Introduction

Conventional single-bundle (SB) anterior cruciate ligament (ACL) reconstruction using the transtibial method is commonly performed worldwide. With this technique [41], the tibial tunnel is placed near the postero-lateral (PL) bundle insertion site. The femoral tunnel is placed, in most cases, in a higher position in the notch, when compared to the anatomical antero-medial (AM) bundle footprint [6, 10, 50]. Therefore, transtibial SB ACL reconstruction can be considered as a non-anatomic technique. Recently, attention has been brought to anatomic double-bundle (DB) reconstruction and the importance of an anatomical tunnel placement in which the tunnels are positioned according to bony landmarks previously described [12] and are positioned within the native ACL footprint [37, 49]. In a recent study it has been shown that anatomic DB reconstruction has better healing and biomechanical characteristics than the non-anatomic DB reconstruction [10]. With this new attention to anatomic graft placement, SB ACL reconstruction must also be re-evaluated to establish the correct tunnel positions for this technique.

The purpose of this study was to compare different SB ACL reconstruction tunnel positions and to determine which would better restore intact knee kinematics and the results would be compared to anatomic DB ACL reconstruction. It was hypothesized that the anatomic SB and DB reconstructions would better restore normal knee kinematics when compared to the non-anatomic reconstruction.

Materials and methods

Ten fresh frozen unpaired adult pig knees (Hampshire, 7–8 months) were used for biomechanical testing. Specimens with a congenital abnormality or arthritis were excluded from the study and the existence of an intact ACL was confirmed arthroscopically. The limbs were disarticulation at the hip joint, frozen at −20°C, and thawed the night before testing. The knees were kept intact and the specimens were kept moist with physiologic saline solution. The tibia and femur were sectioned at the midshaft and the ends potted in heavyweight epoxy putty.

Robotic/UFS testing system

A CASPAR Stäubli RX90 robot (Orto MAQUET, Germany) with a universal force–moment sensor (UFS-Model 4015; JR3 Inc., Woodland, CA) was used for the knee testing. The robot repeatability of motion within ±0.02 mm at each joint and the load cell has an accuracy ±0.2 N and ±0.1 N m according to the manufacturer. The system controls both the displacements and the forces/moments applied to the knee in all six degrees of freedom (DOF) using a mathematical description of knee kinematics [15]. The tibia and femur were secured in aluminum cylinders using epoxy compound and placed in the testing system. The tibial cylinder was connected to the UFS (Fig. 1). The control and data acquisitions were performed using a personal computer and the MATLAB programming environment with a multitask operating system (Math Works Inc., Natick, MA, USA).

Fig. 1
figure 1

The specimen was secured within custom-made aluminum cylinders by using an epoxy compound and fixed to the testing system. The tibial cylinder was connected to the universal force/moment sensor

Full size image

The six DOF path of passive flexion–extension of the intact knee joint was first determined from full extension (30° of flexion) to 90° of knee flexion in 0.5° increments of flexion [35, 42]. The path of passive flexion–extension of the knee joint was determined by minimizing forces and moments in the other degrees of freedom.

Testing protocol

Specimens were tested before and after ACL resection as well as before and after ACL reconstruction. An anterior load of 89 N (simulated KT1000 test) [9, 48] was applied as described by Oster et al. [33] to the specimen at: (a) full extension (~30° between the anatomical axis of tibia and femur), (b) 30° from full extension (FFE) (~60°), and (c) 60° FFE (~90°) while allowing two degrees-of-freedom motion (medial–lateral/superior–inferior) of the tibia. In the second loading case, for a simulated pivot-shift test, a combined 7 N m valgus and 4 N m internal tibial torque was applied to the specimen at full extension and at 30° FFE (~60°) [22, 23, 30].

