Introduction

Transtibial single-bundle (SB) anterior cruciate ligament (ACL) reconstruction has been widely advocated until recent investigations revealed that with this technique, it is difficult to achieve anatomic graft placement [1, 4, 16, 24]. Non-anatomic reconstructions have been shown to result in abnormal knee kinematics and further progress to osteoarthritis over time [14, 27, 31, 39, 44]. Anatomic placement using a far anteromedial or accessory medial portal, that is, transportal technique, has been advocated for both SB and double-bundle (DB) reconstruction in order to achieve anatomic graft placement and improved knee rotational stability [6].

Rotational stability after DB ACL reconstruction has been extensively researched [37, 38, 46, 48, 50], but there is lack of rotational studies of the anatomic SB ACL reconstruction in spite of its increasing popularity [23]. Over-the-top (OTT) ACL reconstruction, in which the graft is passed over the superomedial border of lateral femoral condyle and fixed on the lateral femoral shaft over the condyle, is often used as salvage option for revision cases and in skeletally immature patients [3, 30, 45, 47]. OTT ACL reconstruction also has not been fully examined for rotational stability. Since this procedure is fundamentally non-anatomic, the resultant knee kinematics are assumed to be inferior to the anatomic ACL reconstruction. However, it is still unknown how much these two clinically prevalent techniques effect to the knee stability, especially rotational stability.

The rotational instability of the knee is clinically assessed by the pivot shift test which is basically a subjective evaluation [13, 25, 33], while its quantitative evaluation is yet to be established. However, recent research efforts and advanced technology have facilitated the investigation of quantitative assessment of the pivot shift by measuring dynamic parameters, such as velocity or acceleration, during the manually performed pivot shift test [2, 7–9, 17, 26, 32]. The increased acceleration during the pivot shift test in the ACL-deficient knees by the use of triaxial accelerometry, which can be used as an experimental method for exploring the resultant knee rotational stability after certain types of ACL reconstructions, was reported in previous study [36].

Therefore, the purpose of this study was to investigate the dynamic rotational instability of anatomic SB and the OTT ACL reconstructions during a pivot shift test using triaxial accelerometry. It was hypothesized that the anatomic SB reconstruction would provide better dynamic rotational stability than the OTT procedure.

Materials and methods

Eight fresh frozen human cadaveric knees were used in this study. Each specimen was screened for any bony and ligamentous abnormalities by manual examination, CT scan and arthroscopy. All specimens were stored at −20 °C and thawed at room temperature 24 h before testing. The femur and the tibia were cut approximately 20 cm from the joint line. All soft tissue was removed from the cut ends of femoral and tibial shafts in 10 cm while keeping the knee joint intact. Both exposed femoral and tibial shafts were secured in the cylindrical-shaped epoxy compound (Bondo, Atlanta, Georgia, USA) for mounting in custom-made aluminium clamps.

Rotational instability measurement

Rotational instability measurement using a triaxial accelerometer was conducted during a manually performed pivot shift test, repeatedly in the ACL-intact, ACL-deficient and ACL-reconstructed knees with two different procedures (anatomic SB and OTT). A single examiner performed the pivot shift test in all knees in order to exclude inter-examiner variation. The pivot shift test was performed according to the report of Hughston et al. [18], which reproduces the dislocation phase of the pivot shift phenomenon from flexion to extension [18].

The three-dimensional acceleration (x-axis: anteroposterior; y-axis: mediolateral; z-axis: superoinferior) was measured with a triaxial accelerometer (Kistler, Winterhur, Switzerland) during the pivot shift test. The sensor was rigidly attached to the tibial tubercle with a bone screw (Fig. 1). Signals from the sensor were digitally recorded at a sampling rate of 10 kHz, while the waveforms expressing the acceleration were displayed on a computer screen (Fig. 2). The acceleration in each direction was then analysed and the overall magnitude of acceleration was calculated with use of the following formula: {|a| = √(ax 2 + ay 2 + az 2)}.

Fig. 1
figure 1

Human cadaveric knee mounted in custom fixture. The triaxial accelerometer was attached to the tibial tubercle with a screw

Full size image
Fig. 2
figure 2

A typical example of waveforms in ACL-deficient condition during pivot shift test. Impulses are provoked at the time of pivot shift

Full size image

Measurements were repeated at least three times in each knee state. The median value of three trials was used for comparison, while the other two results were used to assess the repeatability of the measurements. The overall magnitude of acceleration and each directional acceleration were compared between four different ACL states.

