilage margin, while Liu et al. [21] reported it as 6.5 mm below the joint line. Both studies located the centre of the dMCL femoral attachment posterodistal to the ME/sMCL. Although early techniques in ACL surgery treated anteromedial rotatory instability by a pes anserinus transfer [28], those non-anatomical techniques have lost popularity. As no dMCL reconstruction to restore rotatory stability has been described, the present findings provide basic knowledge for further development. Excessive anteromedial rotatory laxity may be left unaddressed by surgeons because they lack the necessary operative techniques, but unaddressed MCL laxity is associated with ACL graft failure [11, 30].

The PMC has an extensive femoral attachment extending posterior from the sMCL attachment around the medial femoral condyle distal to the adductor tubercle. The POL, a distinct band within the PMC, attaches at a mean of 11 mm posterior and 4 mm proximal to the ME. This position is similar to that found in a previous CT study [26]. The POL has also been located 8 mm distal and 6 mm posterior to the adductor tubercle [20].

The present study has some limitations. Some of the small number of specimens were older than typical of MCL injured patients. However, MCL anatomy has not been shown to change over lifetime. Gross inspection and radiographs precluded osteoarthritis and joint line narrowing in all specimens, and thus measurements relative to the joint line were not affected. Even though three investigators were blinded to the others’ results when they identified the ME, this is subjective and there might be a difference in the interpretation by other surgeons. A strength of this study is the analysis of the precision of measuring anatomical locations by both optical digitisation and radiography of fresh specimens.

This study has described the bone attachments of the sMCL, dMCL, and POL, and transferred the anatomical observations to clinically relevant localisation in true-lateral radiographs. These findings should help surgeons to identify landmarks and attachment points to anatomically reinsert an avulsed ligament, to accurately position drill holes for MCL reconstruction, or to critically assess graft positions at review. In addition the oblique orientation of the dMCL implies its importance in restraint of tibial external rotation. This is very relevant to anteromedial rotatory instability in cases combined with ACL rupture.

The locations of the femoral and tibial attachments of the superficial and deep MCLs, and the POL, have been described in relation to overall knee sizes and also in relation to osseous landmarks, using both optical digitisation and radiography. The femoral sMCL attachment envelops the ME and has a long distal tibial attachment. The dMCL has an antero-distally oblique course as it fans out to a wide anteromedial tibial attachment. The POL attaches proximal and posterior to the femoral ME and inserts at the posteromedial tibial rim.