Intraoperative vs. weightbearing patellar kinematics in total knee arthroplasty: A cadaveric study

Available online at www.sciencedirect.com

Clinical Biomechanics 23 (2008) 60–70 www.elsevier.com/locate/clinbiomech

Intraoperative vs. weightbearing patellar kinematics in total knee arthroplasty: A cadaveric study C. Anglin

a,b,*

, J.M. Brimacombe a, D.R. Wilson c,d, B.A. Masri d, N.V. Greidanus d, J. Tonetti e, A.J. Hodgson a a

Department of Mechanical Engineering, University of British Columbia, Vancouver, Canada Centre for Bioengineering Research and Education, University of Calgary, Calgary, Canada Division of Orthopaedic Engineering Research, University of British Columbia and Vancouver Coastal Health Research Institute, Vancouver, Canada d Department of Orthopaedics, University of British Columbia, Vancouver, Canada e Orthopaedic and Trauma Department, Hoˆpital Michallon, Grenoble, France b

c

Received 2 April 2007; accepted 8 August 2007

Abstract Background. During knee replacement surgery, surgeons optimize intraoperative patellar tracking with the aim of optimizing postoperative tracking. This link has not been investigated to date. Our research questions were: (1) How well do patellar kinematics correlate between passive and weightbearing flexion across numerous changes in component placement? (2) How do the kinematics differ between the two loading configurations? Methods. Eight cadaveric knee joints with modified knee components that allowed 11 different femoral, tibial and patellar placements were tested in two experimental rigs simulating intraoperative and weightbearing dynamic flexion. Baseline placement had all components in neutral position. Pearson correlation coefficients were calculated for absolute baseline kinematics and for relative kinematics due to changes in component position (i.e., the 10 altered positions vs. baseline). Findings. Correlations between intraoperative and weightbearing rigs for absolute baseline kinematics were unpredictable, ranging from poor to excellent (mean 0.56 for tilt and mean 0.50 for shift). Correlations between rigs for changes in tilt and shift, i.e. relative kinematics, were strong (>0.8) or very strong (>0.9), with the exception of shift in early flexion (0.54). Differences in relative kinematics, which averaged 2.2° in tilt (standard deviation 1.8°) and 1.6 mm in shift (standard deviation 1.7 mm), were notably smaller and less variable than differences in absolute kinematics, which averaged 4.2° in tilt (standard deviation 3.6°) and 4.3 mm in shift (standard deviation 3.9 mm). Interpretation. The results of this study suggest that, while absolute kinematics may differ between conditions, if a surgeon adjusts a component position to improve patellar kinematics intraoperatively, the effects of such a geometric change will likely carry through to the postoperative joint. Ó 2007 Elsevier Ltd. All rights reserved. Keywords: Total knee replacement; Patella; Kinematics; Knee mechanics; Computer-assisted surgery

1. Introduction

* Corresponding author. Address: Dept. of Civil Engineering, University of Calgary, 2500 University Drive NW, Calgary, AB, Canada, T2N 1N4. E-mail address: [email protected] (C. Anglin).

0268-0033/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.clinbiomech.2007.08.005

During knee replacement surgery, surgeons aim to optimize knee component placement for postoperative joint function. One of the major goals is to achieve central, balanced patellar tracking within the femoral groove. However, surgeons can only judge tracking intraoperatively by observing the patella during passive knee flexion. Since

C. Anglin et al. / Clinical Biomechanics 23 (2008) 60–70

the leg motion during this passive range-of-motion test is controlled by the surgeon rather than the patient, the quadriceps force is limited to the passive tension in the muscles. By contrast, during postoperative weightbearing activities such as walking, climbing stairs or squatting, the quadriceps are required to actively extend the knee or actively resist a flexion moment, leading to substantially higher muscle and joint contact loads. Many researchers have simulated weightbearing in cadaveric specimens to investigate patellar kinematics after total knee arthroplasty (Anouchi et al., 1993; Armstrong et al., 2003; Chew et al., 1997; Ezzet et al., 2001; Jenny et al., 2002; Hsu et al., 1996; Miller et al., 2001; Nagamine et al., 1994; Nelissen et al., 1995; Rhoads et al., 1993, 1990; Tanzer et al., 2001; Yoshii et al., 1992). To our knowledge, however, intraoperative passive joint flexion has not been compared to postoperative weightbearing flexion after total knee arthroplasty either in vivo or ex vivo, especially with regard to the effects of component placement on patellar kinematics. Understanding the link between intraoperative and postoperative kinematics is particularly important for computer-assisted surgery, where measurements are made and actions taken intraoperatively with the intention of improving the postoperative outcome. There are two possible hypotheses regarding the effects of component changes on patellar kinematics: one is that the effects will be similar between the intraoperative and weightbearing conditions due to identical joint geometry; the other is that the effects will be different due to the different loading conditions. Our research questions were: (1) How well do patellar kinematics (tilt and shift) correlate between simulated intraoperative flexion and postoperative weightbearing flexion across numerous changes in component placement? (2) How do the kinematics differ between the two loading configurations? 2. Methods We tested eight cadaveric knee joints, post-arthroplasty, in two different rigs simulating intraoperative and weightbearing flexion, while recording the femoral, tibial and patellar kinematics. 2.1. Specimens and knee components The eight fresh-frozen knee joints (mid-femur to midtibia) included six female and two male specimens, ages 51–80. All specimens were anatomically normal, with no varus or valgus deformities. Our institutional review board approved the study. The specimens were implanted with custom-modified NexGen Legacy posterior-stabilized (LPS) components (Zimmer, Warsaw, USA) by the same surgeon, using a medial parapatellar approach. The allpolyethylene patellar component had a central-dome axisymmetric design that matched the shape of the trochlear groove in mid to deep flexion. Ligament balancing (i.e. reasonable and balanced ligament laxity in both flexion and

