Volume 26, Issue 2 , Pages 202-213, February 2010
Anatomic Double-Bundle Anterior Cruciate Ligament Reconstruction: Kinematics and Knee Flexion Angle–Graft Tension Relation
Article Outline
Purpose
The purpose of this study was to compare the bundle tension curves and resultant knee kinematics between 2 tensioning protocols in anatomic double-bundle anterior cruciate ligament (ACL) reconstruction.
Methods
Anatomic double-bundle ACL reconstruction was performed in 7 male cadaveric knees. Each graft was tensioned to 22 N under 2 conditions: (1) both bundles tensioned at 20° of knee flexion (20/20 protocol) or (2) posterolateral (PL) bundle tensioned at 15° and anteromedial (AM) bundle at 45° (45/15 protocol). Knee kinematics were recorded in response to anterior and combined rotatory loads in the intact, ACL-deficient, and reconstructed states. Bundle tension was recorded dynamically with knee motion and during each loading test.
Results
Tensioning both bundles at 20° of knee flexion resulted in a reciprocal bundle tension pattern that was not statistically different; the PL bundle tension was greater than the AM bundle tension in full extension, and the AM bundle tension was greater than the PL bundle tension from 25° to 120°. In the second tensioning protocol, the AM bundle tension was significantly greater than the PL bundle tension at all flexion angles. Both tensioning protocols restored normal knee kinematics.
Conclusions
Bundle-tensioning protocol is a variable that has a significant effect on the bundle-loading patterns in double-bundle ACL reconstruction. The 20/20 protocol resulted in AM and PL bundle–loading patterns that were equivalent during dynamic testing, whereas the 45/15 protocol led to excessive tension in the AM bundle in full extension. We recommend equal tensioning of both bundles with the knee at 20° of flexion to restore relatively normal tension curves in each bundle and to avoid excessive stress on the AM bundle.
Clinical Relevance
In double-bundle ACL reconstruction, there is no consensus regarding bundle-tensioning protocols. This study provides data on the individual bundle tension curves that result from 2 commonly used tensioning protocols. These data will assist clinicians as the technique and application of double-bundle ACL reconstruction move forward.
Anatomic and biomechanical studies have shown that the anterior cruciate ligament (ACL) consists of 2 distinct bundles that act in a reciprocal fashion during knee flexion. The posterolateral (PL) bundle resists anterior translation and internal rotation of the tibia on the femur in relative knee extension, and the anteromedial (AM) bundle offers more restraint to anterior tibial translation in relative knee flexion.1, 2, 3, 4, 5
Currently, with single-bundle reconstruction techniques, excessive knee laxity and instability may occur.6, 7, 8 An anatomic reconstruction that restores normal kinematics in the knee may lead to better long-term results, with decreased rates of instability and the development of degenerative arthritis. Some surgeons believe that re-creation of both bundles of the native ACL is preferable to restore more normal knee biomechanics, especially in response to rotatory loading,5, 9, 10, 11, 12 although there is no clinical evidence to support this as of yet. This has led to the adoption of anatomic double-bundle anterior cruciate ligament reconstruction (DB-ACLR) in some centers. However, there remains a lack of consensus in the literature regarding technique.
Most authors agree that the tunnels should be placed in the anatomic footprints of the 2 bundles. However, there is no agreement regarding bundle-tensioning protocols. Some authors have suggested that the AM bundle should be tensioned with the knee between 45° and 90° of flexion and the PL bundle near full extension.10, 12, 13, 14, 15, 16 The rationale behind this tensioning protocol is to tension the bundles at the degree of knee flexion at which they function. Other authors have suggested that the bundles be tensioned and fixed at the same degree of flexion at approximately 15° to 20°.4, 11, 17, 18, 19, 20, 21 This tensioning protocol is designed to avoid the generation of excessive stress in either bundle, which could lead to stress shielding of the other. There is little evidence to support the use of any one tensioning protocol over another.
The purpose was to investigate the effect of 2 different graft-tensioning protocols on tension developed in the AM and PL bundles during knee range of motion in a cadaveric model of DB-ACLR, as well as to measure the kinematic effect of these protocols on the reconstructed knee when anterior and rotatory loads are applied. Our hypothesis was that tensioning both bundles at 20° of knee flexion would create reciprocal tension curves and load sharing between the bundles and that tensioning the PL bundle at 15° and the AM bundle at 45° would lead to excessive tension in the AM bundle and fail to restore reciprocal bundle tension curves.
Methods
Specimen Preparation
Ten fresh-frozen cadaveric knees without any evidence of prior knee surgery were harvested from male donors (mean age, 49.7 years; range, 28 to 64 years) with the approval of the Texas State Anatomical Board. The tibia and femur were sectioned 30 cm from the joint, and all overlying soft tissues were left undisturbed for a length of 17.5 cm proximal and distal to the joint line. We potted the femur and tibia in polyvinyl chloride (PVC) pipe (length, 12.5 cm; diameter, 37 mm) in casting resin (Smooth-Cast 300; Smooth-On, Easton, PA), ensuring that the sides of the PVC pipe were parallel to the respective bone axis. The fibula was cut just proximal to the proximal end of the PVC pipe on the tibia and was fixed to the tibia in situ by use of a single transverse bone screw. Heavy suture loops were placed through the rectus femoris, vastus medialis, vastus lateralis, and hamstring group of muscles. A single one quarter–inch 20-anchor insert was placed into the tibial tubercle through a small incision. Its fixation was augmented with bone cement. An eyelet was inserted into this anchor, which was used to apply anterior and posterior loads.
