Patellofemoral kinematic characteristics in anterior cruciate ligament deficiency and reconstruction

Twenty unilateral ACL deficient patients were recruited as preoperative group (10 men and 10 women,20 to 37 years old, average body mass index, 24.1 ± 4.8 kg/m²). Six months after ACL reconstruction, these subjects were included as postoperative subjects. The subjects were included based on intra-operative diagnosis as isolated ACL deficiency. Concomitant knee musculoskeletal disorders or anatomical abnormalities in ACL-D knees were excluded. ACL-D subjects with a history or evidence of injury, or surgery in the contralateral knees were also excluded. For the control group, 10 subjects with healthy knees (6 men and 4 women, 18 to 33 years old, average body mass index, 22.8 ± 6.3 kg/m²) were recruited. No knee trauma, pain, and abnormality of movement were found in the healthy subjects. All healthy subjects choose their right knees as the examined knees. Before testing, all the included subjects signed consent form. The institutional review board in authors institute approved the current study design before the initiation of the study.

The surgical procedure was standardized and was performed by one experienced surgeon in all patients. The tibial tunnel was addressed first. The centre of the tunnel was placed in line with the anterior horn of the lateral meniscus. All the bone tunnels were drilled 0.5 mm larger that the diameter of the respective grafts, which were between 7.5 and 8.5 mm. The ACL graft consisted of 4-stranded semitendinosus and gracilis tendons. Tibial fixation was achieved with a 7–9 mm interference screw (RCI, Smith & Nephew, Andover, Massachusetts, USA). On the femoral tunnel side, the femoral ACL insertion site was marked with an awl in the shallow aspect of the AM bundle insertion site. The femoral tunnel was predrilled with a 4.0-mm sharp non-cannulated drill. Femoral fixation was achieved with an Endobutton (Acufex, Smith & Nephew, Andover, Massachusetts, USA). Graft tensioning was performed at 10° to 20° of knee flexion (Fig. 1).

Segmentation of computed tomography data (CT, SOMATOM Definition; Siemens, Munich, Germany) was performed for the knee joint. Parallel CT digital images with a thickness of 1 mm without a gap and with a resolution of 512 × 512 pixels were obtained. These data were then imported into solid modeling software Mimics 17.0(Materialise, Leuven, Belgium) to outline the contours of the femur, the patella, and tibia manually. The construction of 3D geometric models of these three bone were done by these outlines.

Next, we used a single-plane fluoroscopic imaging system to investigate the 6 degree of freedom (6 DOF) kinematics in the knee joint. This system has been validated for treadmill gait analysis in our previous published study [11]. Laser-positioning devices, were used to align the target knees within the field of view of the fluoroscopes when subjects were ascending the stairs (Fig. 2a). We asked each subject to walk up a custom set of stairs which was validated in a previous study [11]. This custom set of stairs, with each step 18 mm high, 20 mm deep and 40 mm wide, were designed within published ergonomic recommendations [12, 13]. To reduce postural variations, the subjects’ initial positions were carefully examined by an Orthopaedic surgeon. A rhythmic alarm was used to help the patients ascend the stairs at a fixed pace. No constraint was applied to the knees of subjects while they were ascending the stairs. To simulate normal daily activity, subjects ascended the stairs at a self-selected pace. After measurement, fluoroscopic images were processed in the Digital Imaging and Communications in Medicine format.

Selected fluoroscopic images, identified at a specific posture, were imported into the registration software Virtual_knee1.0 (Medmotion, Guangzhou, China). The actual positions of the image intensifiers of the fluoroscopes were then reproduced (Fig. 2b). Meanwhile, the CT image-based 3D knee models were introduced into the fluoroscopic system. To make the outlines of knee joint segmentation match the osseous outlines captured on the single-plane fluoroscopic images, these models were independently translated and rotated in 6DOF. This process was conducted with an established protocol [10]. The models could be manually rotated using the software (Virtual_knee1.0) with an accuracy of 1.57° for rotation and 0.2 mm for translation [14]. Manual matching was first performed, followed by an automated matching. Previously we measured three times on one subject in a same condition using manual matching technique and measurement was done in 5 subjects by one researcher. The intra-rater reliability of manual matching technique is 0.893. After matching, the knee positions during in vivo weight-bearing activities were reproduced, representing the 6DOF kinematics of the knee for each in vivo posture.