Five ACL reconstruction groups were tested: (1) anatomic SB reconstruction (MID–MID) (n = 5), the graft was placed from the middle between the tibial AM and PL footprints to the middle between the femoral AM and PL footprints; (2) anatomic SB reconstruction (AM–AM) (n = 5), the graft was placed from the tibial AM footprint to femoral AM footprint; (3) anatomic SB reconstruction (PL–PL) (n = 5), the graft was placed from the tibial PL footprint to femoral PL footprint; (4) conventional SB reconstruction (PL–high AM) (n = 5), the graft was placed from the tibial PL footprint to femoral high AM position; and (5) anatomic DB reconstruction (DB) (n = 5), grafts were placed in AM–AM and PL–PL positions. Of the 10 knees, five were used for MID–MID and the other five knees were used for the remaining reconstructions (AM–AM, PL–PL, PL–high AM, and DB) (Fig. 2).

Fig. 2
figure 2

Tunnel positions in the five different ACL reconstruction methods

Full size image

The testing protocol and data acquired are detailed in Fig. 3. The simulated KT1000 test was evaluated at the three flexion angles (30, 60, and 90°), and the simulated pivot-shift test was done at two angles (30 and 60°). Anterior tibial translation (ATT) (mm) was measured before and after ACL transection, as well as after ACL reconstruction.

Fig. 3
figure 3

Study design

Full size image

Removal of the ACL allowed measurement of the in situ forces applied to the ACL under the two load cases (simulated KT1000 and pivot-shift tests). After each ACL reconstruction, the two loads were applied to the reconstructed knee. Removal of the reconstructed ACL graft allowed measurement of the in situ force experienced by the grafts. Using the principle of superposition, the change in the force in the same joint positions, before and after ACL transection, represents the in situ force in the ACL [35]. The force in the ACL is the difference between force–moment data recorded in the intact knee and ACL-deficient knee is the force in the ACL under the external loadings, and the difference between force–moment data recorded in the reconstructed knee and the knee without the grafts is the force carried by the graft.

Surgical technique

A three-portal technique was used with anterolateral, anteromedial, and accessory medial portals with a 30° scope [14]. The three ACL bundles (AM, PL, and intermediate (IM) bundles) were identified by tension patterns during knee motion (Fig. 4). The IM and AM bundles were identified at the tibial insertion while the PL bundle was identified at the femoral side. The bundles where removed with a Vulcan EAS Electrothermal Arthroscopy System (Smith & Nephew, Endoscopy, Andover, Mass).

Fig. 4
figure 4

The three bundles of porcine ACL [antero-medial (AM), postero-lateral (PL), and intermediate (IM) bundles] were arthroscopically identified

Full size image

ACL reconstruction

ACL reconstruction was performed with a single-socket technique, as described by Rosenberg and Graf [34]. A guide wire (2.4 mm diameter) was inserted into the center of each tibial ACL bundle footprint (AM and PL) from the anteromedial aspect of the tibia using a tibial drill guide system (Smith & Nephew Endoscopy, Andover, Mass). The wire was then overdrilled with a cannulated reamer (5 mm diameter). Three femoral tunnels (PL, AM, and high AM) were created using a transportal technique [14]. A guide pin was inserted to the center of the AM footprint, PL footprint, and high AM position which is placed at the 11 or 1 o’clock position of the superoanterior portion of the ACL femoral footprint on the lateral wall of the intercondylar notch (Fig. 5a). The pin was then further overdrilled to the anterolateral femoral cortex with a cannulated reamer (5 mm diameter).

Fig. 5
figure 5

a, b 3D-CT images after ACL reconstructions

Full size image

For the MID–MID reconstruction, the tibial and femoral tunnels place was drilled between the center of AM and PL footprints of the tibia and femur (Fig. 5b).

Previously harvested semitendinosus and gracilis tendons from human cadaver knees were used as grafts. A No. 5 braised polyester suture was whip-stitched with a tapered needle into the free ends of the folded grafts. The graft was passed through the EndoButton loop (Smith & Nephew Endoscopy) and made to be two stranded. At the fixation of every SB graft, 80 N-load was applied as an initial tension [8] at 30° of knee flexion [13, 18, 45] using a ligament tension meter (Meira Corp., Nagoya, Aichi, Japan). At the fixation of DB, 40 N-load was applied for each bundle as an initial tension at 30° of knee flexion. Staples were used to fix the graft on the tibia.