After initial kinematic testing, each intact specimen underwent diagnostic arthroscopy and was evaluated for any evidence of chondral, meniscal, or ligamentous pathology. The ACL was examined to determine the demarcation between the AM and PL bundles. Bundle sectioning was performed arthroscopically at the midportion of each bundle.
Kinematic Testing
Each knee was subjected to kinematic testing with the ACL intact and after sectioning of the entire ACL. A mechanical loading frame (“knee simulator”) was designed and constructed to allow physiologic loading and constraint of cadaveric specimens to allow them to simulate lunging (Fig 1). The knee simulator was configured to provide 6 df at the knee to replicate the frontal squat maneuver. A fully extended cadaveric knee was attached to a mechanical hip and ankle joint on the knee simulator and underwent a quasistatic lunging maneuver until 90° of knee flexion was achieved. The ankle joint consisted of 3 independent rotational bearings that permitted ankle flexion, varus/valgus tibial rotation, and internal/external tibial rotation. These bearings were mounted to a translational carriage that allowed the tibia to translate anteriorly and posteriorly. The hip joint was mounted on vertical rails that allowed the hip to translate medially/laterally and slide superiorly/inferiorly toward the ankle. Two high-resolution capacitive inclinometers were mounted rigidly to the mediolateral axis of rotation at the hip and ankle joint to measure the tilt of the femur and tibia with respect to the vertical. An electromechanical actuator was used to simulate body weight and to drive the hip joint superiorly/inferiorly. Tibial torque was applied by use of a computer-controlled servomotor with an inline torque and rotational transducer. Signals from all the transducers were sent to and analyzed with a computerized data acquisition system.
Figure 1.
Knee simulator with mounted specimen. A constant force was applied to the quadriceps and hamstring muscle groups during testing. After the anatomic DB-ACLR, the tibial side of each graft was held with custom-made tensioning devices that allowed precise and separate tensioning of each bundle. For kinematic testing, anterior and posterior loads were applied to the tibia by use of an eyelet inserted into an anchor in the tibial tubercle. Application of a posterior load is shown.
In knees that passively reached 0° or hyperextension, full extension was defined as 0°. In knees with mild flexion contractures, an additional posterior load of 50 N was applied to the knees. Full extension was defined as the angle at which the knee rested with that load applied. Full passive range of motion (PROM) was defined as full extension to 140° of flexion.
Specimens were mounted in the knee simulator and cycled through PROM for 10 cycles. A 134-N anterior load was applied to each specimen at full extension and 15°, 30°, 60°, and 90°. Before the application of the anterior load, a posterior load of 50 N was applied. Anterior translation was measured as the difference between the tibial position with application of the posterior and anterior loads. As a measure of rotational stability, anterior tibial translation and tibial rotation were also measured in response to a combined rotatory load of 10-Nm valgus (applied by a weight on a side arm on the knee simulator) and 5-Nm internal tibial torque22 (applied robotically by the knee simulator) at 15°, 30°, and 60° of knee flexion, after application of a posterior load of 50 N to the tibial tubercle. During all testing, a constant force was applied to the vastus lateralis (4.5 N), vastus medialis (4.5 N), rectus femoris (13.3 N), and hamstring muscles (22.2 N) to simulate resting tension. The magnitude of anterior and rotatory forces was chosen based on the values used in multiple other kinematic studies.3, 5, 13, 16, 22
Anterior tibial translation was measured by use of the high-resolution capacitive inclinometers located at the ankle joint. The amount of tibial translation was calculated from the deflection angle of the inclinometer and the length of the tibial segment from the inclinometer to the tibial plateau. Tibial rotation was measured directly by use of the rotational transducer located in line with the tibial axis in the mounting fixture.
In our analysis of the rotational data, a total arc of rotation that took into account the complex nature of tibial rotation was calculated. The translational and rotational displacements were entered into a formula that calculated an arc of rotation. This method was validated by use of a motional analysis system and computer models to see exactly how the bones were moving. Thin-slice computed tomography (CT) scans with 3-dimensional (3D) reconstructions were obtained for each specimen with a helical scanner. The 3D positions of the femur and tibia at each flexion angle and each loading condition were tracked with optical navigation by use of a 2-camera motion analysis system (NDI Polaris; Northern Digital, Waterloo, Ontario, Canada) in conjunction with a 6-df high-definition laser scanner (Immersion, San Jose, CA).
By use of the CT reconstructions, specimen-specific coordinate systems similar to those defined by Patil et al.23 were defined for each bone in a 3D scanning software package (Insus Technology, Seoul, South Korea) to resolve femoral rollback and tibial rotation with respect to the tibia from the kinematic data. The transcondylar axis of the femur was defined between the centers of 2 spheres fit to the posterior condylar surfaces. The axis of a best-fit cone through the distal third of the femoral extramedullary canal defined the longitudinal femoral axis in the sagittal plane. The cross product of these 2 axes defined the anterior/posterior axis. The origin of the femoral coordinate system was defined as the midpoint of the transcondylar axis of the femur. The transverse plane defined by the femoral coordinate system closely approximated the plane of the tibial plateau. Therefore the tibial coordinate system was defined by displacing the femoral coordinate system along the longitudinal axis of the femur in full extension until its origin contacted the tibial plateau in full extension.
