We performed a prospective, randomised trial of 44 patients to compare the functional outcomes of a posterior-cruciate-ligament-retaining and posterior-cruciate-ligament-substituting total knee arthroplasty, and to gain a better understanding of the in vivo kinematic behaviour of both devices.
At follow-up at five years, no statistically significant differences were found in the clinical outcome measurements for either design. The prevalence of radiolucent lines and the survivorship were the same. In a subgroup of 15 knees, additional image-intensifier analysis in the horizontal and sagittal planes was performed during step-up and lunge activity. Our analysis revealed striking differences. Lunge activity showed a mean posterior displacement of both medial and lateral tibiofemoral contact areas (roll-back) which was greater and more consistent in the cruciate-substituting than in the cruciate-retaining group (medial p < 0.0001, lateral p = 0.011). The amount of posterior displacement could predict the maximum flexion which could be achieved (p = 0.018). Forward displacement of the tibiofemoral contact area in flexion during stair activity was seen more in the cruciate-retaining than in the cruciate-substituting group. This was attributed mainly to insufficiency of the posterior cruciate ligament and partially to that of the anterior cruciate ligament. We concluded that, despite similar clinical outcomes, there are significant kinematic differences between cruciate-retaining and cruciate-substituting arthroplasties.
The role of the posterior cruciate ligament (PCL) in total knee arthroplasty (TKA) has been widely discussed in the orthopaedic literature. Theoretical and functional arguments1–,4 and survivorship5–,7 have all been used to support both its retention and its substitution. Other studies have focused on the degree of deformity4,8 or the underlying cause of arthritis,9 as indications for assessing the PCL and for using a posterior-stabilised design. More elaborate algorithms based upon pathological criteria and derived from retrospective reviews have been described by Lombardi et al10 and Straw et al.11
Recently, kinematic information from radio-stereometry,12 robotic in vitro models13,14 and in vivo fluoroscopic analyses has become available and added to the debate. Those in vivo studies which have used fluoroscopy to investigate knee kinematics after TKA report abnormal kinematics when compared with the normal knee.15–,32 The differences include less posterolateral femoral roll-back as the knee moves from full extension to flexion, abnormal axial rotation between the femur and the tibia throughout the range of movement (ROM), a different centre of rotation of the knee in the horizontal plane, and condylar lift-off. Some authors have specifically compared cruciate-retaining and cruciate-substituting designs.16,23,29,31 The results for the cruciate-retaining arthroplasties are more variable than for the cruciate-substituting arthroplasties. Unphysiological roll-forward of the medial femoral condyle during flexion is a common finding, while rollback on the lateral side during flexion is more common with cruciate-substituting than with cruciate-retaining arthroplasties. Overall cruciate-substituting arthroplasties display more roll-back and have a better range of movement.16,23,29,31,33
To our knowledge, no paper has been published which has combined clinical outcome data and in vivo fluoroscopic analysis in a prospective, randomised study. Strict randomisation is important since patient selection for performing a cruciate-retaining or cruciate-substituting arthroplasty in the normal setting will inherently be biased.11
Our study sought to investigate if there was a difference in the clinical outcome as measured by the commonly-used scoring systems (International Knee Society Score,34 Western Ontario McMasters Osteoarthritis (WOMAC)35 index and the SF-36 health survey36), survivorship at five years, radiological outcome and a kinematic performance between cruciate-retaining and cruciate-substituting arthroplasties with the same surface geometry.
Patients and Methods
We included 44 patients undergoing a primary TKA in our trial. Since we wished to compare the outcome of two versions, cruciate-retaining and cruciate-substituting designs, of the same prosthesis, we required a study cohort which was as homogeneous as possible with respect to the function of the PCL. We, therefore, followed randomisation protocol to exclude selection bias.
Initially, pre-selection was performed in the outpatient clinic. Patients with degenerative arthritis and a coronal deformity of < 15° were selected as potential candidates for the study. Those with post-traumatic arthritis, previous osteotomy, rheumatoid arthritis or sagittal instability were excluded. Selected patients were informed about the nature of the study and agreed to participate. The next level of selection was performed at surgery. After exposure of the knee, the condition of the PCL was assessed both visually and by palpation. If the PCL was present and macroscopically intact, without excessive tightening at maximum flexion of the knee, the patient was included in the study. If, however, the PCL was in any other condition the patient was not included and underwent a routine cruciate-stabilising TKA outside the study protocol. For each patient who met the criteria, a randomisation envelope was opened and the patient was allocated to one of the two groups. Thus all selected patients had a functioning and macroscopically intact PCL.
