Introduction
The anterior cruciate ligament (ACL) consists of two functional bundles [
2,
4,
16]. The anteromedial bundle (AMB) originates anteroproximal in the intercondylar notch, close to the over-the-top position at the posterior edge of the notch, from the deep high part of the femoral attachment area and inserts anteromedial on the anterior intercondylar area of the tibia. The posterolateral bundle (PLB) originates more posteriorly and distally in the notch, from the shallow low part of the femoral attachment area and inserts posterolateral on the anterior intercondylar area of the tibia. The ACL reconstruction aims at restoring normal knee function. Most ACL replacements are performed with the isometric single-bundle technique. Isometric positioning of a single-bundle graft results in replacement of the AMB only. Although tensioned over the complete range of motion, the fibres are mostly tight in flexion. The AMB is the major constraint for anterior tibial displacement of the flexed knee [
30], but cannot restore normal knee laxity and kinematics near extension [
4,
28,
41]. In an effort to improve knee mechanics, double-bundle anatomic ACL reconstructions are now developed with reconstruction of both AMB and PLB [
9,
10,
13,
19,
28,
35,
40,
46,
49,
50,
52,
53]. As presented in previous studies, a reconstructed PLB is able to restore stability in knee angles where an isometrically placed graft fails [
28,
41,
50]. Additional restraint against anterior displacement in 15° of flexion [
48] as well as prevention of the pivot shift is demonstrated [
28,
50]. Therefore, a reconstruction with two bundles should be able to approximate normal ACL function over the complete range of motion [
39,
41,
51].
Tunnel positioning is an important factor for clinical success of ACL reconstructions. Incorrect tibial [
23] and femoral [
29,
55] tunnel placements result in abnormal knee mechanics. Anatomical placement restores normal knee function better than isometric placement [
36,
55]. However, accurate tunnel placement seems difficult. Misplacement between 25 and 65% of the tibial and femoral tunnels is reported [
8,
27,
47]. Double-bundle ACL reconstructions require an anatomical placement of the bone tunnels. It is difficult to identify the ACL remnants in chronic ACL-injured knees. Therefore, detailed information about the approximate native position is essential to determine proper anatomic tunnel placement for the two bundles during arthroscopy [
17]. The anatomical position has been the subject of many studies [
3,
5,
12,
14‐
16,
20,
26,
33,
34,
37,
38,
44]. Only a few recent studies have described the anatomic positions of the AM and PL bundles [
11,
32,
45,
52]. Due to the two-dimensional and limited view on the arthroscopic monitor, the landmarks and descriptions used in the above-mentioned studies seem not sufficient for correct positioning of the two separate bundles in all planes during arthroscopic surgery.
This study is aimed at acquiring quantitative geometric data of the ACL attachments on tibia and femur, such that these data can be used in an arthroscopically guided procedure for reconstruction of the ACL. We hypothesized that reliable guidelines to find the centres of the AMB, PLB and ACL relative to arthroscopically visible landmarks can be established. For the purpose of an anatomically accurate reconstruction of the ACL, the variations should be equal to or less than reported in other studies. As regard to the dimensions of drilled tunnel holes, normally 10 up to 12 mm, 95% (mean ± 2SD) of the attachment centres should be within this range. Therefore the a priori set assumption is that the maximally acceptable SD is 2.5 up to 3 mm.
Discussion
Incorrect tunnel placement, tibial [
22,
25] as well as femoral [
43], is seen as one of the most important causes of clinical failure in single-bundle ACL reconstructions [
1]. In double-bundle ACL reconstruction, exact anatomic tunnel placement seems to be even more essential. Two bundles must be accurately placed relative to the surrounding structures and relative to each other. Although exact tunnel positioning is important, it seems difficult, even for experienced surgeons [
8,
27,
47]. A clear description of the anatomic centres with guidelines to determine the correct tunnel position during arthroscopic procedures can improve the accuracy.
The femoral positions of the two distinct bundles found in this study, the AMB deep high and the PLB shallow low in the notch, are broadly in line with literature [
11,
20,
32,
45,
52]. Compared to others, the landmarks that are used in the present study are more easy to locate during arthroscopy. Harner et al. [
20] quantified the cross-sectional shape and area of the femoral and tibial attachments of both components in 10 knees without describing the positions relative to landmarks. Yasuda et al. [
52] limit their study to the femoral attachment, using five specimens. They describe the centre of the PLB 5 to 8 mm anterior to the edge of the joint cartilage, on the vertical line through the contact point between the femoral condyle and the tibial plateau in a 90° flexed knee. Because this point depends on the position of the knee, it can be difficult to locate accurately during an arthroscopic procedure. Colombet et al. [
11] examined seven specimens, collecting especially data of the attachment dimensions. The femoral results presented by the studies of Mochizuki et al. [
32] and Takahashi et al. [
45] are more suitable for practical use during arthroscopy; however, no exactly defined landmarks were used. The tibial results of Takahashi et al. [
45] cannot easily be transferred to the arthroscopic situation. Finally, above-mentioned studies did not present the centres of the ACL attachment.