The geometry of the posterior femoral condyles was identified with a technique similar to that used in several prior studies24, 25, 26, 27 and was used to measure femoral rollback and rotation. The centers of the spheres fit to the posterior condyles of the femur and the transcondylar axis joining these points were projected to the transverse plane of the tibial coordinate system. Rollback was defined as the displacements of the sphere centers along the anterior/posterior axis of the tibial coordinate system on both the medial and lateral tibial plateaus.
Rotational displacement was defined as the angles subtended between the transcondylar axis vectors in the transverse plane of the tibial coordinate system (Fig 2).
Figure 2.
Calculating rotational displacement. The yellow outline represents the position of the tibial plateau in the unloaded state as viewed from above. The red and blue outlines represent the position of the tibial plateau in the intact and ACL-deficient knee, respectively, in response to the combined rotatory load. The red and blue arrows represent the measured arc of overall rotational displacement in the intact and ACL-deficient specimens, respectively. The red and blue points represent the points about which the tibial plateau rotated relative to the femur in the intact and ACL-deficient knee, respectively, in response to the combined rotatory load.
Anatomic DB-ACLR
After initial kinematic testing, arthroscopically assisted DB-ACLR was performed through anterolateral, anteromedial, and far medial portals (Fig 3). No notchplasty was done. A radiofrequency device was used to dissect and mark the anatomic locations of the AM and PL bundles on the femur and tibia under direct visualization. The center of each femoral footprint thereby determined was marked with an awl. A tibial angle guide was used to drill 2 guidewires into the AM and PL footprints on the tibia through separate incisions on the proximal medial tibia. The guide was set at 55° for the PL bundle and 45° for the AM bundle. Through the far medial portal and with visualization from the anteromedial portal, 2 guide pins were drilled into the AM and PL footprints on the femur. Each pin was drilled out through the lateral femoral cortex and then retracted until its end was flush with the intra-articular surface of the femoral notch. A lateral fluoroscopic image was obtained with all 4 guidewires in place to allow measurement of the tunnel positions as described later (Fig 4A).
Figure 3.
Anatomic DB-ACLR. (A) Placement of tibial guidewires as viewed from anterolateral portal. (B) Position of femoral tunnels with knee at 90° of flexion as viewed from AM portal. (C) Completed DB-ACLR with knee at 90° of flexion as viewed from anterolateral portal.
Figure 4.
Technique for quantification of tunnel positions by use of lateral radiographs. (A) Lateral radiograph with all 4 guidewires in place showing positions of planned tunnels for each bundle. (B) Lateral radiograph with guide bolts in femoral tunnels after reaming. (C) The position of each femoral tunnel was defined in terms of the percentage distance in each axis with the posterior end of the Blumensaat line defined as 0%, with “a” being the distance on the x-axis (along the Blumensaat line) and “b” being the distance on the y-axis of the grid. (D) The position of each tibial tunnel was expressed as a percentage of the maximum tibial sagittal diameter with the anterior-most point defined as 0%.
The tunnels were then drilled over the guidewires. For the femoral tunnels, the guidewires were first overdrilled with a 4.5-mm reamer, followed by 7-mm and 6-mm reamers for the AM and PL bundles, respectively. The positions of the femoral tunnels were measured again on a lateral fluoroscopic image by use of guide bolts with rounded ends placed in the tunnels (Fig 4B). The tibial guidewires were over-reamed with the 7-mm and 6-mm reamers only. Double-looped human semitendinosus hamstring tendon allografts (Musculoskeletal Transplant Foundation, Edison, NJ) were used for grafts. Femoral-side fixation was achieved by use of 6-mm and 5-mm EndoButton Direct devices (Smith & Nephew Endoscopy, Andover, MA) for the AM and PL bundles, respectively.
Quantification of Tunnel Position
The lateral fluoroscopic image was analyzed to quantify tunnel positions on the femur by use of the quadrant technique described by Zantop et al.,28 which was first described by Bernard et al.29 On a direct lateral projection, a grid was created by use of the intercondylar roof as one axis and a perpendicular line from the Blumensaat line to the distal-most subchondral projection of the lateral femoral condyle as the other axis. The position of each tunnel is defined in terms of the percentage distance in each axis with the posterior end of the Blumensaat line defined as 0%, with “a” being the distance on the x-axis (along the Blumensaat line) and “b” being the distance on the y-axis of the grid (Fig 4C).
The lateral fluoroscopic image was also used to define the position of each tibial tunnel in the sagittal plane by use of the technique described by Zantop et al.,28 which was first used by Stäubli and Rauschning.30 This is expressed as a percentage of the maximum tibial sagittal diameter, with the anterior-most point defined as 0% (Fig 4D).
The positions of the tunnels were compared with reference values available in the literature.28, 31
Determination of Knee Flexion Angle–Graft Tension Relation
After reconstruction, the specimen was remounted in the knee simulator so that the femur was resting parallel to the floor, with the tibia freely suspended. Range of motion of the knee was performed manually through a series of ropes and pulleys to avoid any significant rotational or coronal-plane forces. Knee flexion angle was determined by use of electronic inclinometers attached to the femur and tibia.