Group I had a cruciate-retaining TKA. In group II, the PCL was resected and a cruciate-substituting arthroplasty was used. Fifteen patients (eight from the cruciate-retaining group and seven from the cruciate-substituting group) were randomly chosen for further assessment. This was performed two years after surgery. All patients volunteered to participate in the study and were fully informed about it.
All patients were treated in the same centre by a single surgeon (JV). The implant used was the Genesis II (Smith and Nephew, Memphis, Tennessee) TKA in either its cruciate-retaining or cruciate-substituting version. Other than the femoral housing and the absence of the two femoral fixation pegs and tibial post, both systems were identical. They shared the same configuration of patellofemoral groove, the same asymmentrical condylar geometry and the same sagittal radii of the tibial and femoral surface. In the cruciate-retaining group, a standard retaining insert was used for all patients. No dished inserts were used.
We used a medial parapatellar exposure for all TKAs and identical surgical instrumentation. All knees were cemented and all patellae replaced. Intramedullary alignment was used for the femur and extramedullary for the tibia. Sizing of the distal femur was performed with posterior referencing instrumentation in order to restore both the joint line and the flexion gap as accurately as possible and to avoid subsequent release of the PCL because of an altered level of the joint line. Tibial sizing was performed with great care since any change in anteroposterior positioning could have affected the area of tibiofemoral contact. The prosthesis had eight sizes of asymmetrical tibial trays, which allowed good surface cover of the underlying bone. The tibial component was positioned so that in the sagittal plane the front of the tibial tray aligned with the anterior cortex of the tibia.
All patients underwent an identical post-operative care and rehabilitation protocol although the nursing staff and physiotherapists were blinded as to which group the patient belonged.
This was assessed at three months and at one, two and five years post-operatively. Outcome parameters included evaluation of the Knee Society knee and function score,34 the WOMAC index,35 the SF-36 health survey,36 and radiological analysis. All scores were obtained, and measurements made and recorded, with the help of a trained, independent nurse who was blinded to the procedure which had been performed.
In order to measure alignment and to identify radiolucent lines this was performed by an independent orthopaedic surgeon (JB), who used the Knee Society radiological evaluation system.37 Overall limb alignment was assessed pre-operatively and at three months after operation using a digital full-leg standing radiograph. The accuracy of this technique has already been validated.38 Standard radiographs, including anteroposterior, lateral and skyline views, were taken before operation, at three months and at one, two and five years after surgery. Sagittal alignment was measured as the angle between the posterior tibial cortex and the undersurface of the metal-backed tibial tray. All post-operative radiographs were taken under image-intensifier control in order to position the x-ray beam perfectly parallel to the implant.
Patients performed two weight-bearing activities: weight-bearing deep flexion (lunge) and step-up/down. For the lunge, they placed their foot on a 30- to 42-cm riser and were asked to bend to maximum comfortable flexion. They were not constrained and were allowed, for example, to lift their heel if that permitted a greater range of flexion. An investigator offered to hold the patients’ hands or forearms as a safety measure to prevent a fall. A fluoroscopic video for between one and three seconds was acquired once the patients had reached their maximally flexed position. For the single-leg step-up/down activity, they placed their foot on a 30-cm riser. They were instructed to lift themselves repeatedly up and down on the riser and not to swing through with the opposite leg. They were encouraged to reach a fully-extended position of the knee before reversing direction. An investigator again held the patients’ hands or forearms as a safety measure, but did not provide support for them to lift themselves. Images were recorded at 12.5 frames/sec as the patients performed at least four cycles of the step activity at a self-selected pace. The device which we used was a portable C-arm system (OEC 9800 Cardiac; GE-OEC, Salt Lake City, Utah). This was operated in pulse mode. The video signal was recorded on to S-VHS videotape in PAL format. Images were digitised from the videotape using commercially available software and hardware (02 Workstation, Silicon Graphics Inc, Mountain View, California).