Statistical analysis

Differences in ATT and in situ force at the different flexion angles were analyzed using Kruskal–Wallis test for comparison of all groups and Mann–Whitney U test between all the pairwise comparisons. A Bonferroni approach was used to adjust the alpha level for the pairwise post-hoc comparisons and it was assumed that there was statistical significance when P < 0.05 for Kruskal–Wallis test and P < 0.01 for Mann–Whitney U test. The statistical analysis was done using the software package SPSS version 17.0.

Results

A statistical difference was found for the simulated KT1000 test at 60° of knee flexion, at which the PL–PL graft had significantly lower in situ force than the intact ACL (P = 0.008). Although there was no statistical significant difference, the PL–high AM had a lower in situ force at 60° of knee flexion (P = 0.016). At 90° of knee flexion, every reconstructed graft except for the PL–PL graft had significantly lower in situ force when compared to the intact ACL (P < 0.01). The PL–PL graft also had lower in situ force at 90° of knee flexion, although not significant (P = 0.016). At 60° of knee flexion, and the MID–MID graft had significantly higher in situ force than the PL–PL graft (P = 0.009). The MID–MID was not significantly different from the DB reconstruction at 30, 60, and 90° of knee flexion. The PL–PL graft had a lowest in situ force at 60 and 90° of knee flexion, and the PL–high AM graft had a lowest in situ force at 30° of knee flexion (Fig. 6a).

Fig. 6
figure 6

a In situ forces during the simulated KT1000 test; and b in situ forces during the simulated pivot-shift test

Full size image

With the simulated pivot-shift test at 60° of knee flexion, the PL–PL and PL–high AM grafts had a significantly lower in situ force than the intact ACL (P = 0.004 and 0.004, respectively). In every graft, in situ force at 30° of knee flexion was higher than 60° of knee flexion. The PL–high AM graft had a lowest in situ force at both 30 and 60° of knee flexion (Fig. 6b).

During the simulated KT1000 test, the MID–MID and DB reconstructed knees were less translated anteriorly at 30 and 60° of knee flexion, when compared to the intact knee. Even the MID–MID and DB reconstructed knees were significantly translated (P = 0.002 and 0.009, respectively). During the simulated KT1000 test at 30, 60, and 90° of knee flexion, there was a significant difference in ATT between the intact knee and the PL–high AM reconstructed knee (P = 0.000, 0.001, and 0.002, respectively). At 30, 60, and 90° of knee flexion, there was a significant difference in ATT between the intact knee and the PL–PL reconstructed knee (P = 0.002, 0.003, and 0.002, respectively). When compared to the ACL-deficient knee, there was no significant difference in the PL–PL reconstructed knee at 60 and 90° of knee flexion and in the PL–high AM reconstructed knee at 30, 60, and 90° of knee flexion. With the simulated KT1000 test at 30, 60, and 90° of knee flexion, the MID–MID reconstructed knee, the AM–AM reconstructed knee, and the DB reconstructed knee had significantly lower ATT than the ACL deficient knee (P < 0.01) (Fig. 7a).

Fig. 7
figure 7

a ATT during the simulated KT1000 test; and b ATT during the simulated pivot-shift test

Full size image

With the simulated pivot-shift test at 30 and 60° of knee flexion, there was no significant difference among groups. At 30° of knee flexion, the ACL deficient knee was significantly translated during the simulated pivot-shift test (P = 0.010) (Fig. 7b).

