Simulation of patella alta and the implications for in vitro patellar tracking in the ovine stifle joint

Authors

  • Nicky Bertollo,

    Corresponding author
    1. Surgical and Orthopaedic Research Laboratories, Prince of Wales Hospital, University of New South Wales, Level 1, Clinical Sciences Building, Randwick 2031, Australia
    • Surgical and Orthopaedic Research Laboratories, Prince of Wales Hospital, University of New South Wales, Level 1, Clinical Sciences Building, Randwick 2031, Australia. T: +61-2-9382-2687; F: +61-2-9382-2660
    Search for more papers by this author
  • Matthew H. Pelletier,

    1. Surgical and Orthopaedic Research Laboratories, Prince of Wales Hospital, University of New South Wales, Level 1, Clinical Sciences Building, Randwick 2031, Australia
    Search for more papers by this author
  • William R. Walsh

    1. Surgical and Orthopaedic Research Laboratories, Prince of Wales Hospital, University of New South Wales, Level 1, Clinical Sciences Building, Randwick 2031, Australia
    Search for more papers by this author

Abstract

Patella alta is associated with adverse cartilage adaptations, patellofemoral pain, and instability. It is defined by a relatively long patellar tendon and patella positioned in a more proximal location within the patellar groove of the femur. This study used the ovine stifle joint model to investigate the effect of patellar tendon lengthening on the 3D passive kinematics of the patellofemoral and tibiofemoral joints. Eight patellar tendons were lengthened in 2 mm increments up to a maximum of 12 mm (20%) using a device placed in series with the transected patellar tendon. Three-dimensional kinematics were measured in the intact joint and at each increment of patellar tendon length (LT) during passively induced tibiofemoral flexion. Patellar flexion angle was linearly correlated with tibial flexion angle in the intact joint, and this correlation persisted after tendon lengthening (R = 0.897–0.965, p < 0.01). Patellofemoral kinematics expressed as a function of tibial flexion angle were significantly altered by LT increases >9%. In contrast, when patellofemoral kinematics were expressed as a function of patellar flexion angle they were not significantly altered by increases in LT. Tibiofemoral kinematics were not affected by the LT increases. These results demonstrate that for a given tibial flexion angle, patellar tendon lengthening alters the patellar flexion angle. However, for a given patellar flexion angle, the orientation of the patella in the remaining five degrees of freedom is unchanged, implying a repeatable path of patellar motion. © 2012 Orthopaedic Research Society. Published by Wiley Periodicals, Inc. J Orthop Res 30:1789–1797, 2012

Patella alta is a condition characterized by a patella which resides in an abnormal, proximal location within the trochlear groove of the femur, as indicated by an Insall–Salvati (ISI) index of 1.2 or more.1 The ISI of the normal, asymptomatic knee, calculated by dividing patellar tendon (PT) length (LT) with longitudinal patellar height on a lateral radiograph with the leg in 30° of flexion is approximately 1.0.1–6 A 20% or more intrinsic increase in LT can induce patella alta, which has been associated with a number of patellofemoral (PF) pathologies including PF pain,4, 7–10 retropatellar cartilage degeneration2, 3, 11, 12 and patellar instability.2, 4, 13, 14 LT increases of a lesser 10% (ISI of 1.1) have also been implicated in PF pain4, 9 and degenerative retropatellar cartilage changes.2, 3 The relationship between intrinsic PT lengthening, PF pain and cartilage degeneration is not fully understood.

Whilst PF pain syndrome is often present in the absence of patella alta15–17 a proximally displaced patella has been linked to a reduction in retropatellar contact area and elevated PF stress.18–22 It has been postulated that this is the catalyst for degenerative cartilaginous changes leading to PF pain in these patients. Ward and Powers22 found decreased contact area but unaltered joint reaction force during gait in a patient population with patella alta. In vivo measurements obtained by Sheehan et al.10 demonstrated patellar kinematic changes to be prevalent in patients with patella alta exhibiting PF pain. These changes manifested as modified flexion, and a net proximal shift and lateral translation of the patella.