61

extension), was performed with the femoral and tibial components in their neutral positions. We modified the components and the bone to enable us to vary component placement in five ways (as shown in Figs. 1–3): femoral component rotation (±5°), tibial component rotation (±5°), patellar resection angle (±7.5° and 15° lateral), patellar component medialization (2.5 and 5.0 mm), and increased patellar thickness (+3 mm), for a total of 10 clinically representative variations (Brimacombe, 2006) relative to the baseline setting where all components were in their neutral positions (0°, 0 mm). These component variables are all under the surgeon’s direct control to potentially influence patellofemoral kinematics. 2.2. Design of modified components Key features of the components are described below. A more extensive description can be found elsewhere (Brimacombe, 2006). Femoral component axial rotation (Fig. 1a and b): A long threaded rod was welded onto a 6° valgus block, which was pinned securely to the femoral component box. A 6° valgus cut was made distally on each specimen, and sufficient bone was removed anteriorly and posteriorly to allow at least ±5° rotation. During the experiment, the femoral rod was inserted through a long tube that was cemented into the intramedullary canal of each specimen. The rod was accessible above the resected femur. A nut was threaded onto the rod and used to fix the femur in the desired orientation, with the tube acting as a long washer pushing against the component as with the design used by Armstrong et al. (2003). A lockwasher was used for extra connection strength. Metallic wedges were placed against the posterior bone to fill the gap

Femoral rod of modified femoral component; attaches to testing rig Pointer, attached to rod Nut Lockwasher Protractor, attached to tube

Femoral bone

Femoral tube cemented into intramedullary canal

Block providing 6º valgus angle Modified femoral component

Fig. 1a. Schematic assembly of the adjustable femoral component.

62

C. Anglin et al. / Clinical Biomechanics 23 (2008) 60–70

Fig. 1b. Adjustable femoral component ( 5°, 0°, +5° axial rotation): (1) original femoral component, (2) 6° valgus block fixed to the component, (3) rod passing through the femoral tube, (4) femur, (5) tube cemented into the femoral bone (acting like a long washer), (6) protractor fixed to the tube, (7) pointer fixed to the threaded rod, and (8) nut and lockwasher.

Fig. 2a. Adjustable tibial component ( 5°, 0°, +5° axial rotation): (1) baseplate, (2) sets of screw-holes for rotations, (3) screw-holes to attach the baseplate to the bone, (4) screw-hole to attach the insert to the bone, (5) tibial tray, (6) tibial insert, and (7) location of the neutral or offset washer to accommodate the changing position of the screw with rotation.

Washer Tibial spacer Tibial tray

3-part tibial construct

Baseplate Bone (tibial plateau)

Fig. 2b. Schematic assembly of the adjustable tibial component.

on the opposing side created by the rotation, thereby securing the position. Angles were accurate to within 0.5° by reading the location of a pointer (fixed to the rod) relative to a protractor (fixed to the tube). Tibial component axial rotation (Fig. 2a and b): Tibial rotation was achieved to within 0.1° using a customdesigned 3 mm baseplate with two sets of three screw holes (medial and lateral) corresponding to 0°, +5° and 5°. An extra 3 mm was resected from the bone to accommodate the baseplate. The baseplate itself was attached to the bone using two 4 cm long screws. A trial insert was screwed into the bone through the tibial tray and baseplate using a 5 cm long screw; washers machined with offset holes accommo-

dated the rotation angles. A trial insert was used rather than a snap-in insert to permit repeated access to the baseplate screws to change the rotation angle. Patellar baseplate (Fig. 3): All patellar modifications were facilitated by using a hexagonal baseplate on the patellar bone, aligned in the mediolateral direction. The 2.5 mm thick baseplate was securely attached to the bone by inserting six pins through the baseplate at 30° to the bone surface. Modified patellar component (Fig. 3a): The bottom 2.5 mm cylindrical portion of the patellar component was removed such that the height of the total construct (baseplate + component) was the same as the original component. Since this portion of the component was cylindrical, the modification had no effect on the component diameter. Patellar resection angle (Fig. 3b): To simulate the asymmetric patellar resection angles, we added either a neutral disk or a wedge of the designated angle, ±7.5° or 15°, with machining tolerances of 0.1°. The disk and all wedges had the same central thickness (4.5 mm), which was removed from the patellar bone to maintain the original patellar thickness. ‘Medial’ wedges were thicker on the medial side, corresponding to underresection of the medial patellar