The AM and PL bundles were held on the tibial side with custom-made tensioning devices that allowed precise and separate tensioning of each bundle. Each tensioning device included a load cell (LSB200; Futek, Irvine, CA) that measured the tension in each bundle (Fig 5). Each bundle was pre-tensioned to 50 N with the knee at 20° of flexion, and the knee was brought through PROM for 10 cycles. The tension in each bundle was then set at 22 N, for a total tension of 44 N, with the knee placed in the following positions: (1) PL bundle tension set at 15° of knee flexion and AM bundle tension set at 45° of knee flexion (15/45 protocol) and (2) tension in both bundles set at 20° of knee flexion (20/20 protocol). The PL bundle was tensioned first in every case. After tensioning, each specimen was cycled 10 times through PROM. The grafts were then re-tensioned. Tension in each graft was measured as the knee was brought through PROM for 20 seconds at a rate of approximately 0.15 Hz. During this testing, muscle groups about the knee were attached to the testing apparatus to simulate resting tension only. The tension curve of each graft was dynamically recorded as a function of knee flexion angle at a rate of 100 Hz. Both tensioning protocols were tested in each specimen. All tension was released in each bundle before each tensioning pattern tested.
Figure 5.
Dynamic testing setup. Custom-made tensioning devices that allowed precise and separate tensioning of each bundle were used to hold the tibial side of each graft. Each tensioning device included a load cell (LSB200; Futek) that dynamically measured the tension in each bundle as the knee was brought through range of motion.
Kinematics After Anatomic DB-ACLR
After dynamic testing, the kinematics of the knee were again measured as previously described. After initial tensioning, cycling was performed for 10 cycles from 20° to 90° for the 20/20 protocol and from 15° to 90° for the 15/45 protocol. The grafts were then re-tensioned before kinematic testing. In addition to measurement of tibial translation and rotation, the tension in each bundle was recorded in response to each anterior and combined rotatory load.
Data Analysis
Mean values and SDs were calculated for each data point. Analysis of the kinematic data was performed by use of multivariate analysis of variance with a Fisher Least Significant Difference post hoc test to determine significant differences between groups. The load-sharing data were analyzed by use of the unpaired t test. The level of significance for the kinematic and load-sharing data was set at P < .05. The dynamic data were analyzed by use of repeated-measures analysis of variance with a Bonferroni post hoc correction to determine the level of significance. The maximum tension in each bundle during dynamic testing was compared by use of the unpaired t test, with the level of significance set at P < .05.
Results
Specimens
Of the 10 specimens, 2 were excluded. One was excluded before any testing because of a significant flexion contracture of 20°, and the other was excluded after initial arthroscopy because of a partial ACL deficiency (the PL bundle was torn). All data related to the reconstruction were excluded for an additional specimen after all testing had been completed because of significantly outlying data related to the PL tunnel position on the femur and the PL bundle tension curve; the data from the initial kinematic runs were retained to compare the intact, single bundle–deficient, and ACL-deficient states. The remaining 7 specimens ranged in age from 28 to 64 years, with a mean of 53 years. All of these showed easily identifiable double-bundle ACL anatomy. None had any evidence of prior surgery or pathologic laxity. There were grade 2 or 3 chondral changes in 1 or more compartments in 3 specimens. The remaining 4 specimens had minimal or no evidence of degenerative changes. There was no significant meniscal pathology.
Kinematic Data
Data for anterior tibial translation in response to the 134-N anterior load are shown in Table 1. In all specimens anterior translation peaked at 15° or 30° of knee flexion. The increases in anterior tibial translation were statistically significant at each flexion angle between intact and ACL-deficient specimens. In the reconstructed knees anterior translation was similar to that seen in the intact knees at all flexion angles for both tensioning protocols.
Table 1. Anterior Tibial Translation in Response to 134-N Anterior Load
| Anterior Tibial Translation (mm) | ||||
|---|---|---|---|---|
| Intact | ACL Deficient | 20/20 | 45/15 | |
| Knee flexion angle | ||||
 FE | 5.7(1.4)† | 12.2(2.2)⁎ | 6.5(1.1)† | 6.0(1.1)† |
| 15° | 8.3(2.2)† | 15.7(1.8)⁎ | 8.3(0.8)† | 8.1(0.9)† |
| 30° | 8.8(3.1)† | 16.1(2.8)⁎ | 8.0(1.0)† | 7.7(1.3)† |
| 60° | 4.7(2.0)† | 9.2(2.4)⁎ | 5.0(1.5)† | 4.3(1.0)† |
| 90° | 3.7(1.7)† | 7.4(2.0)⁎ | 4.2(1.3)† | 3.8(1.4)† |
⁎P < .001 compared with intact. |
†P < .001 compared with ACL deficient. |
Data for overall rotational displacement of the tibia in response to rotatory loading are shown in Table 2. When we analyzed the specimen used to formulate the calculation to determine rotational displacement, it was apparent that the center of rotation of the tibia was variable depending on the state of the ACL in that specimen. In the intact state the center of rotation was close to the center of the medial tibial plateau. In the ACL-deficient state the center of rotation moved medially to a point at the edge of the plateau. By use of this calculation, the increase in rotational displacement between the intact and ACL-deficient specimens was statistically significant at 15°. Both tensioning protocols produced rotational displacement that was, on average, less than that seen in the intact knees, although these differences were not statistically significant. The rotational displacement measured with both tensioning protocols was statistically significantly less than that in the ACL-deficient specimens.