Decreased patellar flexion in patella alta may indeed, be borne of the linear relationship between patellar and tibial flexion angles (referenced to the femoral coordinate frame) which is known to exist in the human knee.10, 17, 23, 24 It is conceivable that a proximal patellar shift caused by PT lengthening would delay patellar flexion for a given tibial flexion angle. If this assumption held true throughout tibial flexion then PT lengthening would cause a mean offset or, lag of the patellar flexion pattern. This flexion lag phenomenon has been demonstrated in an ovine model, where resection of the central one-third of the PT in sheep induced an increase in LT, which persisted up to 6 months postoperatively.25

By virtue of the linear relationship between patellar and tibial flexion angles,10, 17, 23, 24 and considering that the patella follows a circular path of motion around an axis in the femur,23, 26 it follows that each finite position of the patella along this path may be associated with a unique amount of patellar flexion. Furthermore, this flexion pattern may be independent of LT increases, such as in patella alta. This study sought to investigate whether the remaining five degrees of freedom (DOF) of patellar motion also remain unaffected by LT increases of up to 20%, which would imply that the patella follows a repeatable path of motion in its articulation with the femoral trochlea. To this end, we defined and analyzed patellar kinematics as a function of both patellar and tibial flexion angle relative to the femoral coordinate frame, denoted as PFpat and PFtib, respectively. A convention in biomechanics literature is the analysis of patellar kinematics and PF mechanics as a function of tibial flexion angle, that is, PFTib.

In this study a simple device placed in series with the transected PT (Fig. 1) was used to lengthen LT in 2 mm increments, up to maximum of 12 mm (20%), to test the null hypothesis that LT increases would have no effect on in vitro PFtib and PFpat kinematics in the ovine stifle joint during passively induced flexion-extension. The ovine stifle joint model represents a practical alternative to human cadaveric tissue for biomechanical studies due to its size and robustness, and is being increasingly utilised in such research.26, 27–34

Figure 1.

(a) Exploded view of the patellar tendon lengthening device. (b) Following attachment of the device the patellar tendon was transected and patella alta induced. [Color figure can be seen in the online version of this article, available at http://wileyonlinelibrary.com/journal/jor]

MATERIALS AND METHODS

Eight un-paired intact hindlimbs were obtained from sheep (crossbred Merino Wethers, 1.5 years, 54.5 ± 1.4 kg) euthanized for other ethically approved studies that did not involve or impact the stifle. Hindlimbs were disarticulated at the hip, immediately frozen (−20°C) and defrosted at room temperature approximately 24 h prior to experimentation.

The methods employed for the capture, derivation and analysis of ovine stifle joint kinematics during passive flexion-extension has previously been described in detail.25 Briefly, hindlimbs were secured in a perspex human knee loading frame and an extensor moment applied using a weight-cable-pulley system. Stifle joint kinematics were measured using an electromagnetic (EM) tracking system (3Space Isotrack II, Polhemus, Colchester, VT) consisting of a transmitter and two receivers, which were anchored to bone-screws implanted into both the anterior and anteromedial aspects of the patella and tibia, respectively. The transformation matrix describing the position of the femur was obtained by averaging two datasets from the patellar receiver whilst intermittently anchored to an additional bone-screw implanted into the lateral aspect of each fixed femur prior to, and at the completion of each test series. We have previously determined the mean translational and rotational root mean square errors within the measurement field when using this system to be 0.5 mm (range: 0.2–1.0 mm) and 1° (range 0.3–2.7°), respectively (unpublished data).

For each test condition in this experimental design stifle joints were passively taken through their full range of motion (terminal extension to full flexion to terminal extension—the cycle) a total of 3× (approximately 20 s per cycle) and results averaged. A sample size of eight was chosen based on an a priori power analysis which indicated that a 10% change in patellar flexion angle could be detected following a 10% change in LT with a power of 0.8 and alpha of 0.01. Input data was obtained from a pilot dataset.