C. Anglin et al. / Clinical Biomechanics 23 (2008) 60–70

bone. The disk or wedges were attached to the baseplate via metal pegs that fit into machined holes in the baseplate. An embedded magnet in the baseplate resisted separation of the components. Patellar medialization (Fig. 3c): Patellar positioning was achieved to within 0.1 mm by inserting the metal pegs into different sets of holes on the baseplate, at 0.0, 2.5 and 5.0 mm. Patellar thickness (Fig. 3d): A 3 mm thick disk, with protruding and receiving pegs, was added between the neutral disk and the baseplate to create the additional height. 2.3. Definitions of neutral placement (baseline) Neutral, baseline alignment was defined as follows: for femoral component rotation neutral was defined as 3°

63

externally rotated from the posterior condylar line because this was considered to be more reproducible than aligning the prosthesis with the transepicondylar axis (TEA); this line was within 2° of the TEA in all cases, confirming the lack of substantial condylar deformity. For tibial component rotation, neutral placement was defined by aligning the central axis of the tibial plate with the junction between the mid and medial thirds of the tibial tubercle. A neutral patellar resection angle was defined by having equal medial and lateral patellar thicknesses at one-quarter distance across the mediolateral width. Neutral patellar placement occurred when the patellar component was centred mediolaterally on the patellar bone. Neutral patellar thickness was defined as being within 1 mm of the original thickness. Each of these definitions mimics the standard surgical goal with the exception that some surgeons routinely medialize the patellar component. 2.4. Testing order We tested the component variation types in a randomized order, in order to minimize any bias from order effects. Baseline measurements, with all components in their neutral position, were taken at the beginning of each setup and between each type of component variation. The medial incision was closed following each component change using two towel clamps; the locations of the clamps were recorded using a marker to ensure reproducibility. Two flexion cycles were performed for all trials. We were unable to obtain weightbearing data for one specimen because both the tibial component and the patellar baseplate loosened due to poor bone quality and could not be resecured. Specimens were tested first with all variations in the intraoperative rig and then with all variations in the weightbearing rig. To minimize any frictional differences between the trial components used and standard implanted components, and to mimic the clinical situation more closely, we lubricated the patella, femur and tibia regularly using a Teflon-based dry film lubricant (DryTef, Walter Tool Company Inc., Norwell, MA). To prevent drying of the tissues, the specimens were sprayed routinely with water. 2.5. Intraoperative (horizontal, passive) apparatus

Fig. 3. Adjustable patellar component (0°, ±7.5°, 15° resection / 0, 2.5, 5.0 mm medialization / 0, +3 mm thickness): (a) Patellar construct with (1) baseplate attached securely to the bone via 6 pins, (2) neutral disk; the central thickness was identical for all resection angles (the equivalent thickness was removed from the bone), and (3) patellar component with the bottom portion removed to accommodate the baseplate thickness. Example positions: (a) Baseline position and orientation, (b) 15° asymmetric resection, (c) 5 mm component medialization, and (d) 3 mm additional thickness.

Our custom-designed ‘intraoperative’ rig (Fig. 4) simulated the range-of-motion test performed by surgeons to evaluate patellar tracking once all trial components are in place at the end of the surgery. During in vivo surgeries, the surgeon moves the leg from full extension to full flexion and back again while watching the tilt and shift of the patella. Our closed-chain experimental apparatus simulated this assessment, using mechanical linkages for the ‘ankle joint’ and ‘hip joint’. The ankle joint had all three rotational degrees of freedom; the hip joint had two rotational degrees of freedom (flexion and internal/external rotation) and one translational degree of freedom (horizontal). This configuration prevented the leg from fall-

64

C. Anglin et al. / Clinical Biomechanics 23 (2008) 60–70

Fig. 4. ‘Intraoperative’ passive-flexion rig. (1) Hip joint, (2) ankle joint, (3) threaded rod, (4) rail, (5) quadriceps clamp, (6) spring, (7) marker array.

ing sideways, while maintaining the full six degrees of freedom of the knee joint itself such that the kinematics were controlled solely by the structures of the knee, not by the rig. This retention of the joint’s six degrees of freedom was demonstrated in a kinematic analysis of the vertical closed-chain Oxford rig with the same number of degrees of freedom for the hip and ankle joints (Zavatsky, 1997). A custom-designed clamp was firmly attached to the quadriceps tendon and then connected via a spring to a point distal to the ‘hip’ joint. The spring was intended to mimic the passive resistance of the quadriceps tendon and to ensure that the patella remained in contact with the femur. From extension to full flexion, the spring force increased from 11 N to 90 N with an extension of 7.0 cm, corresponding to a spring stiffness of 11.3 N/cm. 2.6. Weightbearing (vertical, active) apparatus Our custom-built Oxford-type weightbearing rig (Fig. 5) actively loaded the knee by loading the quadriceps tendon during continuous flexion and extension to simulate postoperative deep knee bends (squats) or stair climbing. Without the resistance of the quadriceps muscle, the knee would flex under the hip load (54 N). The simulated ankle joint allowed rotations about three orthogonal axes and no translations. The simulated hip joint allowed vertical translation, flexion and ab/adduction resulting in two rotational and one translational degree of freedom. Therefore, as with the intraoperative rig, the knee retained its full six degrees of freedom, constrained only by its soft-tissue envelope. 2.7. Kinematic measurements To determine the patellofemoral tracking characteristics throughout flexion and extension, we firmly attached marker arrays, each with four infrared-emitting diodes

Fig. 5. Weightbearing loaded-flexion rig. (1) Hip joint, (2) ankle joint, (3) threaded rod, (4) rail, (5) quadriceps clamp, (6) motor, pulley, cable, (7) marker array.