The reliability, repeatability and accuracy of this measurement were previously reported in the studies of dynamic instability [11, 34–36]. In the previous studies, the triaxial accelerometer showed a high repeatability of the measurements with a correlation coefficient over 0.95. The accuracy of the measurement for acceleration is given by the sensitivity of the accelerometer device, which is 1,000 mV/g.

Surgical technique

Both ACL reconstructions were performed arthroscopically using a three-portal technique comprising anterolateral, central and accessory anteromedial portals [10, 43].

For the anatomic SB ACL reconstruction, a guide pin was passed at the centre of the femoral insertion site of the ACL through the accessory anteromedial portal. The femoral tunnel was drilled with 8-mm-diameter acorn reamer. Subsequently, the 8-mm-diameter tibial tunnel was drilled at the centre of the tibial insertion site of the ACL. An 8-mm-diameter hamstring graft was passed through both femoral and tibial tunnels and fixed with EndoButton CL (Smith & Nephew, Andover, Massachusetts, USA) on the femoral side. The graft was preconditioned by cyclic flexion–extension loading for 10 cycles while applying a 44 N of graft tension. Tibial fixation was then performed at 30° of knee flexion using a staple with an initial tension of 44 N and augmented by suture to screw post-fixation.

For the OTT ACL reconstruction [47], a lateral incision was made at the distal femur to identify the OTT position, that is, posterolateral aspect of the distal femoral metaphysis, by palpation. A small hole was made in the posterior capsule from inside out by a curved clamp into the OTT position. The same size of hamstring graft was passed through the same location of the tibial tunnel as the anatomic SB procedure and fixed onto the OTT position with CL device (Smith & Nephew, Andover, Massachusetts, USA) and a post-screw at the lateral femoral metaphysis. The graft was preconditioned and fixed to the proximal tibia in the same fashion as the anatomic SB technique (Fig. 3).

Fig. 3
figure 3

Arthroscopic view of left knee after anatomic SB a and OTT b reconstruction. ACL anterior cruciate ligament, PCL posterior cruciate ligament, LFC lateral femoral condyle, MFC medial femoral condyle, OTT over-the-top

Full size image

Statistical analysis

Statistical analysis for differences in the median acceleration during the pivot shift test among the four different ACL states was performed using the Kruskal–Wallis test, and differences between two groups were then tested by a post hoc Mann–Whitney U test. Statistical significance was assumed when P < 0.05 for Kruskal–Wallis test and P < 0.01 for Mann–Whitney U test. All statistical results were calculated using a statistical software package (SPSS ver. 17.0).

Results

A post hoc power analysis showed that a sample size of 8 specimens for each group would yield statistical power of 0.81.

The overall magnitude of acceleration was significantly different between the ACL-intact knees and the ACL-deficient knees (P < 0.01, Table 1). Both the anatomic SB and OTT ACL reconstructions significantly reduced the overall magnitude of acceleration compared to the ACL-deficient knees (P < 0.01), but still had larger accelerations compared to the ACL-intact knees (P < 0.01). There was no significant difference between anatomic SB and OTT reconstruction procedure (Fig. 4).

Table 1 Acceleration measurement results
Full size table
Fig. 4
figure 4

Overall magnitude acceleration in 4 different knee conditions. *P value < 0.01

Full size image

The acceleration in the x direction, or anterior direction of the tibial tubercle, was not significantly different between four different ACL states. The acceleration along the y-axis, lateral direction of the tibial tubercle, was significantly different between the ACL-intact knees and the ACL-deficient knees (P < 0.01). Both the anatomic SB and OTT ACL reconstruction significantly reduced the acceleration in y direction compared to the ACL-deficient knees (P < 0.01), but still had larger accelerations compared to the ACL-intact knees (P < 0.01). There was no significant difference in the acceleration in the y direction between anatomic SB and OTT reconstruction procedure. For the acceleration in z direction, proximodistal direction of the tibial tubercle, there was a significant difference between the ACL-intact knees and the ACL-deficient knees (P < 0.01); however, neither the anatomic SB nor OTT ACL reconstructions significantly reduced the acceleration from the ACL-deficient status (P = 0.09 and 0.06, respectively) and had larger accelerations compared to the ACL-intact knees (P < 0.01). There was no significant difference for the acceleration in z-axis between anatomic SB and OTT reconstruction procedures.

Discussion

The most important findings of this study are that anatomic SB and OTT ACL reconstruction significantly reduced the overall magnitude of acceleration compared to the ACL-deficient knees, but still had larger accelerations compared to the ACL-intact knees. Contrary to the hypothesis, SB and OTT ACL reconstruction were not significantly different in controlling the rotational stability. In addition, the largest difference was in the lateral component of acceleration of the tibial tubercle, which could represent rotation of the tibia.