An identical sequence of tests was then performed on each specimen. Firstly, kinematic data with the native LT was obtained (LNative). The fat pad was then excised and a custom-made PT Lengthening Device (PTLD; Fig. 1) attached. This device was machined from black acetal and assembled with nylon screws in order to eliminate it as a potential source of EM interference. The PT was then transected, transferring the tensile load to the PTLD and data (denoted as L0) obtained. Kinematic and LT changes were referenced to this control pattern. The medial and lateral peripatellar retinaculi were then partly transected (extending approximately 1.5 cm from the PT margin) to ensure that effective LT could be increased and the patella displaced proximally without resistance or tension generated in the adjacent soft tissues. Incremental 2 mm increases in LT up to a maximum level of 12 mm (L0 + 12 mm) were then induced, with the exact amount of lengthening controlled by measuring the inner distance between the two clamps with calibrated digital calipers.

A maximum lengthening of 12 mm required to produce a 20% increase was calculated from radiographic data obtained from stifles of similar age and weight processed in our laboratory for other unrelated experiments. Incremental 2 mm increases were incorporated to provide the necessary resolution to detect statistically significant changes in patellar kinematics (if any) at LT increases of less than 20%. The precise percentage increase for each incremental lengthening step was calculated using the instantaneous LT determined from the kinematic data as described below.

Three-dimensional kinematics were determined from the raw data as previously described.25, 30 CT-scans of disarticulated bones and associated registration blocks were segmented in a slice-by-slice manner and geometric representations generated (Amira, Visualise Software, Berlin, Germany). These were imported into ProEngineer (PTC, Needham, MA) where local body-fixed coordinate systems were established and bones digitized. The femoral-fixed coordinate system origin was defined as the intersection of three surfaces: a cylinder of best-fit applied to the posterior femoral condyles, a plane perpendicular to- and bisecting a line connecting the most posterior points of the femoral condyles and the distal cortical surface of the trochlear groove (Fig. 2a). The longitudinal (z) axis was defined passing through the volumetric centroid of the femoral head. The anterior (y) axis was perpendicular to a frontal plane passing through the z-axis and parallel to the cylinder of best-fit axis; the right femoral x-axis pointed laterally. For the tibia, the origin was coincident with the center of the tibial spines. A longitudinal (z) axis was created, with the anterior axis coincident with a plane passing through the z-axis and coincident with the tibial tuberosity; the right tibial lateral (x) axis was directed laterally. Finally, the origin of the body-fixed coordinate system was coincident with the longitudinal patellar z-axis and a plane perpendicular to this axis but passing through the volumetric centroid. Anterior (y) and lateral (x) axes were established based on the retropatellar geometry. A depiction of the orientation and position of the anatomically based coordinate systems established in the femur, tibia and patella is given in Figure 2.

Figure 2.

Establishment of anatomically based coordinate systems in the femur (a), patella (b), and tibia (c). A cylinder of best fit (blue surface) was applied to the posterior femoral condyles. [Color figure can be seen in the online version of this article, available at http://wileyonlinelibrary.com/journal/jor]

Using a custom-written script for Matlab 2009a (Mathworks, Natick, IL) raw kinematic data was transformed into clinically identifiable tibial translations and rotations as per the joint coordinate system (JCS) of Grood and Suntay,35 which was also adapted to model patellar kinematics (Fig. 3). Patellar rotations included flexion (about a laterally directed axis in the femur), tilt (about the longitudinal patellar axis) and spin (about a floating axis). Patellar translations were defined as medial–lateral (ML) shift (along a laterally directed axis fixed in the femur), anterior–posterior (AP) drawer (along an anteriorly directed axis in the patella) and proximal–distal (PD) shift (along the patellar longitudinal axis).

Figure 3.

Application of the joint coordinate system to the modeling of patellar kinematics. Rotational and translational axes are indicated. The components of translation, S1–3, are the projections of the vector, H, extending from the origins of the tibial to the patellar coordinate system, along each axis.

LT was defined as the linear distance from the inferior patellar pole to the tibial tuberosity, and was assumed to be a straight line throughout passively induced flexion-extension. The percentage increase in LT for each 2 mm incremental increase was calculated by dividing LT by L0 at 120°, which was the tibial flexion at which the lowest variability in LT was observed in a pilot dataset.