(IREDs), to the femur, tibia and patella using K-wires. The three-dimensional position and orientation of each bone were tracked with an Optotrak 3020 optoelectronic camera system (Northern Digital, Waterloo, Canada). The root-mean-square (RMS) accuracy for each IRED was 0.1 mm in the plane of flexion and 0.15 mm perpendicular to the plane of flexion (Northern Digital Inc., 2006). The weight of the patellar tracker and associated cable

C. Anglin et al. / Clinical Biomechanics 23 (2008) 60–70

We extracted data at 15° increments, resulting in six flexion angles between 15° and 90° in both flexion and extension; only the extension data will be presented for graphical clarity. Absolute patellar tilt and shift were normalized to the average baseline value (i.e. with all components in their neutral position) at 90° flexion during the extension phase in order to facilitate comparisons between specimens and rigs. Whereas Katchburian et al. (2003) nor-

a

Relative Correlation Example VAR1 VAR9

9

VAR8 VAR6 VAR10 VAR4 BASE VAR5 VAR2 VAR3 VAR7

6 3 0 0

15

30

45

60

75

90

105

75

90

105

-3 -6

Flexion Angle (°)

b

12 9 VAR1

6

VAR9 VAR4 VAR8 VAR6 BASE VAR3 VAR7 VAR5 VAR10

3 0 0 -3

15

30

45

60

10 differences at each flexion angle

Intraoperative

Absolute Patellar Shift (mm)

12

Weightbearing

2.8. Data analysis

malized data to 0° flexion, we chose to normalize the baseline values to 90° flexion because in the raw data, the greatest deviations from neutral tilt and shift occurred in early flexion rather than late, and therefore normalizing to 90° represented patellar motion more accurately. This is explained by the geometry of the femoral component groove, which is more central and deeper distally than proximally. This normalization had no effect on the relative data or correlations, since the baseline values were subtracted from the component variable data. We tested the statistical hypothesis that there were correlations between the kinematic results of the two rigs using the Pearson product-moment correlation coefficient, r, with a = 0.05. For each specimen, absolute baseline intraoperative vs. weightbearing tilt and shift were correlated across the six extracted flexion angles (Fig. 6, Base). For each specimen and flexion angle, relative tilt and shift were correlated using the difference between each variation and the average baseline, i.e. using 10 data points (Fig. 6, VarBase). The correlations were averaged to represent the degree to which postoperative weightbearing joint kinematics relate to intraoperative kinematics for an individual patient. For illustration purposes, all of the absolute tilt and shift data are shown together (Fig. 7a and b), and all of the relative tilt and shift data are shown together

Absolute Patellar Shift (mm)

was less than 15 g; the effect of this on the kinematics would therefore be easily offset by the quadriceps load (passive or active) and the towel clamps that were used to close the medial incision. We identified four landmarks (medial, lateral, superior and inferior) in each bone with small screws. For the femur, the landmarks were the medial and lateral epicondyles plus two proximal points that formed a line parallel to the femoral shaft; for the tibia, the landmarks were on the medial and lateral sides of the tibial plateau and at two distal locations to form a line parallel to the tibial axis; for the patella, the landmarks were at the superior and inferior patellar poles and at medial and lateral points perpendicular to the superoinferior axis. Using a tracked pointer, we digitized these landmarks to define an anatomical reference frame for each bone. The origin was defined halfway between the medial and lateral landmarks of each bone. The kinematics of each bone were recorded throughout the range-of-motion, and the anatomical axes digitized before and after all tests in each rig. We calculated the kinematics using both digitizations, and then averaged the results. Patellofemoral and tibiofemoral kinematics were calculated using a custom-written Matlab program (Version 6.5; The Mathworks, Natick, MA, USA) based on the joint coordinate system (JCS) recommended by Cole et al. (1993) which supports the proposal by Grood and Suntay (1983), namely using the flexion axis of the proximal segment (the femur) as the first axis, the axial rotation axis of the distal segment (patella or tibia) as the third axis, and an axis perpendicular to both of these as the second axis. Translations were measured along these axes. In the Euler system, this corresponds to rotation about the common (i.e. aligned) patellofemoral or tibiofemoral mediolateral axis, followed by rotation about the patellar or tibial anteroposterior axis, followed by rotation about the patellar or tibial superoinferior axis. Of the six components of patellar movement relative to the femur, we selected tilt and shift as the most clinically relevant, as these are the easiest to judge intraoperatively and the easiest to measure radiographically. In accordance with the system above, patellar tilt was defined as rotation about the superoinferior axis of the patella while patellar shift was defined as translation parallel to the epicondylar (flexion) axis of the femur. Movement in the lateral direction was defined as positive (Katchburian et al., 2003).