The effect of tunnel placement on knee rotational laxity using SB ACL reconstruction has been reported with a favourable result in lower femoral placement, although it remains unknown if the tunnel location is truly anatomic. Scopp et al. [41] found that rotational laxity of cadaveric knees was restored to normal with an oblique femoral tunnel placement [41]. Loh et al. [31] reported that grafts in the 10-o’clock position were able to resist rotator loads more effectively than grafts in the 11-o’clock position [31]. Thus, low graft placement has been recommended for SB ACL reconstruction with the expectations of better rotational laxity, but the tunnel location suggested by the clock-face reference cannot be consistent and reliable for three-dimensional structure of the intercondylar notch [12]. Meanwhile, this study is different in that the femoral and tibial tunnel placements were directly confirmed at the centre of the femoral and tibial original footprint using the transportal technique, and the dynamic rotational instability was assessed instead of the rotational laxity.

Although the OTT is a non-anatomic procedure, it is still used as a salvage option for limited cases to avoid drilling in the lateral condyle, such as a revision cases with a significant bone loss or a skeletally immature individual with an open physis [15, 21, 42, 45]. However, little is known about the resultant knee kinematics or rotational stability after the OTT ACL reconstruction. Clinical reports demonstrated that there was no significant difference in hamstring reconstructions by either OTT or via a femoral tunnel [19, 20]. Radford et al. [40] compared the static laxity at 20° and 90° knee flexion with the biomechanics of single-bundle OTT and femoral tunnel techniques and showed that the OTT reconstruction gave better anteroposterior stability and functioned like an intact ACL at 20° of knee flexion [40]. A recent study by Lertwanich et al. [29] reported that both anatomic SB and OTT techniques closely restored the intact knee kinematics, and there was no difference between anatomic SB and OTT technique in the kinematics with an anterior tibial load and a combined rotatory load [29]. Rotational instability is clinically evaluated by pivot shift test [13, 33], which depends on the examiner’s technique for the pivot shift movement during dynamic testing procedure. The presence of a positive pivot shift test is predictive of osteoarthritis of the knee, failure to return to previous level of play, patient reported instability and poor subjective outcome scores following ACL reconstruction [5, 22, 28, 49]. Bignozzi et al. [7] concluded that the dynamic evaluation of a pivot shift is better able to describe dynamic rotational instability [7]. Hoshino et al. [17] measured the tibial anterior translation and calculated the acceleration of the tibial posterior reduction using an electromagnetic device [17], and Lane et al. [26] measured similar acceleration of the tibial movement using a computer navigation system [26]. They both demonstrated increased acceleration of the ACL-deficient knees [17, 26], similarly to this study. Furthermore, this study evaluated acceleration after different ACL-reconstructed procedures, anatomic SB and OTT procedure, and found that both procedures could successfully reduce the dynamic instability of ACL deficiency, but failed to restore the intact level.

There are some limitations in this study. First, the knees were evaluated immediately after surgery, and consequently, the results represent time-zero conditions. The healing of the graft at the OTT position may be different from that of the femoral tunnel technique which utilizes circumferential tendon-to-bone contact. After time to develop adequate graft corporation into the bone, anatomic SB ACL could have different biomechanical properties than the replaced ACL on the OTT position. Second, manual tests like the pivot shift have a wide variability among examiners. In order to exclude such variability, the manual tests were performed by a single experienced examiner and the repeatability was re-evaluated in this study. The repeatability, that is, 331 ± 218 mm/s2 as a standard deviation, was consistent with our previous reports and other studies [17]. Last, since the acceleration was measured on the tibial tubercle, the direction of the acceleration change could not be directly described on the basis of 6°-of-freedom knee coordinate system which is commonly used for defining knee motion. Interestingly, our result indicated increase in the acceleration in the medial movement of the tibial tubercle, which appears to represent tibial rotation, while most of the previous report using 6°-of-freedom coordinate system demonstrated the increase in the acceleration in the tibial anteroposterior direction [17, 26]. Although the evaluated phase of the pivot shift test was also different, that is, the dislocation phase in this study and the reduction phase in the previous studies [17, 26], further investigation is warranted to examine how our measured acceleration can be explained by the common 6°-of-freedom knee kinematics.

Anatomic ACL reconstruction has been increasing its popularity. For revision cases and in skeletally immature patients, OTT ACL reconstruction is often used as salvage option. This study showed that both reconstructions had no significant differences in dynamic stability. OTT reconstruction might be one of the options for revision cases and in skeletally immature patients.

Conclusion

Although OTT reconstruction is used for limited cases, it provides comparable result to anatomic SB reconstruction in terms of controlling the dynamic rotational instability.