The three-dimensional (3D) position and orientation of the components were determined from model-based shape-matching techniques,15 manual matching, and image space optimisation routines. The images were digitised and corrected for distortion.15 The optical geometry of the image-intensifier system (principal distance, principal point) was determined from images of a calibration target.15 The implant model was projected on to the geometrically-corrected image, and its 3D position was adjusted to match its silhouette with the silhouette of the TKA components for the patient. The results of this shape-matching process had an error of between 0.5° and 1° for rotation and 0.5 and 1.0 mm for sagittal translation.
The kinematics of the joint were determined from the 3D position of each TKA component using Cardan angles as described by Tupling and Pierrynowski.39 The points of condylar contact were estimated as the lowest point on each femoral condyle relative to the transverse axis of the tibial baseplate. The mean centres of axial rotation were determined from lines connecting the medial and lateral contact locations during the step-up/down activity.40
The lunge data represent the 3D pose in a single position. Statistical comparisons of the cruciate-retaining and cruciate-substituting results were performed using the Student’s t-test.
Multiple trials of step data were acquired for each knee. For each knee, the range of flexion was separated into 10° portions and the accumulated data were then used to generate a mean for each knee. Statistical comparisons for the step data were performed using an analysis of variance (ANOVA). If this determined a significant difference (p < 0.05), Tukey’s Honestly Significant Difference method was used to compare pairs.
Measurements of outcome.
Both groups had similar clinical and pre-operative data (Table I⇓), except for pre-operative pain as measured by the WOMAC score only, not by the Knee Society score. The patients in both groups had a similar height, weight, and male-to-female distribution. The magnitude of the deformity was almost identical and the difference in varus/valgus distribution was not significant (p > 0.5) (Table II⇓). A pre-operative fixed flexion deformity of between 5° and 10° was present in nine of the cruciate-retaining patients and in 11 of the cruciate-substituting patients. Two patients in the cruciate-retaining group and one in the cruciate-substituting group had a fixed flex-ion contracture of between 10° and 15°. No patient had a post-operative fixed flexion contracture of more than 5° by the five-year follow-up. At the time of surgery, a similar number of patients (ten cruciate-retaining and 12 cruciate-substituting) required a partial release of the medial collateral ligament in order to balance their joint.
The results of the SF-36 Health survey are shown in Table I⇑. The physical score improved rapidly at three months, with slower recuperation up to one year after operation. Afterwards, this value stabilised. Both groups showed the same pattern. The mental score remained virtually unchanged in both groups throughout the follow-up period.
The results of the WOMAC35 score are also seen in Table I⇑. This score is subdivided into pain, stiffness and function. The pre-operative values showed a difference between the cruciate-retaining and cruciate-substituting groups, with more pain in the cruciate-substituting group. This difference was still present at three months (p = 0.03), despite a clear fall in the amount of pain which was experienced by the patient, but levelled off at one year after surgery. Stiffness was similar for both groups, although there was a greater improvement by one and two years post-operatively. This was not significant.
The function score had a trend towards worse pre-operative function in the cruciate-substituting group, without being statistically significant. It remained somewhat worse for this group but the difference was not significant by two and five years after surgery.
Table I⇑ also shows the values for the Knee Society scoring system. As with the WOMAC score, pain was greater pre-operatively in the cruciate-substituting group but this was not significant. The pain score improved sharply by three months but with little change thereafter. The cruciate-retaining group scored better than the cruciate-substituting group by five years but this difference, too, was not significant. The Knee Society functional score also showed worse pre-operative function for the cruciate-substituting group. Post-operatively, a clear improvement was seen at three months, which had improved even further by one year, although the difference between the two groups was not significant at this stage.
The overall mechanical alignment deviated by less than 3° from the neutral (0°) mechanical axis. Sagittal alignment varied between 88° and 84°, with 2° and 6° of posterior slope, respectively. Radiolucent lines, of less than 1 mm were detected in zone 137 on the tibia in ten patients (five cruciate-retaining and five cruciate-substituting) at two years. No line had progressed by five years. One cruciate-substituting knee had a radiolucent line of 2 mm in zone 737 on the tibia. This had not progressed by five years. The original deformity for this patient had been valgus.
No patient was lost to follow-up. Between the two- and five-year follow-up examinations, ten patients had died, five in the cruciate-substituting group and five in the cruciate-retaining group. No surviving patient underwent or needed revision surgery.