65

VAR2

-6

Flexion Angle (°)

Fig. 6. Relative data used for the correlation calculation for a sample specimen: for a given flexion angle, the 10 differences in patellar shift relative to baseline in the intraoperative rig, (a), were correlated with the corresponding 10 differences in the weightbearing rig, (b). This was repeated for the six flexion angles at which measurements were extracted.

66

C. Anglin et al. / Clinical Biomechanics 23 (2008) 60–70

(Fig. 7c and d). Fig. 7e and f represents how the reported correlations were calculated, i.e. by determining the correlation for each specimen at the given flexion angle, then calculating the mean and standard deviation (SD) of these correlations. The absolute and relative differences in tilt and shift between the two rigs across all flexion angles and component placement variables were also calculated to describe the magnitude of the differences in addition to the correlation.

3. Results In both rigs, the patella tilted and shifted laterally in most specimens in early flexion, and then tracked neutrally once supported by the groove (Fig. 8). Medial shift and tilt did occur in some specimens. Variability in tilt and shift between specimens was greater in the intraoperative rig than in the weightbearing rig (Figs. 7 and 8). Correlations between the rigs for absolute baseline tracking ranged from

Tilt

Shift 20

25

15

15

S1 S2 S3 S4 S5 S7 S8

10 5 0 -20

-15

- 10

-5

0

5

10

15

20

25

-5 -10

Absolute Shift (mm) WEIGHTBEARING RIG

Absolute Patellar Tilt (°) WEIGHTBEARING RIG

Absolute

20

10

5

0 - 15

-10

-5

a

-15

5 0 -10

0

10

20

-5 - 10

c

- 15

Change in Shift After Change in Component Position (mm) WEIGHTBEARING RIG

Change in Tilt After Change in Component Position (°) WEIGHTBEARING RIG

Relative

S1 S2 S3 S4 S5 S7 S8

10

10

5

0 - 15

-10

-5

5

10

15

20

d

- 10

- 15

- 20

Change in Shift After Change in Component Position (mm) INTRAOPERATIVE RIG 20

25

Example at: 45¼ Flexion

20 15 10 5 0 -5

0

5

10

15

20

-5 - 10

e

- 15 - 20

Change in Tilt After Change in Component Position (°) INTRAOPERATIVE RIG

25

Change in Shift After Change in Component Position (mm) WEIGHTBEARING RIG

Change in Tilt After Change in Component Position (°) WEIGHTBEARING RIG

0 -5

Change in Tilt After Change in Component Position (°) INTRAOPERATIVE RIG

Correlations

b

15

15

- 10

20

20

20

-15

15

Absolute Shift (mm) INTRAOPERATIVE RIG

25

- 20

10

- 15

-20

Example at: 45° Flexion

5

- 10

Absolute Patellar Tilt (°) INTRAOPERATIVE RIG

- 20

0 -5

15

10

5

0 - 15

-10

-5

0

5

10

15

20

-5

- 10

f

- 15

Change in Shift After Change in Component Position (mm) INTRAOPERATIVE RIG

Fig. 7. Scatterplots showing intraoperative vs. weightbearing kinematics for all specimens combined for: (a) absolute tilt, (b) absolute shift, (c) relative tilt and (d) relative shift. Correlation calculations were performed for individual specimens at each flexion angle; examples across specimens are shown for (e) relative tilt at 45° and (f) relative shift at 45°. Seven specimens were tested in both rigs; the eighth failed in the weightbearing rig. Correlation coefficients were averaged across specimens to represent the correlations that would be seen for an individual patient.

C. Anglin et al. / Clinical Biomechanics 23 (2008) 60–70 20

12

Lateral

Intraoperative Weightbearing

Lateral

Intraoperative Weightbearing

Patellar Shift (mm)

10

15

Patellar Tilt (°)

67

10

5

0

8 6 4 2 0 -2 -4

-5 0

15

30

45

60

75

90

105

0

15

30

Flexion angle (°)

45

60

75

90

105

Flexion angle (°)

Fig. 8. Baseline tilt and shift, with all components in their neutral position, for the intraoperative and weightbearing rigs; mean ± standard deviation. Lateral is positive.

excellent and highly significant (P < 0.001) in some specimens to very poor and non-significant in others (Fig. 9). The change in tilt or shift due to a change in component position (i.e. relative tilt or shift) was very similar between intraoperative and weightbearing flexion (see Fig. 7c and d), except for shift near extension. Correlations between the rigs for both relative tilt and relative shift were strong (>0.8) or very strong (>0.9), with the exception of shift in early flexion, which was only moderate (0.54), as shown in Fig. 10. Statistical significance was high in all cases (P < 0.001). Relative changes in kinematics were therefore better predicted between rigs than absolute tracking. Differences in absolute tilt and shift between rigs for all 10 component variations and the baseline averaged 4.2° (SD 3.6°) in tilt and 4.3 mm (SD 3.9 mm) in shift, in comparison to a total range of absolute tilt and shift of approximately 35° and 25 mm. Differences between rigs were greatest in early flexion. Relative differences (i.e. changes