Pooled PFtib, TF and PFpat kinematic data for each degree of freedom (DOF) was analyzed as a function of LT increase (LT + 1 mm to 12 mm) using general linear regression models with contrasts, which accounted for nested intrasubject variation, repeated measures and flexion angle (either tibial or patellar). TF and PFtib kinematics were analyzed at 5° iterations of tibial flexion whilst PFpat kinematic data was analyzed at 2.5° iterations of patellar flexion to ensure that the measurement resolution was approximately equal. Correlations between independent variables, both between and within test groups were quantified using Pearson's correlation coefficient, R. All statistical analyses were performed using PASW Statistics 18 (SPSS Inc., Chicago, IL) and differences considered to be significant where p < 0.01.

RESULTS

For the intact state (LNative) a strong and significant correlation between tibial flexion and patellar flexion angle, measured relative to the femoral coordinate frame, was observed (Fig. 3). The polynomial regression equation obtained from pooled data relating patellar (y) to tibial (x) flexion was y = −0.0029x2 + 1.3237x − 34.324 (R = 0.972, p < 0.001). A linear regression analysis yielded y = 0.755x − 8.008 (R = 0.965, p < 0.01). Furthermore, the degree of this correlation remained unaffected by the simulated LT increase (R = 0.897–0.965, p < 0.01). Progressive PT lengthening was associated with a negative shift of terminal patellar extension and flexion values, whereas terminal tibial values remained unaffected. As such, common ranges of tibial and patellar flexion across all lengthening levels and all samples was 65–135° and 55–90°, respectively. L0 across all stifles was 54.2 ± 1.7 mm (mean ± SD), and the mean difference between LNative and L0 was 0.3 ± 0.1 mm.

Across all samples the 2, 4, 6, 8, 10, and 12 mm simulations equated to 3 ± 0%, 6 ± 2%, 9 ± 1%, 12 ± 1%, 16 ± 1%, and 20 ± 1% increases in LT (mean ± SD). Combined PFtib kinematics for all stifle joints as a function of the simulated length increases is presented in Figure 4, where it is apparent that incremental LT increases were associated with offsetting of control (L0) tracking patterns of all kinematic parameters by varying degrees. Results from the linear regression model applied to this data are presented in Table 1. Significant changes in patellar flexion, tilt, AP drawer and PD shift were encountered for LT equal to L0 + 2 mm (p < 0.01). All PFtib kinematic parameters were significantly different from the control tracking pattern for all LT equal to or greater than L0 + 6 mm (mean 9% increase in LT).

Figure 4.

Patellar kinematics as a function of tibial flexion (PFtib) at simulated values of patellar tendon length. Left column; rotations. Right column; translations (mean ± SD). Note: To improve clarity error bars have been selectively removed from some series.

Table 1. Linear Regression of PFtib Kinematics as a Function of Simulated Increases in LT
  Dependent variable
Rotations (°)Translations (cm)
FlexionSpinTiltML shiftAP drawerPD shift
  1. Linear contrasts estimate the effect of patellar tendon lengthening for each DOF of patellar position when pooled over the range of tibial flexion angles from 65° to 135°. The contrast estimates, significance and 99% CI values are indicated. Significant values are indicated in bold.

 Contrast estimate−2.60.21.50.070.11−0.11
L0+ 2 mmp-value0.0000.2200.0000.0110.0000.000
 99% CI−4.0 to −1.1−0.3 to 0.70.7 to 2.4−0.01 to 0.140.05 to 0.17−0.13 to −0.09
 Contrast estimate−6.30.43.10.140.28−0.12
L0+ 4 mmp-value0.0000.040.0000.0000.0000.102
 99% CI−7.8 to −4.8−0.1 to 0.92.3 to 4.00.07 to 0.210.22 to 0.34−0.31 to 0.07
 Contrast estimate−8.10.74.10.170.34−0.21
L0 + 6 mmp-value0.0000.0010.0000.0000.0000.000
 99% CI−9.5 to −6.60.2 to 1.23.3 to 5.00.10 to 0.240.28 to 0.40−0.23 to −0.19
 Contrast estimate−9.91.25.00.190.43−0.31
L0 + 8 mmp-value0.0000.0000.0000.0000.0000.000
 99% CI−11.4 to −8.40.7 to 1.74.1 to 5.80.12 to 0.260.37 to 0.50−0.34 to −0.30
 Contrast estimate−11.71.75.90.220.50−0.36
L0 + 10 mmp-value0.0000.0000.0000.0000.0000.000
 99% CI−13.2 to −10.21.2 to 2.25.1 to 6.80.15 to 0.290.44 to 0.56−0.39 to −0.34
 Contrast estimate−13.52.47.00.230.58−0.42
L0 + 12 mmp-value0.0000.0000.0000.0000.0000.000
 99% CI−15.0 to −12.11.9 to 3.06.2 to 7.80.16 to 0.300.52 to 0.64−0.44 to −0.40