Table II⇑ summarises the details of the patients and lunge kinematics for the subgroup of 15 knees which had been randomly chosen for image-intensifier analysis. There were no significant differences in ROM between cruciate-retaining and cruciate-substituting knees, either passively or during the lunge activity. There was very close agreement between the measured, maximum passive flexion and the maximum knee flexion during the lunge activity. All knees showed condylar contact posterior to the anteroposterior midline during the lunge activity (Fig. 1⇓). Contact was a mean of 8 mm more posteriorly in the cruciate-substituting knees, which was a significant difference for both medial and lateral condyles. Both types of TKA showed approximately 6° of tibial internal rotation in the maximally flexed position.
There was a weakly significant correlation between the anteroposterior position of the femur relative to the tibia and the maximum flexion angle achieved for all knees (Fig. 2⇓). This suggested that the flexion range increased approximately 1.3° for each additional millimetre of posterior femoral translation. However, there was an extremely strong linear relationship (R = 0.89, p = 0.006) between the anteroposterior position of the femur and maximum flexion for the cruciate-substituting knees when these were analysed separately.
During step-up/down activities, there was net femoral external rotation (tibial internal rotation) with knee flexion, of between 5° and 7° (Fig. 3⇓) with no significant differences between the two groups for axial rotations during the step-up/down activity. The groups were neutrally rotated in extension, and reached approximately 5° of tibial internal (femoral external) rotation at 80° of knee flexion. There was a significant difference in medial antero-posterior contact location between the two groups in full extension (Fig. 3b⇓). Cruciate-retaining knees had a mean of anterior medial condylar slide of 4 mm as the knee flexed to 80° compared with posterior translation of 3 mm for cruciate-substituting knees (p < 0.005, Table III⇓). Medial condylar contact was significantly more posterior in the cruciate-substituting knees at 70° to 80° of flexion compared with the cruciate-retaining knees (Fig. 3b⇓).
The cruciate-substituting knees showed significantly more lateral femoral roll-back than the cruciate-retaining knees (Table III⇑, p < 0.005). Lateral condylar contact was also significantly more posterior in the cruciate-substituting knees at 70° to 80° of flexion compared with the cruciate-retaining knees (Fig. 3c⇑). A difference visual representation of these data is given in Figure 4⇓.
As the knee flexed, the cruciate-substituting group showed posterolateral translation, with a little medial translation, resulting in a mean centre of rotation at + 29% medial of the tibial width, or a medial pivot. Forward translation of the medial condyle with flexion and posterolateral translation resulted in a mean centre of rotation at + 1% of the tibial width, or a central pivot, for the cruciate-retaining knees (Fig. 5⇓).
Our study has two main limitations. First, the follow-up time for clinical assessment was relatively short. A follow-up for five years certainly allowed for detection of technical errors, such as malalignment or ligament imbalance, but in the absence of these errors, differences in outcome were hard to detect with the current scores. However, differences in survivorship analysis could not be expected, apart from infection, since revision surgery is rarely needed at this early stage. These differences may appear later. The increased sliding on the medial side which was seen in the cruciate-retaining group may lead to a greater wear of polyethylene and potentially earlier revision for these patients.41,42
Secondly, the number of enrolled patients was relatively small and image-intensifier analysis was only performed on a subgroup of 15 patients. This limited the potential for finding statistically significant differences. Since the kinematic behaviour between cruciate-retaining and cruciate-substituting arthroplasties was so different, there was sufficient statistical power.
The role of the PCL has been a controversial issue since the early days of TKA. Numerous authors have shown good clinical outcomes for both cruciate-retaining and cruciate-substituting designs.3,6,10,11,23,43 Straw et al11 showed similar results for patients with cruciate-retaining and cruciate-substituting arthroplasties. Significantly worse results were reported for patients with a cruciate-retaining arthroplasty when a tight PCL had been released. We avoided this potential bias in our study because of the specific randomisation method since only patients with a macroscopically intact and functional PCL were included. Since our surgical technique aimed to restore the joint line and the flexion gap correctly with the use of posterior referencing instrumentation for sizing the femoral component, no further releases of the PCL were required at the end of the operation in the cruciate-retaining group. Our results showed a similar post-operative outcome for the classical outcome measurement tools of WOMAC, SF-36, and the Knee Society scores. No significant differences were detected at three months and at one, two and five years although this may have been because these scoring systems were too crude to identify important differences. This effect can be caused by the non-parametric character and ceiling effect of these scores.44 Survivorship at five years and the appearance of radiolucencies were identical for both groups, except for one radiolucency in zone 7 on the tibia in a cruciate-substituting knee. Since all knees were well aligned and balanced, this finding is not surprising.