in tilt and shift relative to the baseline values) averaged 2.2° (SD 1.8°) in tilt and 1.6 mm (SD 1.7 mm) in shift, in comparison to a total range of relative tilt and shift of approximately 25° and 15 mm. Repeatability between each of the two flexion cycles was high in both rigs: absolute differences in tilt between cycles averaged 0.7° (SD 1.3°) in the weightbearing rig and 0.8° (SD 1.3°) in the intraoperative rig, the greatest differences occurring at small flexion angles. Absolute differences in shift between cycles averaged 0.3 mm (SD 0.5 mm) in the weightbearing rig and 0.6 mm (SD 1.6 mm) in the intraoperative rig. The root-mean-square (RMS) differences between the 5 pairs of baseline measurements taken over the course of testing were 1.3°/2.4° for tilt and 0.9/ 2.1 mm for shift, for the weightbearing and intraoperative rigs, respectively. The specific effects of each component variable are reported elsewhere (Anglin et al., 2007). 4. Discussion

1.0

Pearson Correlation Coefficient

Baseline Tilt 0.8

0.6 Baseline Shift 0.4

0.2

0.0

-0.2

Fig. 9. Correlations between the intraoperative and weightbearing rigs for absolute baseline tilt and absolute baseline shift, mean and standard deviation across specimens.

In this study, we compared patellar kinematics in simulated intraoperative and weightbearing conditions to evaluate whether intraoperative tracking predicts weightbearing tracking and whether intraoperative changes to component placement can be used to predict the effects on patellar tilt and shift during weightbearing. The relationship between absolute intraoperative and weightbearing tracking was unpredictable; it was often excellent, but sometimes poor. By contrast, the kinematic effects of changing component positioning were well-correlated between the two rigs except for patellar shift near extension. This improved correlation in later flexion is consistent with the view that patellar kinematics are predominantly controlled by the geometry of the femoral groove, as has been found in the natural knee (Ahmed and Duncan, 2000), i.e. the path is more tightly controlled in later flexion than in early flexion. Remaining differences in patellofemoral mechanics between the two rigs are likely explained by differences in soft-tissue forces in the different loading configurations. The higher quadriceps force (and hence contact force) in

68

C. Anglin et al. / Clinical Biomechanics 23 (2008) 60–70

1.0

Pearson Correlation Coefficient (r)

0.9 0.8

Change in Tilt

0.7 0.6 0.5 Change in Shift

0.4 0.3 0.2 0.1 0.0 15

30

45

60

75

90

Flexion Angle (°)

Fig. 10. Correlations between the intraoperative and weightbearing rigs for tilt relative to baseline and shift relative to baseline, throughout extension, mean and standard deviation across specimens (see Figs. 6 and 7 for calculation procedure); P < 0.001 in all cases.

the weightbearing rig appeared to stabilize the patella, resulting in less variability in tilt and shift in the weightbearing rig. By contrast, given a lower quadriceps force in the intraoperative rig and assuming similar retinacular tensions between rigs, the contributions of the lateral and medial retinacula to tilt and shift would be greater in the intraoperative rig than the weightbearing rig. This may explain the greater range of tilt and shift observed in the intraoperative rig. The range of tilt over the flexion cycle in the weightbearing rig was similar in magnitude to many previously reported ranges (Armstrong et al., 2003; Chew et al., 1997; Hsu et al., 1996; Katchburian et al., 2003; Miller et al., 2001; Omori et al., 1997; Rhoads et al., 1990) with only one exception, in which the change was almost twice as large (Nagamine et al., 1994). The range of shift was similar in magnitude to some studies (Chew et al., 1997; Miller et al., 2001; Omori et al., 1997; Rhoads et al., 1990), while different than others (Anouchi et al., 1993; Armstrong et al., 2003; Hsu et al., 1996; Nagamine et al., 1994). The patterns of patellar tilt and shift are quite variable among studies, in some cases becoming increasingly medial during flexion, in other cases increasingly lateral. We are unaware of any previous studies investigating intraoperative patellar mechanics either ex vivo or in vivo for comparison. Since the marker arrays were attached to the femoral and patellar bones, the reported kinematics represent the tilt and shift of the patellar bone with respect to the femoral bone. Changes in the position and orientation of the patellar bone affect both the soft-tissue tensions and the Q-angle. The kinematics of the patellar component with respect to the femoral component were generally smaller, and often in the opposite direction to the bone vs. bone kinematics (Anglin et al., 2007). For example, in early flex-

ion, 5 mm patellar component medialization led to a 3 mm lateral shift of the patellar bone (thereby reducing lateral retinacular tension) combined with a 2 mm medial shift of the patellar component in the femoral groove. Our findings that the intraoperative results showed greater variability than the weightbearing results as well as a tendency towards more lateral shift are consistent with two recent studies that compared weightbearing with nonweightbearing kinematics. A clinical radiographic study (Baldini et al., 2006, 2007) found that tilt and subluxation were reduced in an axial weightbearing radiograph compared to the standard non-weightbearing Merchant’s view; in the case of asymmetric resection, the reduced subluxation often led to bony impingement of the medial facet. The same study reported that anterior knee pain correlated with the weightbearing results but not the non-weightbearing results. A magnetic-resonance imaging (MRI) study of the natural patella (Powers et al., 2003) likewise found less patellar displacement during weightbearing compared to non-weightbearing in symptomatic subjects with lateral subluxation. The authors characterized patellar kinematics during non-weightbearing as the patella rotating on the femur and during weightbearing as the femur rotating underneath the patella; this could explain some of our differences in absolute kinematics, but should have a minor impact on changes in kinematics due to changes in component placement, i.e. relative kinematics. The present study is distinct from the radiographic study in that an entire range of flexion was investigated; it is distinct from the MRI study in that the resurfaced knee was investigated. It is distinct from both in that we parametrically examined the effect of numerous changes in component placement. Differences between intraoperative and postoperative kinematics may be clinically significant in some cases. More than 5 mm of displacement or 5° of tilt from neutral