Table 2. Rotational Displacement in Response to Combined Rotatory Load of 10-Nm Valgus and 5-Nm Internal Tibial Torque
| Overall Rotational Displacement(°) | ||||
|---|---|---|---|---|
| Intact | ACL Deficient | 20/20 | 45/15 | |
| Knee flexion angle | ||||
| 15° | 16.1(1.2)† | 19.4(3.6)⁎ | 14.6(1.6)† | 14.8(2.9)† |
| 30° | 17.9(2.9) | 21.2(6.2) | 15.3(1.9)† | 15.9(1.7)† |
| 60° | 14.7(3.4) | 17.2(7.8) | 11.9(1.9)† | 12.0(2.0)† |
⁎P < .05 compared with intact. |
†P < .05 compared with ACL deficient. |
Tunnel Positions
Tunnel positions are shown in Table 3. The average tunnel positions were similar to those previously reported.28, 31 There was some significant variation in tunnel position that influenced the results of the tension curves. The 1 specimen that was discarded because of significantly outlying data was found to have PL bundle tunnels that were more than 1 SD away from the mean values for each of the parameters used to describe the tunnel positions. The femoral PL bundle tunnel was relatively anterior and distal, and the tibial PL bundle tunnel was relatively posterior compared with the mean values. This resulted in a tensioning pattern in which the PL bundle tension was high in extension and low at 20° to 40° but was at its highest at 140°.
Table 3. Tunnel Positions
| Femoral Tunnels | Tibial Tunnels | |||||
|---|---|---|---|---|---|---|
| AM | PL | |||||
| a | b | a | b | AM | PL | |
| Specimen | ||||||
| 48793L | 16.4 | 21.7 | 34.5 | 40.0 | 36.1 | 50.6 |
| 51322L | 15.7 | 32.8 | 27.8 | 59.4 | 30.7 | 47.7 |
| 51241R | 18.5 | 30.0 | 32.1 | 57.5 | 27.7 | 44.9 |
| 50507R | 19.2 | 30.0 | 32.1 | 55.0 | 35.6 | 47.8 |
| 52237R | 23.0 | 32.6 | 32.2 | 52.2 | 31.3 | 48.5 |
| 51275R | 24.7 | 23.3 | 34.8 | 46.5 | 30.3 | 46.5 |
| 52242L | 21.2 | 20.0 | 31.8 | 52.5 | 27.3 | 44.4 |
| Mean | 19.8 | 27.2 | 32.2 | 51.9 | 31.3 | 47.2 |
| SD | 3.3 | 5.4 | 2.3 | 6.7 | 3.5 | 2.1 |
Knee Flexion Angle–Bundle Tension Curves
The knee flexion angle–bundle tension curves showed a consistent and reproducible hysteresis effect, with the graft tension being slightly different between flexion and extension of the knee. There was no significant decrease in bundle tension between cycles. For the purpose of analysis, a single complete flexion-extension cycle was used and the bundle tension values averaged at each flexion angle to create a single line curve for each bundle for each tensioning protocol. This curve was averaged for all specimens. The resultant flexion angle–bundle tension curves are shown in Fig 6. To average the curves for all specimens, this figure only shows data starting at 6° of knee flexion, because this was the largest flexion contracture seen in any of the specimens. For the 20/20 protocol, the bundle tensions intersect at 23°. The PL bundle had more tension in full extension than the AM bundle, and the AM bundle had more tension beyond 25° of flexion, although these differences were not statistically significant. In contrast, for the 45/15 protocol, the AM bundle tension was higher than the PL bundle tension for the entire flexion-extension arc. This difference was statistically significant throughout the flexion arc. The AM bundle tension peaked at 67.5 N (SD, 8.5 N) in the 20/20 protocol and 90.0 N (SD, 21.3 N) in the 45/15 protocol (P = .024). The PL bundle tension peaked at 74.3 N (SD, 6.0 N) in the 20/20 protocol and 64.4 N (SD, 7.6 N) in the 45/15 protocol (P = .018). The difference between the peak AM and PL bundle tension in the 45/15 protocol was statistically significant (P = .011). The difference between the peak AM and PL bundle tension in the 20/20 protocol was not statistically significant (P = .109).
Figure 6.
Knee flexion angle–bundle tension relation during dynamic testing. For the 20/20 protocol, the bundle tensions intersect at 23°. The AM and PL bundle tensions were statistically equivalent throughout knee range of motion. In contrast, for the 45/15 protocol, the AM bundle tension was higher than the PL bundle tension for the entire flexion-extension arc, and the difference was statistically significant throughout knee range of motion.
Load Sharing
Load sharing between the bundles for each tensioning protocol is shown in Table 4. These data are listed as percentages of the total load generated in both bundles during each of the kinematic tests as described by Mae et al.4 With 134 N of anterior loading, the AM bundle carried more of the load than the PL bundle at all flexion angles by use of the 45/15 protocol and at all angles except full extension by use of the 20/20 protocol. These differences were statistically significant. In response to the combined rotatory loading, the AM bundle carried more load than the PL bundle at all flexion angles by use of both tensioning protocols. The proportion of load shared by the AM bundle in the 20/20 protocol was significantly less than that in the 45/15 protocol at all flexion angles.