The aim to reproduce normal knee kinematics after implantation of a TKA has been questioned.30 However, other authors share our view that the reproduction of normal kinematic patterns is the best option for preserving stability and movement.13 Since most modern knee arthroplasties are surface replacements which mimic the anatomical form of the human knee, this is a logical assumption. Normal knee kinematics under loaded conditions (deep knee bend) have recently been studied by dynamic MRI,45 biplanar image-matching of radiographs25 and fluoroscopy.26,46 It was shown that the posterior part of the medial femoral condyle had a single radius of curvature, acting like a ball in a socket from between 20° and 110° of flexion, and allowing the lateral condyle to pivot around it.45,47 This positions the lateral tibiofemoral contact point in deep flexion well posterior on the tibia. Asano et al25 used a biplanar image-matching technique to describe the kinematic behaviour of the normal human knee. They found a medial centre of rotation in the horizontal plane as the knee moved from full extension to 120° of flexion. The mean axial rotation of the femur relative to the tibia was 22° at a flexion angle of 90°, and 23.8° at a flexion angle of 120°. The medial contact point moved backwards by 6.9 mm and the lateral contact point by 27.4 mm at 120° of flexion. Differences between these image-intensifier findings and the dynamic MRI data with respect to the kinematics on the medial side of the knee can be attributed to the different ways of describing the kinematic behaviour of the knee. Depending upon the use of either the tibiofemoral contact area or the centre of rotation when describing the movement of the knee, the kinematics of the medial side are different.45,47
Mahfouz et al26 used an image-intensifier to compare the kinematics of the normal and the anterior cruciate-deficient knee during a loaded, deep knee bend. Their ten normal knees displayed femoral roll-back as the knee moved into flexion. At 120° of flexion the mean posterior translation of the medial condyle was 1.9 mm and the mean posterior translation of the lateral condyle was 21 mm. The mean axial rotation of the femur relative to the tibia was 23.7°. In the anterior cruciate-deficient knee, reduced magnitudes of anteroposterior translation and a greater variation in kinematic patterns were seen, with reversed axial rotation between 30° and 45° of flexion.
These reported kinematic patterns are hard to reproduce in patients with a TKA. Several studies have compared the kinematic behaviour of cruciate-retaining and cruciate-substituting implants.6,23,25,26,29,31 Udomkiat et al23 found a mean value for femoral roll-forward during flexion on the medial side of 2.7 mm and a mean roll-back of -2.2 mm (laterally for cruciate-retaining knees). In the cruciate-substituting design, the mean value medially was 0 mm but −1.3 mm laterally. Dennis et al29 compared the findings of cruciate-retaining and cruciate-substituting implants from their multicentre data. They reported better lateral roll-back for the cruciate-substituting group. Forward sliding of the medial femoral condyle during flexion was present in 50% of the cruciate-retaining patients and 70% of the cruciate-substituting patients. Bertin et al32 claimed to have found better kinematic behaviour for a cruciate-retaining arthroplasty. In 20 subjects who had undergone a cruciate-retaining NexGen TKA (Zimmer, Warsaw, Indiana), 13 experienced femoral roll-back medially and 19 laterally. The mean values were −3.1 medially and −3.9 laterally. Although forward sliding of the medial femoral condyle during flexion was less common than in previous reports,16,22,23,30,31,33 the absolute value on the lateral side of −3.1 mm was still much removed from the normal mean values which range between 21 and 27 mm.25,26 Despite differences in the arthroplasties which were used, a clear trend appears from these studies. In the replaced knee, axial rotation is less pronounced than in the human knee, and forward sliding of the medial femoral condyle during flexion on the medial side is present for all cruciate-retaining types of knee while lateral femoral rollback is better for cruciate-substituting than for the cruciate-retaining devices. Maximum flexion tends to be better in the cruciate-substituting than in the cruciate-retaining groups.