A coordinate system based on a protocol by Suntay et al. [15] was used to calculate the kinematics of knees based on the matched bone models (Fig. 3). The algorithm was developed based on our previous study [16]. A “four-points” method was applied to build coordinate systems in the femur, the patella and the tibia. In the femur, two prominent points located on the medial and lateral femoral epicondyles were selected. Another two points that located parallel to the wall of the femur shaft were selected. The transepicondylar line was obtained by linking the most pivot points on the medial and lateral condyles. The femoral origin was located at the midpoint of the transepicondylar axis. The line that is parallel to the shaft of the femur was defined as the long axis of the femur. In the patella, the first two points were the most medial and lateral points on the patella; the other two points were the most proximal and distal points on the patella. The line connecting to the most medial and lateral points on the patella was defined as the medial-lateral axis, and the midpoint of this line was defined as the origin of the patella. In the tibia, the most external points on sides of the medial and lateral tibia plateau were selected. Another two points that located parallel to the wall of the tibia shaft were selected. The line connecting to the most pivot points on the medial and lateral tibia plateau was defined as the medial-lateral axis, and the midpoint of this line was defined as the origin of the tibial coordinate system. Patellofemoral rotation and translation was defined as the patella center move with respect to the origin in the femur coordinate system We measured the 6 DOF kinematics in the PFJ, including 3 rotational freedom (medial-lateral patellar tilting; medial-lateral patellar rotations; and patellar flexion-extension) and 3 translational freedom (medial-lateral patellar translations; anterior-posterior patellar translations; and proximal-distal patellar translations).

A two-way repeated ANOVA was used to compare the patellofemoral kinematics. The dependent variables were 6 DOF kinematics in the PFJ as mentioned above. The independent variables were the test conditions (ACL-I, ACL-D and ACL-R) and flexion angles (0°, 30°, 60°, 90°, 120°). We applied the Bonferroni to post hoc multiple comparisons when a statistically significant difference was detected (0.05 significance level). The statistical analysis was performed using commercially available software (SPSS for windows 13.0, Chicago, IL, USA).

In ACL-I group, the patellar flexion angle is about a half of knee flexion (Fig. 4a). Patellar flexion angle significantly reduced in ACL-D group compared with other two groups with from 0°~ 60°(P < 0.05). In ACL-D group, patellar flexion was 4.3°±. 4.0 at 0° of knee flexion and increased to 64.9 ° ± 6.3 at 120°of knee flexion. Compared to control knees, ACL-R knees showed significant decreased patellar flexion at lower flexion (P < 0.05). There is no significant difference detected in these three groups after 60°.

Generally, the patella showed increased external rotation when the knee flexed (Fig. 4b). For internal/external rotation, ACL-R exhibited an increase of about 3~5°in external rotation compared with other two groups at lower flexion (P < 0.05). ACL-R group peaked at patellar external flexion (8.3 ± 13.5°) when the knee extended and then continuously decreased. Significant difference was only detected at 60 of knee flexion between ACL-I group and ACL-D group (2.4 ± 7.6° VS 5.8 ± 5.4°).

ACL-D group and ACL-R group showed sharp increased lateral tilt during the whole knee flexion procedure (Fig. 4c). From 0° to 90° ACL-I group revealed a significant increased patellar flexion angle compared with the other two groups (P < 0.05) and no significant difference was shown between the ACL-D group and ACL-R group. At 120° of knee flexion, all three groups exhibited highest lateral tilt (5.3 ± 1.4°, 7.0 ± 1.2°, and 6.4 ± 1.3° respectively) in their corresponding ACL status.

In general, the patellar showed anterior patellar translation during the knee flexion (Fig. 5a). No significant difference was found among the groups through the whole measured range, even though ACL-D group and ACL-R group showed more anterior translation compared to control knees.

Compared to control knees, ACL-R group showed more lateral translation (about 3 mm) than the figure from control knees (P < 0.05, Fig. 5b). ACL-R group exhibited 10.1 of patellar flexion when the knee extended, followed by decreasing patellar lateral translation as the knee flexed. In particular, ACL-D group showed a significant reduce in lateral translation compared to ACL-R group at 30° of knee flexion (4.8 ± 0.7 mm VS 7.7 ± 1.8 mm).

The patella exhibited a distinguished movement pattern with respect to the description of proximal-distal translation (Fig. 5c). Initially the centre of patella was beneath the femur when the knee extended, up to 16.3 ± 14.7 mm. The patella moved up and was above the femur as the knee extended, up to 20.9 ± 2.3 mm. A significant reduce (about 1.4~7 mm) of distal translation was found in ACL-D group compared with the other groups (P < 0.05) when the knee extended. As the knee flexed, similar difference was shown at 60°, 90°, and 120° of knee flexion. Even though no significant proximal-distal translation was detected between ACL-I group and ACL-R group, the latter group showed less proximal patellar translation of about 3 mm.

There are some limitations in this study. First, a small number of sample size was included. We didn’t perform a statistical power analysis to do sample size calculation as there was no similar study using the same 2D-3D matching technology to evaluate PFJ kinematics. Second, we only performed stair climbing and other functional activities should also be included to evaluate the kinematic alterations of PFJ such as level walking, stair descending, and single leg hop. Third, the mean follow up after ACL-R was about 6 months, which might introduce selective bias in inclusion of subjects. In future study, patients should be followed up at various time intervals to investigate the change in PFJ kinematics over time. Fourth, PFJ kinematics represent the spatial alterations between the patella and the femur. However, as an in-vivo study, some biomechanical parameters such as contact pressure and contact area which are directly associated with biomechanical changes were unable to be collected.