C. Anglin et al. / Clinical Biomechanics 23 (2008) 60–70

shift or tilt is generally considered abnormal (Shih et al., 2004). Given that mean differences in absolute tilt and shift between rigs (4.2° and 4.3 mm, respectively) approached this level, a patella with tilt or shift under the 5 mm/5° threshold in one situation may have tilt or shift over the threshold in the other situation. The primary limitation of the current study, as with all cadaveric studies, is that neither the intraoperative simulation nor the weightbearing simulation fully replicated physiological loading at the joint. Nevertheless, we believe that this experiment captured the fundamental differences between the two loading conditions (lower, passive, horizontal loading vs. higher, active, vertical loading) and that the conclusion regarding strong correlations between the rigs for changes in component placement would therefore remain unchanged under physiological conditions. The repeatable, parametric changes of many different component placements on the same specimen would not be possible to implement in vivo. A second limitation is that we only investigated a single component design. Nevertheless, while some studies have identified differences due to component design (Singerman et al., 1997; Yoshii et al., 1992), a study comparing three different component designs, including the NexGen, did not reveal differences among the designs, only differences in kinematics between the resurfaced knees and the intact knee (Chew et al., 1997). Now that the importance of a deeper groove for patellar tracking is well-recognized (Chew et al., 1997; Yoshii et al., 1992), most current designs, including the NexGen, include this feature. The component design used in the present study is therefore representative of most current designs. Absolute kinematics may also depend on the surgical approach (Bindelglass and Vince, 1996), the experimental protocol (Katchburian et al., 2003), and ligament balancing, and will certainly depend on the individual person, given the range of patellar tracking patterns that have been found in the natural knee in vivo (Katchburian et al., 2003; MacIntyre et al., 2006). Our primary emphasis on the effects of changes in component position minimizes the impact of these differences in absolute kinematics. 5. Conclusion The results of this study suggest that, while absolute kinematics may differ between intraoperative and weightbearing flexion, if a surgeon adjusts a component position to improve patellar kinematics intraoperatively, the effects of such a geometric change will likely carry through to the postoperative joint. Acknowledgements We wish to thank the Natural Sciences and Engineering Research Council of Canada (NSERC), Praxim (Grenoble, France), the Canadian Arthritis Network and the Michael

69

Smith Foundation for Health Research for their generous financial support, and Zimmer Canada for the donation of the components and use of the instruments. We would also like to thank Dr. Donald Garbuz for his input into the development of the experimental protocol. References Ahmed, A.M., Duncan, N.A., 2000. Correlation of patellar tracking pattern with trochlear and retropatellar surface topographies. Journal of Biomechanical Engineering 122 (6), 652–660. Anglin, C., Brimacombe, J.M., Hodgson, A.J., Masri, B.A., Greidanus, N.V., Tonetti, J., Wilson, D.R., 2007. Surgeon control of patellar tracking: effects of femoral, tibial and patellar component position. Transactions of the 53rd Orthopaedic Research Society, San Diego, February 11–14. Anouchi, Y.S., Whiteside, L.A., Kaiser, A.D., Milliano, M.T., 1993. The effects of axial rotational alignment of the femoral component on knee stability and patellar tracking in total knee arthroplasty demonstrated on autopsy specimens. Clinical Orthopaedics and Related Research 287, 170–177. Armstrong, A.D., Brien, H.J.C., Dunning, C.E., King, G.J.W., Johnson, J.A., Chess, D.G., 2003. Patellar position after total knee arthroplasty. Journal of Arthroplasty 18 (4), 458–465. Baldini, A., Anderson, J.A., Zampetti, P., Pavlov, H., Sculco, T.P., 2006. A new patellofemoral scoring system for total knee arthroplasty. Clinical Orthopaedics and Related Research 452, 150–154. Baldini, A., Anderson, J.A., Cerulli-Mariani, P., Kalyvas, J., Pavlov, H., Sculco, T.P., 2007. Patellofemoral evaluation after total knee arthroplasty. Validation of a new weight-bearing axial radiographic view. Journal of Bone and Joint Surgery–American 89 (8), 1810–1817. Bindelglass, D.F., Vince, K.G., 1996. Patellar tilt and subluxation following subvastus and parapatellar approach in total knee arthroplasty. Implication for surgical technique. Journal of Arthroplasty 11 (5), 507–511. Brimacombe, J.M., 2006. Effects of component placement on patellar kinematics and loading in intraoperative and postoperative loading configurations. M.A.Sc. thesis, University of British Columbia, Vancouver. Chew, J.T.H., Stewart, N.J., Hanssen, A.D., Zong-Ping, L., Rand, J.A., An, K.-N., 1997. Differences in patellar tracking and knee kinematics among three different total knee designs. Clinical Orthopaedics and Related Research 345, 87–98. Cole, G.K., Nigg, B.M., Ronsky, J.L., Yeadon, M.R., 1993. Application of the joint coordinate system to three-dimensional joint attitude and movement representation: a standardization proposal. Journal of Biomechanical Engineering 15, 344–349. Ezzet, K.A., Hershey, A.L., D’Lima, D.D., Irby, S.E., Kaufman, K.R., Colwell, C.W., 2001. Patellar tracking in total knee arthroplasty – Inset vs. onset design. Journal of Arthroplasty 16 (7), 838–843. Grood, E.S., Suntay, W.J., 1983. A joint coordinate system for the clinical description of three-dimensional motions: application to the knee. Journal of Biomechanical Engineering 105, 136–144. Hsu, H.-C., Zong-Ping, L., Rand, J.A., An, K.-N., 1996. Influence of patellar thickness on patellar tracking and patellofemoral contact characteristics after total knee arthroplasty. Journal of Arthroplasty 11 (1), 69–80. Jenny, J.Y., Lefebvre, Y., Vernizeau, M., Lavaste, F., Skalli, W., 2002. In vitro analysis of the continuous active patellofemoral kinematics of the normal and prosthetic knee. Revue de Chirurgie Orthope´dique et Re´paratrice de l’Appareil Moteur 88 (8), 797–802. Katchburian, M.V., Bull, A.M.J., Shih, Y.-F., Heatley, F.W., Amis, A.A., 2003. Measurement of patellar tracking: assessment and analysis of the literature. Clinical Orthopaedics and Related Research 412, 241–259. MacIntyre, N.J., Hill, N.A., Fellows, R.A., Ellis, R.E., Wilson, D.R., 2006. Patellofemoral joint kinematics in individuals with and without