In our study, the difference in maximum flexion during lunge between cruciate-retaining and cruciate-substituting patients was not significant although the gain in flexion compared with the pre-operative situation was greater in the cruciate-substituting (14°) than in the cruciate-retaining (5°) group. The better flexion in the cruciate-substituting group correlated with earlier findings.11,29 The explanation probably lies in the greater posterior translation of the femur on the tibia.33 The cruciate-substituting group displayed a greater posterior contact area in flexion than the cruciate-retaining group (Fig. 1⇑). Also, within the cruciate-substituting group, a clear linear relationship strongly suggested that the anteroposterior position of the femur relative to the tibia was a key factor in maximum knee flexion. This can be explained anatomically since a posterior position of the femur relative to the tibia clears the back of the knee and prevents impingement of soft-tissues or polyethylene. This phenomenon works synergistically with the posterior condylar offset as described by Bellemans et al.48
No difference was found between the clinically measured maximum passive flexion (unloaded) and the image-intensifier measured maximum flexion during lunge (loaded). In two studies,19,27 the maximum weight-bearing flexion was reported to be reduced when compared with passive maximum flexion. This phenomenon was most pronounced in cruciate-retaining devices. The difference from our findings may be because of the difference in activity (squat vs lunge) or the different surface geometry of the prosthesis (LCS vs Genesis II).
The total amount of axial rotation was similar for cruciate-substituting and cruciate-retaining arthroplasties. The pattern of condylar movement leading to this axial rotation, however, is statistically different between cruciate-retaining and cruciate-substituting knees (Table II⇑ and Fig. 3⇑), consistent with earlier studies of other designs.16,17,19,20,22,23,29–32,41 Internal rotation of the tibia with increasing flexion was mainly caused by anterior sliding of the medial femoral condyle (Fig. 3⇑) in the cruciate-retaining knees whereas the same internal rotation of the tibia was caused by posterior roll-back of the lateral condyle in the cruciate-substituting knees (Fig. 5⇑). The latter more closely resembles the natural kinematic pattern of the human knee.25,45 As a result of these condylar patterns, the mean centre of axial rotation was central for the cruciate-retaining knees, consistent with earlier findings.49 The forward sliding of the medial femoral condyle during flexion of the Genesis II cruciate-retaining knee which we found was less than has been reported for other designs.16,23,29,31,32 The mean centres of rotation reported by Banks and Hodge49 were −9% for cruciate-retaining and +14% for cruciate-substituting knees, very different to our values of 1% and 29%, respectively. The cruciate-retaining design which came closest to our data was the NexGen knee.32 It is possible that the asymmetrical condylar design of both the Genesis II and the NexGen implants plays a role in the axial rotational pattern of the device.
The forward sliding of the medial femoral condyle in cruciate-retaining knees during flexion may be explained by the initial force of the patellar tendon keeping the tibia anterior, the quadriceps active test.50 With increasing flexion, the direction of pull changes to a more vertical position. The gastrocnemius and hamstrings muscles then exert a posterior pull on the tibia and lead to posterior tibial sub-luxation if there is an insufficient PCL. This phenomenon is partially compensated for in cruciate-substituting arthroplasties since the cam/post mechanism prevents posterior subluxation of the tibia. The fact that it still partially occurs in cruciate-substituting arthroplasties can be explained by the absence of the stabilising function of the anterior cruciate ligament (ACL). In early flexion, the patellar tendon force vector can translate the tibia anteriorly. This positions the femur relatively posteriorly with respect to the tibia. Reduction of this position to the midline causes anterior sliding of the femur on the tibia, which is stopped eventually when the cam-post mechanism engages. This is supported by the work of Komistek et al24 who compared the in vivo kinematics of an ACL-sparing knee arthroplasty with those of a classical posterior-stabilised (ACL-sacrificing) design. They related a posterior tibiofemoral contact position near full extension to an ACL which was either not functioning properly or absent (cruciate-substituting group). Further confirmation of this hypothesis was found in studies which dealt with kinematics after TKA20,30 in which the authors reported the highest posterior translation in ACL-sparing designs.
In conclusion, our randomised controlled trial was not able to demonstrate clinical differences between cruciate-retaining and cruciate-substituting TKAs, nor was there a difference in survivorship or prevalence of radiolucent lines at five years. In kinematic analysis however, it was shown that the cruciate-substituting group had more consistent and more natural function than the cruciate-retaining group, without replicating the kinematics of the normal knee.
A further opinion by Mr Timothy Wilton is available with the electronic version of this article on our web-site at www.jbjs.org.uk
We wish to thank Patrick Stragier for his co-operation and practical support.
No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article.
- Received April 29, 2004.
- Accepted November 10, 2004.
- © 2005 British Editorial Society of Bone and Joint Surgery