Combined PFpat kinematics for all stifles as a function of the simulated length increases are displayed in Figure 5. Results from the linear regression model applied to the data are presented in Table 2. Increases in LT up to 20% had no effect on patellar spin, ML shift or PD shift (p > 0.01). The change in patellar tilt and AP drawer at a 20% increase in LT was within the accuracy of the measurement system employed (0.5 mm and 1°) and not significant.

Figure 5.

Patellar kinematics as a function of patellar flexion (PFpat) at simulated values of patellar tendon length. Left column; rotations. Right column; translations (mean ± SD). Note: To improve clarity error bars have been selectively removed from some series.

Table 2. Linear Regression of PFpat Kinematics as a Function of Simulated Increases in LT
  Dependent variable
Rotations (°)Translations (cm)
SpinTiltML shiftAP drawerPD shift
  1. Linear contrasts estimate the effect of patellar tendon lengthening for each DOF of patellar position when pooled over the range of patellar flexion angles from 55° to 90°. The contrast estimates, significance and 99% CI values are indicated. Significant values are indicated in bold. Significant changes whose contrast estimate values were within the experimental measurement error are denoted as ns (not significant).

 Contrast estimate−0.1−0.00.01−0.000.01
L0 + 2 mmp-value0.7650.9180.4470.9730.731
 99% CI−0.6 to 0.5−0.4 to 0.4−0.04 to 0.07−0.04 to 0.04−0.09 to 0.12
 Contrast estimate−0.4−0.50.010.030.06
L0 + 4 mmp-value0.076ns0.5290.0420.127
 99% CI−0.9 to 0.2−0.9 to −0.1−0.0 to 0.1−0.0 to 0.1−0.0 to 0.2
 Contrast estimate−0.3−0.60.030.020.03
L0 + 6 mmp-value0.121ns0.1160.1640.408
 99% CI−0.9 to 0.2−0.9 to −0.2−0.02 to 0.08−0.02 to 0.07−0.07 to 0.14
 Contrast estimate−0.3−0.80.020.050.09
L0 + 8 mmp-value0.229ns0.360ns0.025
 99% CI−0.8 to 0.3−1.2 to −0.5−0.03 to 0.070.00 to 0.10−0.01 to 0.20
 Contrast estimate−0.3−0.80.020.050.08
L0 + 10 mmp-value0.148ns0.270ns0.047
 99% CI−0.9 to 0.2−1.2 to −0.5−0.03 to 0.070.01 to 0.10−0.02 to 0.18
 Contrast estimate−0.3−1.00.030.060.09
L0 + 12 mmp-value0.118ns0.1980.0010.023
 99% CI−0.9 to 0.2−1.4 to −0.7−0.03 to 0.080.01 to 0.10−0.01 to 0.19

Despite maximum LT increases ranging from 17% to 22% no significant changes were induced in tibial kinematics (p > 0.8). Mean intra- and intersubject variability for patellar tracking and tibial kinematics are presented in Table 3. Intersubject variability was greater than intrasubject variability for all patellar and tibial kinematic parameters.

Table 3. Average Range of Motion and Intra- and Intersubject Variability for the 6 DOF In Vitro Patellar and Tibial Stifle Joint Kinematics
 MaxMinRangeIntrasubjectIntersubject
  1. The percentage range of motion, which the respective variabilities represent are presented in parentheses.