Table 4. Load Sharing
| 20/20 | 45/15 | |||
|---|---|---|---|---|
| AM% | PL% | AM% | PL% | |
| 134-N anterior load | ||||
| Knee flexion angle | ||||
| FE | 51.1(3.7)⁎ | 48.9(3.7) | 61.2(4.6) | 38.8(4.6) |
| 15° | 57.0(3.7) | 43.0(3.7) | 67.5(4.5) | 32.5(4.5) |
| 30° | 64.1(5.7) | 35.9(5.7) | 74.9(5.0) | 25.1(5.0) |
| 60° | 75.0(10.4) | 25.0(10.4) | 84.6(8.0) | 15.4(8.0) |
| 90° | 70.2(18.2) | 29.8(18.2) | 80.6(15.4) | 19.4(15.4) |
| Combined rotary load(10-Nm valgus + 5-Nm internal rotation) | ||||
| Knee flexion angle | ||||
| 15° | 55.3(3.5) | 44.7(3.5) | 66.0(4.7) | 34.0(4.7) |
| 30° | 63.7(4.8) | 36.3(4.8) | 74.9(5.6) | 25.1(5.6) |
| 60° | 78.9(7.8) | 21.1(7.8) | 86.6(4.3) | 13.4(4.3) |
⁎P = .31 compared with 20/20 PL%. P < .01 for all other values comparing AM% with PL%, AM% with AM%, and PL% with PL%. |
Discussion
The bundle tension curves in this study closely resemble the tension curves directly measured in the entire intact ACL by Markolf et al.32 despite the fact that each bundle was analyzed separately. However, the relation of bundle tension to knee flexion angle was significantly different between the 2 tensioning protocols. The 20/20 protocol allowed for statistically equivalent bundle tensions through PROM. With the 45/15 protocol, the AM bundle tension was significantly higher than the PL bundle tension throughout range of motion. When the knee was loaded for kinematic testing, the AM bundle absorbed a greater proportion of the load than the PL bundle except at full extension with anterior loading in the 20/20 protocol. However, the proportion of increased load absorbed by the AM bundle was less with the 20/20 tensioning protocol compared with the 45/15 protocol. Together, these results indicate that the forces to which the AM bundle was exposed after DB-ACLR were significantly greater by use of the 45/15 tensioning protocol compared with the 20/20 protocol. In contrast, although the forces to which the PL bundle was exposed were increased by use of the 20/20 protocol compared with the 45/15 protocol, this difference was not significant at full extension, where the bundle tension was greatest.
Normal knee kinematics were restored with both tensioning protocols even though the loading of the individual bundles differed significantly between them. The differential loading patterns of the bundles could have important implications for the short-term and long-term success of the reconstruction, leading to stress shielding of one bundle or attenuation and failure of the other and, ultimately, loss of stability and clinical failure of the reconstruction. Different bundle tension patterns could also affect the distribution of forces to the articular cartilage or menisci, which may have implications in the development of degenerative arthritis. The differences observed in the bundle tension patterns would not have been detected if the bundle tensions had not been directly measured. In light of these findings, studies investigating the importance of bundle tensioning must not rely purely on kinematic analysis to determine the potential success of DB-ACLR.
An unanticipated additional finding of this study was that the tension patterns of the bundles were extremely sensitive to variation in tunnel position. As described, one of the specimens yielded tension curves that did not at all resemble the others in the study. When this was investigated, it became clear that the PL bundle tunnels were substantially off from the mean in every parameter described, though only by a small amount in terms of actual distance. One other specimen showed somewhat irregular tension curves, with the PL bundle tension equal to the AM bundle tension at full extension in the 20/20 protocol but greater than the AM bundle tension for the rest of the flexion arc. However, the tension values did not differ significantly from the rest of the specimens and so were included in the analysis. The femoral tunnel of this specimen was relatively closer to the Blumensaat line compared with the mean values, and the tibial tunnel was relatively more posterior. These were the first 2 reconstructions performed in the study; as such, they likely represent the learning curve associated with this technically challenging procedure.
In other studies that report tibial rotation in response to loading, the point of rotation is considered to be the center of the tibial plateau, which is an oversimplification of what actually occurs. As seen in the computer-reconstructed images in 1 specimen in this study, the point about which the tibia rotates changes depending on the status of the ACL. In the ACL-deficient knee, this point moves medially (Fig 2). An anatomic ACL reconstruction should restore this normal point of rotation. Because we performed the computer reconstruction for only 1 specimen to correlate our other measurements with our calculated total arc of rotation, the data obtained in this study do not allow for precise calculation of the location of the point of rotation in all specimens. However, by calculating the total arc of motion, we have avoided the arbitrary definition of that point to the center of the plateau. Additional studies will be required to investigate this further.
The double-bundle anatomy of the ACL was first described in 193833 and was confirmed by Girgis et al.34 in 1975. The bundles are believed to act in concert to provide anterior translational and rotational stability to the tibia on the femur. Although some clinical studies have shown improved results for double-bundle versus single-bundle reconstructions in terms of anterior laxity and pivot-shift examination findings,10, 11, 14, 19 several have shown equivalent results15, 35 and none has shown improved patient outcomes. Analysis of these studies is difficult because of widely variable surgical and experimental methods. The tensioning protocols described are highly variable, with tensioning of the grafts at a wide range of knee flexion angles, using different amounts of initial tension, which is not uniformly measured. Yet, the tensioning protocol used has been shown to affect initial loading of ACL grafts.13, 16, 17, 36, 37, 38
Yasuda et al.39 investigated the effect of differential tensioning at 20° in an anatomic DB-ACLR using sutures rather than grafts before final reconstruction in 30 patients. They found that tensioning both suture bundles equally at 30° resulted in a reciprocal tensioning pattern through range of motion, with the AM suture bundle being tighter in flexion and the PL suture bundle being tighter in extension relative to the other bundle; whereas if the sutures were tensioned in a 2:1 ratio (AM to PL), this pattern was not restored, and the tension in the AM suture bundle was higher than that in the PL suture bundle at all flexion angles except full extension, where they were equal. These findings are similar to those in our study. However, tension in each bundle was not determined dynamically, and knee kinematics were not measured. Furthermore, the tension profile of sutures in an anchor may not sufficiently simulate the tension profile of soft-tissue grafts in tunnels.