Discussion

This study has shown that the anatomic MID–MID reconstruction better restores normal knee kinematics when compared to the non-anatomic ACL reconstructions. The purpose of this study was to compare different tunnel positions for SB ACL reconstruction to determine the position that best restores intact knee kinematics. It was shown that the MID–MID reconstruction was well-balanced for antero-posterior and rotational stability. This reconstruction provided the most stability among the anatomic SB ACL reconstructions and more closely restored normal knee kinematics. It was hypothesized that the AM–AM reconstruction would have a better antero-posterior stability and higher in situ force during the simulated KT1000 test, but contrary to this, the MID–MID reconstruction was superior. The reason for this may be that the MID–MID graft is positioned between the AM and PL insertion sites causing a synergic effect. Brophy et al. [3] found a trend toward increased PL bundle strain when compared to the AM bundle at 30° of knee flexion, suggesting that the AM and PL bundles have a synergistic relationship as the knee comes to extension. In the present study, the AM–AM graft had a relatively high in situ force during ATT at every angle of knee flexion. The PL–PL graft had the lowest in situ force during ATT at 90° of knee flexion. These are in accordance with prior studies that have reported the AM bundle to be the dominant bundle in controlling translation when the knee is flexed [3, 16, 48].

ACL reconstruction has been widely performed for many years and innovations of surgical instruments and improvement of surgical techniques [9, 27] have occurred, like the single-incision endoscopic transtibial approach [29, 41]. Studies based on graft isometry have supported placement of the femoral tunnel high in the intercondylar notch in a non-anatomical location [6, 50]. Often the tibial tunnel must be placed slightly posteriorly [5, 19, 31, 38] in order to reach the back-wall for placement of the femoral tunnel [3] and to avoid roof impingement [19]. These recommendations place the tibial tunnel within the PL bundle of the ACL footprint [49]. Therefore, the conventional SB transtibial ACL reconstruction is actually a non-anatomical technique. Even though it is a non-anatomical tunnel placement, the results of SB ACL reconstruction using either a patellar tendon or hamstring autograft has consistently yielded high stability rates, subjectively high patient satisfaction, and low revision rates in the hands of experienced surgeons [25].

Several recent in vivo kinematic studies [4, 17, 40] and biomechanical studies [26, 43], however, have shown that conventional SB ACL reconstruction cannot restore tibial rotation to normal levels and does not have any significant effects against combined rotational loads. To address these concerns, there has been a clinical trend toward placing the femoral tunnel lower in the notch into the anatomical footprint of the ACL [20, 24, 39, 44] in order to improve rotational stability [26, 32, 36]. These findings supported our results that anatomic SB ACL reconstruction (except PL–PL) had better stability than the PL–high AM. The PL–PL reconstruction was superior to the PL–high AM reconstruction only in terms of rotational stability.

Recently, attention has been brought to anatomic DB ACL reconstruction and the importance of an anatomical tunnel placement [14, 46]. Advocates [25] suggest the DB technique improves rotational control [40], improves overall function, and possibly decreases radiographic evidence of postoperative degenerative joint disease, although there are reports that the clinical outcomes of DB reconstruction are not always significantly different than that of SB reconstruction [1, 2]. In this study, anatomic MID–MID SB reconstruction had similar results to the DB reconstruction (P = 1.000). Although several studies have shown that DB ACL reconstruction is superior to SB ACL reconstruction [7, 10, 11, 21], there are indications for SB ACL reconstruction and these include, among others, small knees and narrow notch [28]. The results from this study suggest that a MID–MID SB reconstruction should be performed when a SB reconstruction is warranted.

There are limitations to this study. The use of porcine knees is the main limitation. In spite of this being a commonly used animal model for ACL reconstruction [7, 10, 11, 21], the results found in this study may not be reproducible in the human knee. The porcine ACL has three bundles (AM, IM, and PL bundles); although we carefully marked the insertion site of each bundle before they were sectioned, the IM bundle was not included in this study as its role has not been completely elucidated. Notwithstanding it is believed that this is a good animal model as this study and other authors have found similar roles of the AM and PL bundles in human ACL [10, 11, 21, 47]. Further studies on human knees will be needed in the future.

Conclusion

A robotic/universal force–moment sensor testing system was used to evaluate the graft in situ force with different tunnel positions in ACL reconstructions. The MID–MID SB ACL reconstruction reproduced the in situ force of the normal ACL better than any other SB ACL reconstruction. These findings suggest that anatomic MID–MID SB ACL is recommended for anatomic SB ACL reconstruction.