Patella
 Rotations (°)
  Spin3.7−18.121.70.7 (3%)2.3 (11%)
  Tilt19.3−32.051.20.4 (1%)5.0 (10%)
 Translations (cm)
  ML1.4−0.41.80.1 (3%)0.1 (7%)
  AP2.50.61.9<0.1 (2%)0.2 (10%)
  PD−0.3−3.63.30.1 (3%)0.2 (7%)
Tibia
 Rotations (°)
  Abd/add16.4−8.825.20.5 (2%)1.5 (6%)
  Int-ext6.5−19.325.81.1 (4%)2.4 (10%)
 Translations (cm)
  ML0.0−0.40.4<0.1 (6%)0.1 (18%)
  AP2.2−1.13.20.1 (2%)0.3 (10%)
  PD2.31.11.20.1 (4%)0.2 (15%)

DISCUSSION

Whilst maximum LT increases of 12 mm were ultimately simulated to produce a 20% increase in LT as per the experimental hypothesis our results demonstrated that all PFtib kinematic parameters were significantly (p < 0.01) affected beyond a much lower mean lengthening level of 9% (L0 + 6 mm). Furthermore, patellar flexion, tilt, AP drawer and PD shift were significantly (p < 0.01) altered at an even lower mean 3% increase (L0 + 2 mm). Conversely, TF kinematics remained unaffected by the simulated increases in LT (p > 0.8). Kinematics were highly reproducible for the same subject with variability less than 1.1° and 1 mm. Intersubject variability was marginally higher at less than 5° and 3 mm.

In this, and in our previous work,25 a linear correlation between patellar and tibial flexion angles relative to the femoral coordinate frame during passive flexion-extension was identified. A similar relationship is also evident in the human knee.10, 17, 23, 24 The current study demonstrates that this relationship is LT-dependent, such that PT lengthening causes a decrease in patellar flexion for a given tibial flexion angle. A similar degree of correlation between patellar and tibial flexion angles persisted at increasing values of LT (R = 0.897–0.965, p < 0.01). These results suggest that each finite position of the patella along its articulation with the femoral trochlea is associated with a unique amount of flexion rotation about the laterally directed femoral axis. Measurement of patellar flexion about this axis in the human knee is associated with minimal variation, as compared to the alternative measurement of patellar flexion about a body-fixed patellar axis.36

Since patellar flexion was closely correlated with tibial flexion, even after tendon lengthening, we sought to understand the impact of PT lengthening on the position and orientation of the patella in the remaining five degrees of freedom at a given patellar flexion angle by defining PFpat motion. We found that changes in PFpat kinematics following PT lengthening up to 20% were within the resolution of our measurement system and, therefore, not significant. Our results suggest that patellar motion within the trochlear groove follows a repeatable 3D path and that PT lengthening shifts the position of the patella along this path.

Recently, Sheehan et al.10 compared the in vivo knee kinematics of patients presenting with PF pain to that of asymptomatic controls, where LT between groups differed by 5.8 mm. The PF pain group (mean 25% increase in LT) was deemed to be mal-tracking based on the detection of significant differences in PFtib parameters including PD and ML shifts, as well as patellar flexion-extension in the measurement range (full extension to 35° tibial flexion). Comparable changes in patellar shifts were detected beyond a mean 9% increase in LT in the current study, but which were not evidence of an altered patellar tracking pattern, as demonstrated by the lack of significant changes in PFpat kinematics at this lengthening level.

Contrary to our finding of a decrease in patellar flexion accompanying simulated patella alta, Sheehan et al.10 reported an increase in patellar flexion from extension to 35° knee flexion in the patella alta group. This and other subtle differences between the two studies may be attributed to differences in species and anatomy, definition and description of coordinate and measurement systems (i.e., alternative rotation about patella-fixed axes), as well as the range of knee flexion over which patellar motion was examined. In the human knee, the trochlear groove is a major determinant of patellar motion and stability from 30° to 105° of tibial flexion where fully engaged,37 whereas the patella is engaged throughout virtually the full range of tibial flexion in the ovine stifle joint. These combined factors may account for variation in shift of the pattern of patellar flexion accompanying patella alta in early tibial flexion between the two studies.

In their study, Sheehan et al.10 also examined patellar kinematics in a subset of patients exhibiting severe patella alta and diagnosed with Ehlers Danlos Syndrome, a connective tissue disorder characterized by excessive joint laxity. Values of patellar tilt and shift not reached by the control group were encountered, and thus it would be impossible to assume that a change in the reference angle to patellar flexion would eliminate those differences.