In a cadaveric model Cuomo et al.17 investigated the effects of different tensioning protocols on anterior and rotational laxity. They used 3 tensioning protocols: sequential tensioning of the AM bundle at 90° and PL bundle at 20°, sequential tensioning of the PL bundle at 20° and AM bundle at 90°, and tensioning of both bundles at 20°. The amount of tension was determined by the amount required to match the intact knee's anteroposterior (AP) laxity. They found that tensioning the PL bundle first required the most tension to restore AP laxity (48 N). The tensions in each bundle required to restore normal laxity in the other tensioning protocols were 17 to 19 N. Tensioning both bundles at 20° best restored normal AP and rotational laxity. Tensioning the PL bundle first at 20° led to too much tension on it; tensioning the AM bundle first at 90° led to unpredictable laxity restoration. The results of this study are consistent with those of our study, although the techniques were different and the sequence of tensioning was not varied in our study. Similarly, Mae et al.38 investigated the effect of flexion angle on relative motion of the tibia on the femur in single-bundle ACL reconstructions in cadaveric knees. They tensioned the grafts to 44 N at 0°, 20°, and 90° of knee flexion. They found that the relative motion of the tibia on the femur as the knee was brought through PROM was closest to the intact knee when the grafts were tensioned with the knee at 20°. In addition, they measured the load between the femur and tibia and found that this was also closest to the values in the intact knee when the grafts were tensioned with the knee at 20°.
In this study bundle-tensioning protocols were shown to significantly affect bundle-loading patterns in anatomic DB-ACLR. Because the bundle tensions were statistically equivalent in the 20/20 protocol, the biomechanical benefit of reconstructing 2 separate bundles in 4 tunnels is unclear, especially given the increased technical complexity of the procedure. Long-term clinical studies are necessary to determine whether there is any clinical benefit of an anatomic DB-ACLR over single-bundle techniques. Additional research is needed to further investigate the importance of tunnel position on loading patterns because small variations in tunnel position appeared to have notable effects in 1 specimen in this study and dramatic effects in another. Additional information is also needed regarding the assessment of rotational stability and the potential clinical implications of the observation that the center of rotation of the tibia on the femur was variable depending on the state of the ACL.
This study has several limitations. The overall number of specimens was low, which increases the possible effect of outlying data. Although the change in center of rotation was seen by use of CT reconstructions for 1 of the specimens, an exact measurement of this point would yield valuable data if done for all specimens. As with any cadaveric reconstruction model, this study is only useful to investigate the time-zero reconstruction characteristics of DB-ACLR. It has been shown that over time, grafts tend to lose some of their initial tension.40 Biological and material changes that occur related to the grafts and fixation devices after surgery are critically important to investigate and understand in the quest to restore a normal knee to patients after ACL injury.
Conclusions
Bundle-tensioning protocol is a variable that can have a significant effect on the bundle-loading patterns in DB-ACLR. The 20/20 tensioning protocol resulted in AM and PL bundle–loading patterns that were equivalent during dynamic testing, not truly reciprocal as hypothesized, whereas the 45/15 protocol led to excessive tension in the AM bundle in full extension. We recommend equal tensioning of both bundles with the knee at 20° of flexion to restore relatively normal tension curves in each bundle, as well as to avoid excessive stress on the AM bundle.
Acknowledgment
The authors acknowledge Matthew T. Thompson, M.S., and Sabir K. Ismaily, B.S., Institute of Orthopedic Research and Education, Houston, Texas, for assisting in statistical analysis and rotational data analysis.
Supplementary data
Supplementary data.