Tunnel positioning in the femoral notch is often determined by the ‘clock’ method [
13,
18,
35,
36,
52,
54]. However, this only determines the high–low position from the roof, in the transversal plane along the arc of the femoral notch [
17,
54]. The accuracy in the sagittal plane, deep–shallow along the roof of the notch, seems to affect the functional outcome more [
17,
21,
29,
55,
56]. This position is often determined with a femoral guide placed behind the posterior edge of the intercondylar notch, at the over-the-top position, not to be confused with the Resident’s ridge [
24]. This seems sufficiently accurate for determining the AMB centre, when a guide with a 7 mm offset is placed in the 10.30 or 1.30 o’clock position. The 7 mm offset of the guide places the tunnel in the sagittal plane close to the AMB centre position, found in this study, i.e. 7.2 mm shallow along the roof. Although we did not translate the measured distance in a clock position, the 10.30 clock position seems close to the measured position in the transversal plane, 1.4 mm low from the roof on the condyle wall. This corresponds with the results of Mochizuki et al. [
32] and Yasuda et al. [
52]. However, the clock method is sensitive to subjective interpretation of positioning the face of the clock [
7]. It is not sufficient to find the correct position for the PLB, especially in the sagittal plane. Therefore Colombet et al. [
11] defined an extra guideline to position the PLB tunnel: 8 mm lower and ‘shallower’ relative to the AMB centre, found with the clock method. The present study is more precise: 1.3 mm more shallow along the notch and 5.3 mm lower from the roof on the condyle wall, relative to the above-described position of the AMB centre. These positions can best be approached through an anteromedial arthroscopic portal [
6].
Various methods can be used to determine the correct drill hole position at the tibia. Some authors prefer placement of the tibial tunnel based on avoiding graft impingement against the roof of the femoral notch with knee extension [
22,
34,
44]. Others prefer guides that use the posterior cruciate ligament (PCL) attachment as reference point [
31]. Based on these methods, positioning in anteroposterior direction, the sagittal plane, is defined. However, placement of the tunnel in mediolateral direction, the transversal plane, which is also important [
26], is not determined. Some studies describe the mediolateral position, relative to the width of the tibial plateau [
23,
45]. However, this guideline cannot be used during an arthroscopic procedure. The results of this study can be used for arthroscopic positioning in both directions.
The division of the AMB and PLB on the tibia in anteroposterior direction is similar to Harner et al. [
20], resulting in a medial AMB and lateral PLB. This deviates only marginally from other definitions, where the division is in an anteromedial and a posterolateral part [
3,
16,
38]. The position of the centres in anteroposterior direction correspond with the results of Takahashi et al. [
45], who used similar reference points to define the anterior margin of the articular surface of the tibial condyles. The centres of the two bundles were more close to each other on the tibia, than on the femur.
The size of the femoral ACL attachment 184 ± 52 mm
2 was similar to the result of Odensten and Gillquist (200 mm
2) [
38]. Other studies found smaller attachment areas, 132 mm
2 [
45] and 113 mm
2 [
20]. This difference may be caused by a different measuring method, using the largest projection in a 2D image plane. Our study measured the actual three-dimensional attachment area. The partition of the femoral attachment area was nearly equal with 45%, 55%, respectively, for the AMB and PLB. This was comparable to Harner et al. [
20] (AMB = 52%, PLB = 48%) and Takahashi et al. [
45] (AMB = 50%, PLB = 50%). The average tibial attachment area found in this study was larger than the femoral attachment. This is in line with the results of Girgis et al. [
16] and Fuss et al. [
15]. On the other hand, Takahashi et al. [
45] found that the femoral attachment was larger than the tibial attachment (132 vs 119 mm
2). Harner et al. [
20] and Odensten and Gillquist [
38] found similar sizes for the tibial and femoral attachment areas. In our study the AMB occupied 59% of the tibial attachment, nearly similar to the 57% of Takahashi et al. [
45], slightly more than the 52% Harner et al. [
20] found.
There was a large variation in knee sizes and dimensions of the reference lines. However, we did not find a correlation between the absolute distance of the attachment centres and the size of the reference lines, i.e. knee size. The absolute positions of the attachment centres were more or less similar for large and small knees. Therefore both absolute as well as relative data can be used as guidelines to find the anatomical positions. At the arthroscopic view, the absolute positions are more useful than the relative positions.
As regards to the dimensions of drilled tunnel holes (10–12 mm), the standard deviations of the attachment centres within 3 mm are acceptable. The femoral positions had less variations than the tibial positions. The positions for the entire ACL attachment had less variations than the attachment positions for the separate bundles. The results of this study seem to produce accurate guidelines to find the anatomic tunnel position during arthroscopic reconstruction. However, some limitations must be mentioned. The subdivision of the ACL in AMB and PLB is not based on anatomically distinct fibre bundles surrounded with fibrous issue [
5,
15,
33,
38]. Fibres of both bundles are twisted around each other [
33]. Division of the ACL in separate bundles is not easy [
3]. Therefore utmost care was taken to divide the two functional bundles according to a previously described method [
11,
20]. The method, based on observed tension variation during passive flexion–extension movement was reported to be consistent [
32]. This is expressed in the standard deviations (SD) of the mean attachment centres. The variation of the population is expressed by the SD of the ACL centre. The SD of the mean PLB and mean AMB centres show this variation combined with the division error. For the tibia the SDs of the PLB and AMB attachment centres are equal to that of the ACL. For the femur the SDs of the two bundle centres are slightly larger. However, these results do not point to large errors in the identification of the two bundles. The variations of attachment centre positions are actually rather small and similar to other studies [
11,
45].
The age of the cadaveric knee specimens was more than 60 years. However, only knees without severe arthritic signs were evaluated. It seems likely that the data can be transferred to the, mostly younger, population receiving an ACL reconstruction [
3,
11,
34,
44]. Perfusion fixation preserves the outer contours of shapes i.e. ligament bundles, so neither the identification nor the measurements of the various bundles can be assumed to have been hampered by the fixation.