Materials
Six intact fresh-frozen human cadaver legs were examined. Donors were aged between 21 and 53 years old, 2 females and 4 males, 3 knees were left and 3 were right. All knees showed normal anatomy and a normal passive range of motion. Specimens were stored at -20°C from the time of retrieval and were thawed overnight at room temperature prior to testing, thereby undergoing a single freeze cycle. In each specimen the complete femur, the tibia and the foot were maintained intact during the experiment. The skin and muscles surrounding the femur, more than 20 cm from the joint line, were removed so that the bone was exposed for better fixation to the experimental set up. The specimens were kept moist during preparation and testing, by spraying with 0.9% normal saline solution. The anatomical and kinematic acquisitions were performed using FlashPoint optical navigation system (Image Guided, Boulder, Colorado, USA) to accurately record the relative motion of the tibia and the femur and to digitize anatomical data [10, 11].
Experimental procedure
The femur was fixed to the experimental desktop with a clamping device at 90° of flexion, by suitable supports at the thigh. The tibia was left free to move as in the standard operating room set-up. The intact foot enabled the surgeon to check the internal-external alignment of the limb and use mobile supports at the foot to stabilize the leg better during laxity tests in flexion (Fig. 1).
Two custom rigid bodies supporting a set of standard FlashPoint infrared emitters were fixed respectively to the femur and tibia, to record their relative position during passive motion tests. In order to minimize possible interactions with surgical acts, the femoral rigid body was fixed in the proximal part of the femur and the tibial one distally in the medial part of the tibia (Fig. 1). The accuracy of the system evaluated in vitro is 0,3 mm and 0,3° [12].
First, the intact knee was examined by kinematic tests simulating the standard clinical examination used to diagnose ACL deficiency. In particular, we acquired anterior-posterior drawer tests and internal-external rotations at 90° of knee flexion performed manually at maximum force. This has shown to be quite repeatable in literature [12]. All tests were repeated three times by the same surgeon. We decided to not perform the test at 30° because the mainteinance of the position during tests could be difficult and the gravity of the specimens as well as the meniscal and cartilagineous conditions of the knees could have in certain way affected the repeatability of our measurements. With our set up we consider that the examination at 90° should be the one more consistent during all tests. These 6 degrees of freedom (DOF) were recorded at 18 Hz with the FlashPoint system. Since all the kinematic data were analyzed off-line, the surgeon was blinded with regards to the result of laxity tests. Then the ACL was transected with a minimal incision and kinematic tests were repeated identically for the ACL-deficient knee.
Subsequently, two ACL reconstructions were performed on the same knee, using previously harvested semitendinosus and gracilis tendons sutured together with the tibial insertion left intact [13]. Both ACL reconstructions used the same single tibial tunnel, drilled 50 mm deep from the tibial spine, medially with respect to the cresta tibialis. The tibial tunnel was directed to the ACL area by outside-in, oriented from medial to lateral in the frontal view and from anterior to posterior in the sagittal view [13]. Two femoral tunnels were made starting from the same area, at the so-called "10:30 hours" position inside the femoral ACL insertion. Femoral tunnels were performed in-out from the antero-medial portal with the knee flexed at 130° using a Kirschner wire and were drilled using a 6-mm diameter cannulated drill (Acufex, Smith&Nephew Inc., Andover, D). For the first femoral tunnel, hereafter "horizontal tunnel" (HT), the surgeon directed the tunnel towards the end of lateral posterior condyle, while for the second femoral tunnel, hereafter "vertical tunnel" (VT), the tunnel was directed 20 mm more proximal with respect to Horizontal Tunnel exit, laterally on the femoral shaft (Fig. 2).
The graft was pulled through the tibial tunnel and passed into the femoral notch as a postero-lateral bundle by an over-the-top passage performed using a curved Kelly clamp. The graft was then taken from the lateral side of the knee and passed through one femoral tunnel back distally in the joint and into the tibial tunnel again, thus building the antero-medial bundle of the reconstructed ACL (Fig. 3a, 3b).
The procedure was repeated for the second reconstruction passing the graft in the other femoral tunnel. In 3 alternated cases the horizontal tunnel was used first and in 3 alternated cases the vertical tunnel was used first. The prepared graft was preconditioned by moving the knee through 5 cycles of the full range of flexion applying 22 N tension to the graft. Both ACL reconstructions were fixed to the tibia by means of two titanium barbed staples (Citieffe Inc., Bologna, Italy) set at 10° flexion after applying maximum load. The laxity tests were recorded respectively after the horizontal tunnel and vertical tunnel reconstruction. The knee was dissected only after surgery, thus exposing ligaments and bone surfaces in order to digitize accurate data about knee anatomy. A standard joint coordinate reference system (i.e. mechanical axis of femur and tibia, transepicondylar line, joint line and knee center), ACL insertions, tunnel entrance and exit holes, distal femur and proximal tibia surfaces were acquired.
Analysis
Analysis of the knee joint kinematics was performed off-line with a custom dedicated software [14, 15, 2, 16]. Three-dimensional anatomical data were displayed in a reference system defined on the femur as follows: (1) proximo-distal axis was defined as the tibial mechanical axis in full extension; (2) medio-lateral axis was defined as the transepicondylar line, normalized with respect to (1); antero-posterior axis was defined as the instantaneous cross product of (1) and (2) (Fig. 4). Kinematic data were used to compute the antero-posterior (AP) laxity and the internal-external (IE) laxity of the joint at 90°, using Grood and Suntay decomposition algorithm [17, 18].
We also computed the angles described by the two different tunnels in the frontal plane with respect to the femoral mechanical axis and the final length of the graft trajectory in the two reconstructions in extension as the sum of the following values: twice the length of the tibial tunnel, the length of the femoral tunnel, the length of the over-the-top passage of the graft around the posterior condyle, the linear length of the two bundles of the graft replacing the ACL in the femoral notch.
Statistical analysis was performed to assess the precision of the results and the differences between the reconstructions using the two tunnels. In particular, the repeatability of each test was estimated as the percent mean standard error of the three acquisitions of the test on each specimen and a statistical descriptive analysis was made for all results, using the mean values of repeated tests for each specimen.
The global effect of the two reconstructions was estimated by performing the Wilcoxon test with the Montecarlo method for small samples, applied between ACL-deficient knees and horizontal tunnel reconstructions and between ACL-deficient knees and vertical tunnel reconstructions, respectively. Additionally, the functional restoration was evaluated by performing Wilcoxon test with the Montecarlo method for small samples between ACL-intact knee and horizontal tunnel reconstructions and between ACL-intact knee and vertical tunnel reconstructions, respectively. Moreover the Wilcoxon test with the Montecarlo method for small samples was performed to evaluate the difference between horizontal tunnel reconstruction and vertical tunnel reconstruction.