Various investigators have demonstrated that patella alta is associated with decreased contact area and increased PF stress at a given tibial flexion angle.18, 21, 22, 38 Degradation of articular cartilage is thought to result from changes (increases or decreases) in PF joint contact stress, which is affected by contact area and magnitude of the resultant force. The PF joint reaction force varies also as a function of tibial flexion angle. Although not investigated here, our results suggest that since the 3D path of the patella is not altered in patella alta, the joint contact area for a given patellar flexion angle may also be unchanged. Of greater consequence may be that for a given patellar position in the knee with patella alta, the tibial joint flexion angle is modified, and therefore the joint reaction force at that patellar position is likely altered. Changes in joint mechanics would result in either increases or decreases in PF joint contact stress, and may play a role in the development of PF pain and cartilage degeneration.

PF pain syndrome is a complex, multi-factorial problem which often persists in the absence of patella alta.15, 16 Wilson et al.17 recently showed that PF pain syndrome was associated with altered PFtib kinematics and not a proximally displaced patella.

Our results clearly reflect the characteristic nature of the articulation of the patella within the trochlear groove in the ovine stifle joint. Unlike in the human knee where a curve passing through the deepest points of the trochlear groove lies in the sagittal plane,26 a similar curve for the ovine trochlea does exist but is positioned in a more valgus orientation. Accordingly, the patella tracks in a distinctive pattern from lateral to medial with stifle flexion. On this basis, it can be appreciated that with progressive LT increases a net lateral patellar shift would occur. Similarly, curvature of the trochlear groove projected to the sagittal plane dictates that increasing LT be associated with a net anterior patellar shift. Both these effects were observed in PFtib kinematics.

A certain limitation in the current study was the lack of loading of the medial and/or lateral quadriceps components. It is well known that patellar motion in the human knee is dependent on the shape of the bony components, the force vectors from the quadriceps and passive soft tissue constraints. Quadriceps components produce significant off-axis moments on the patella, producing motions such as patellar spin in the frontal plane.39 Powers et al.40 have shown the importance of multi-plane loading on PFtib kinematics and retropatellar contact pressure as well as on the patellar ligament force/quadriceps force ratio.41

In this study a proximal displacement of the patella could not have been achieved without transection of the peripatellar retinaculum. Recent human studies have demonstrated that the patella may rely heavily on such passive constraints, particularly near full extension when less constrained by the femoral trochlea.

This study has several further limitations. Firstly, the study of passive, unloaded in vitro motion of the knee may not be representative of the motion that occurs in the loaded, in vivo knee. Knees were passively flexed in an open kinematic chain setup by the same person in all tests. Closed kinematic chain testing may have generated different results. The range of tibial flexion common to all stifle joints in the current study was 65–135°, which differs from the full extension values of 35–55° reported in other studies. This is likely to be attributed to differences in condition (in vivo and in vitro), the definition of the femoral and tibial coordinate frames, as well as the nature and magnitude of the applied loading. Whilst anatomy and biomechanics of the ovine stifle joint are analogous with that of the human knee, the results obtained in this study are not directly transferable and similar trends may not be produced. Finally, the mid-substance of the PT has been substituted with an element, which does not possess the same viscoelastic material properties as the native tissue and the effect(s) that this may have had on the results remains unknown.

Patellar tracking is the motion path of the patella along its articulation with the femoral trochlea,42 and can either be expressed relative to the femoral trochlea, or a combination of body-fixed and floating axes.36 Studies of the prior are essentially limited to static studies of a radiographic nature. For the latter kinematic studies, a convention in the literature is the expression of patellar motion as a function of tibial flexion angle, that is, PFtib.42, 43 PFtib kinematics in this study were particularly sensitive indeed to the simulated increases in LT, due to the modification of the patellar flexion angle at a given tibial flexion angle.

Patella alta in the human knee has been associated with alterations in both retropatellar contact mechanics and PFtib kinematics, and which are believed to initiate a cascade of events leading eventually to cartilage degeneration and PF pain. Our study has demonstrated that motion of the patella follows a repeatable path, which is independent of mean simulated LT increases of up to 20%.