References
- . Functional anatomy of the ACL. J Bone Joint Surg Br. 1991;73:260–267
- . Direct measurement of strain in the posterolateral bundle of the ACL. J Biomech. 1997;30:281–283
- Distribution of in situ forces in the anterior cruciate ligament in response to rotatory loads. J Orthop Res. 2004;22:85–89
- Force sharing between two grafts in the anatomical two-bundle ACL reconstruction. Knee Surg Sports Traumatol Arthrosc. 2006;14:505–509
- . The role of the anteromedial and posterolateral bundles of the anterior cruciate ligament in anterior tibial translation and internal rotation. Am J Sports Med. 2007;35:223–227
- . Anatomic, radiographic, biomechanical and kinematic evaluation of the ACL and its 2 functional bundles. J Bone Joint Surg Am. 2006;88:2–10
- Excessive tibial rotation during high-demand activities is not restored by anterior cruciate ligament reconstruction. Arthroscopy. 2005;21:1323–1329
- Abnormal rotational knee motion during running after anterior cruciate ligament reconstruction. Am J Sports Med. 2004;32:975–983
- Strain pattern comparison of double- and single-bundle anterior cruciate ligament reconstruction techniques with the native anterior cruciate ligament. Arthroscopy. 2007;23:1210–1217
- . Double-bundle versus single-bundle anterior cruciate ligament reconstruction: A prospective randomized clinical study. Knee Surg Sports Traumatol Arthrosc. 2007;15:500–507
- A prospective randomized study of 4-strand semitendinosus tendon anterior cruciate ligament reconstruction comparing single-bundle and double-bundle techniques. Arthroscopy. 2007;23:618–628
- . Anatomic double-bundle ACL reconstruction using tibialis anterior allograft. Oper Tech Sports Med. 2007;15:62–67
- . Effects of knee flexion angles for graft fixation on force distribution in double-bundle anterior cruciate ligament grafts. Am J Sports Med. 2006;34:577–585
- . Prospective randomized comparison of double-bundle versus single-bundle anterior cruciate ligament reconstruction. Arthroscopy. 2008;24:137–145
- . Reconstruction of the ACL with a semitendinosus tendon graft: A prospective randomized single blinded comparison of double-bundle versus single-bundle technique in male athletes. Knee Surg Sports Traumatol Arthrosc. 2008;16:232–238
- . Determination of a safe range of knee flexion angles for fixation of the grafts in double-bundle anterior cruciate ligament reconstruction: A human cadaveric study. Am J Sports Med. 2007;35:1513–1520
- . The effects of different tensioning strategies on knee laxity and graft tension after double-bundle anterior cruciate ligament reconstruction. Am J Sports Med. 2007;35:2083–2090
- . The in vivo kinematics of the anteromedial and posterolateral bundles of the anterior cruciate ligament during weightbearing knee flexion. Am J Sports Med. 2007;35:547–554
- . A retrospective study of the midterm outcome of 2-bundle ACL reconstruction using quadrupled semitendinosus tendon in comparison with 1-bundle reconstruction. Arthroscopy. 2006;22:252–258
- Arthroscopic evaluation of ACL grafts reconstructed with the anatomical two-bundle technique using hamstring tendon autograft. Knee Surg Sports Traumatol Arthrosc. 2007;15:720–728
- Anatomic reconstruction of the AM and PL bundles of the ACL using hamstring tendon grafts. Arthroscopy. 2004;20:1015–1025
- . The effect of axial tibial torque on the function of the anterior cruciate ligament: A biomechanical study of a simulated pivot shift test. Arthroscopy. 2002;18:394–398
- . Can normal knee kinematics be restored with unicompartmental knee replacement?. J Bone Joint Surg Am. 2005;87:332–338
- Three-dimensional morphology and kinematics of the distal part of the femur viewed in virtual reality. J Bone Joint Surg Am. 2003;85(Suppl 4):97–104
- . Tibiofemoral movement 1: The shapes and relative movements of the femur and the tibia in the unloaded cadaver knee. J Bone Joint Surg Br. 2000;82:1189–1195
- . Tibiofemoral movement 2: The loaded and unloaded living knee studied by MRI. J Bone Joint Surg Br. 2000;82:1196–1198
- . Kinematic analysis of conventional and high-flexion cruciate-retaining total knee arthroplasties. J Arthroplasty. 2005;20:529–535
- . Tunnel positioning of anteromedial and posterolateral bundles in anatomic anterior cruciate ligament reconstructions: Anatomic and radiographic findings. Am J Sports Med. 2008;36:65–72
- . Femoral insertion of the ACL (Radiographic quadrant method). Am J Knee Surg. 1997;10:14–21
- . Tibial attachment area of the anterior cruciate ligament in the extended knee position (Anatomy and cryosections in vitro complemented by magnetic resonance arthrography in vivo). Knee Surg Sports Traumatol Arthrosc. 1994;2:138–146
- . Anatomical and non-anatomical double-bundle anterior cruciate ligament reconstruction: Importance of femoral tunnel location on knee kinematics. Am J Sports Med. 2008;36:678–685
- . Direct measurement of resultant forces in the anterior cruciate ligament: An in vitro study performed with a new experimental technique. J Bone Joint Surg Am. 1990;72:557–567
- . On the injuries to the ligaments of the knee joint. Acta Chir Scand. 1938;91:282
- . The cruciate ligaments of the knee joint: Anatomical, functional and experimental analysis. Clin Orthop Relat Res. 1975;216–231
- Reconstruction of the ACL: Single vs. double-bundle multistranded hamstring tendons. J Bone Joint Surg Br. 2004;86:515–520
- The effect of graft tensioning in anatomic 2-bundle ACL reconstruction on knee joint kinematics. Knee Surg Sports Traumatol Arthrosc. 2007;15:508–514
- . Optimization of graft fixation at the time of anterior cruciate ligament reconstruction: Part 1: Effect of initial tension. Am J Sports Med. 2008;36:1087–1093
- . Optimization of graft fixation at the time of anterior cruciate ligament reconstruction: Part 2: Effect of knee flexion angle. Am J Sports Med. 2008;36:1094–1100
- . An in-vivo biomechanical study on the tension-versus-knee flexion angle curves of 2 grafts in anatomic double-bundle anterior cruciate ligament reconstruction: Effects of initial tension and internal tibial rotation. Arthroscopy. 2008;24:276–284
- . The effect of initial graft tension on the biomechanical properties of a healing ACL replacement graft: A study in goats. J Orthop Res. 2003;21:708–715
The authors report no conflict of interest.
PII: S0749-8063(09)00613-6
doi:10.1016/j.arthro.2009.07.014
© 2010 Arthroscopy Association of North America. Published by Elsevier Inc. All rights reserved.
Volume 26, Issue 2 , Pages 202-213, February 2010