70

C. Anglin et al. / Clinical Biomechanics 23 (2008) 60–70

patellofemoral pain syndrome. Journal of Bone and Joint Surgery– American 88 (12), 2596–2605. Miller, M.C., Zhang, A.X., Petrella, A.J., Berger, R.A., Rubash, H.E., 2001. The effect of component placement on knee kinetics after arthroplasty with an unconstrained prosthesis. Journal of Orthopaedic Research 19 (4), 614–620. Nagamine, R., Whiteside, L.A., White, S.E., McCarthy, D.S., 1994. Patellar tracking after total knee arthroplasty: the effect of tibial tray malrotation and articular surface configuration. Clinical Orthopaedics and Related Research 304, 263–271. Nelissen, R.G.H.H., Weidenheim, L., Mikhail, W.E.M., 1995. The influence of the position of the patellar component on tracking in total knee arthroplasty. International Orthopaedics 19, 224–228. Northern Digital Inc. (NDI), 2006. Optotrak Technical Specifications. (Accessed 24.09.06). Omori, G., Koga, Y., Bechtold, J.E., Gustilo, R.B., Nakabe, N., Sasagawa, K., Hara, T., Takahashi, H.E., 1997. Contact pressure and three-dimensional tracking of unresurfaced patella in total knee arthroplasty. The Knee 4 (1), 15–21. Powers, C.M., Ward, S.R., Fredericson, M., Guillet, M., Shellock, F.G., 2003. Patellofemoral kinematics during weight-bearing and nonweight-bearing knee extension in persons with lateral subluxation of the patella: a preliminary study. Journal of Orthopaedic and Sports Physical Therapy 33, 677–685.

Rhoads, D.D., Noble, P.C., Reuben, J.D., Mahoney, O.M., Tullos, H.S., 1990. The effect of femoral component position on patellar tracking after total knee arthroplasty. Clinical Orthopaedics and Related Research 260, 43–51. Rhoads, D.D., Noble, P.C., Reuben, J.D., Tullos, H.S., 1993. The effect of femoral component position on the kinematics of total knee arthroplasty. Clinical Orthopaedics and Related Research (286), 122–129. Shih, H.-N., Shih, L.-Y., Wong, Y.-C., Hsu, W.-W., 2004. Long-term changes of the nonresurfaced patella after total knee arthroplasty. Journal of Bone and Joint Surgery 86A (5), 935–939. Singerman, R., Pagan, H.D., Peyser, A.B., Goldberg, V.M., 1997. Effect of femoral component rotation and patellar design on patellar forces. Clinical Orthopaedics and Related Research 334, 345–353. Tanzer, M., McLean, C.A., Laxer, E., Casey, J., Ahmed, A.M., 2001. Effect of femoral component designs on the contact and tracking characteristics of the unresurfaced patella in total knee arthroplasty. Canadian Journal of Surgery 44 (2), 127–133. Yoshii, I., Whiteside, L.A., Anouchi, Y.S., 1992. The effect of patellar button placement and femoral component design on patellar tracking in total knee arthroplasty. Clinical Orthopaedics and Related Research 275, 211–219. Zavatsky, A.B., 1997. A kinematic-freedom analysis of a flexed-kneestance testing rig. Journal of Biomechanics 30 (3), 277–280.