Robot-Assisted Unicompartmental Knee Arthroplasty

The Journal of Arthroplasty Vol. 25 No. 2 2010

Robot-Assisted Unicompartmental Knee Arthroplasty Andrew D. Pearle, MD, Padhraig F. O'Loughlin, MD, and Daniel O. Kendoff, MD

Abstract: The outcomes of unicompartmental knee arthroplasties (UKAs) have demonstrated inconsistent long-term survival. We report the first clinical series of UKA using a semiactive robotic system for the implantation of an inlay unicondylar knee arthroplasty. Ten patients were selected for this study. Preoperative mechanical leg alignment values ranged from 0.3° varus to 9.8° varus. A haptic guidance system was used; a detailed description is given in the manuscript. The setup time for the robot was 41 minutes; intraoperative registration process, 7.5 minutes (6-13 minutes); skin incision, 8 cm; robot-assisted burring, 34.8 minutes (18-50 minutes); mean tourniquet time, 87.4 minutes (68-113 minutes); and overall operation time, 132 minutes (118-152 minutes). The planned and intraoperative tibiofemoral angle was within 1°. The postoperative long leg axis radiographs were within 1.6°. Haptic guidance in combination with a navigation module allows for precise planning and execution of both inlay components in UKA. Keywords: robot assisted, uniknee arthroplasty, haptic guidance system, medial osteoarthritis. © 2010 Elsevier Inc. All rights reserved.

Medial unicompartmental knee arthroplasty (UKA) has been shown to give good midterm to long-term results [1,2]. The total number of implanted UKA in the United States is constantly growing; although in 1997, only 1% of all knee implants were UKAs, in 2000, 6% of all implanted knee prostheses were UKAs [3]. Recently, minimally invasive techniques have achieved an overall reduction in soft tissue and bone trauma; however, it has been noted that minimal invasive techniques are not as accurate as open UKA with regard to the anteroposterior-tibial placement and the postoperative leg alignment and the overall revision rate [4-6]. Although the optimal leg alignment has not been fully established, a slight undercorrection is recommended [7,8]. However, substantial undercorrections should be avoided so as to minimize the incidence of early implant failure and polyethylene wear. Postoperative mechanical axis alignments of 8° varus or more have shown to be significantly higher associated with an early revision [9]. Varus undercorrections of 5° are also associated with significantly higher polyethylene wear rates than neutral alignments postoperative [10]. Valgus overcorrections above a desired 180° leg axis revealed

From the Orthopaedic Department, Hospital for Special Surgery, New York, New York. Submitted April 12, 2008; accepted September 14, 2008. No benefits or funds were received in support of this study. Reprint requests: Daniel O. Kendoff, MD, Orthopaedic Surgery, Computer Assisted Surgery Center, Hospital for Special Surgery, 532 East, 72nd Street, New York, NY 10021. © 2010 Elsevier Inc. All rights reserved. 0883-5403/08/2502-0012$36.00/0 doi:10.1016/j.arth.2008.09.024

230

significantly higher risks of lateral compartment degeneration including significantly higher cartilage wear rates [10]. Achieving the optimal medium between undercorrection and overcorrection is challenging and often dependent on the experience of the operating surgeon. In addition to coronal plane alignment, sagittal alignment is also a factor in long-term implant survival because recent data have demonstrated that tibial slope higher than 7° is associated with significant increased loosening rates [11]. Compared with the outcome of total knee arthroplasties, UKAs do show inconsistent long-term survival. Revision rates of between 10% and 20% have been reported [12,13]. Patient-specific factors play a major role, for example, age, weight, and comorbidities. However, correct alignment of the femoral and tibial components has been shown to be the most objectively quantifiable factor [11,14-17]. Recent technical innovations in UKA have included the use of computer-assisted navigation technology. This has been shown to improve postoperative leg alignment compared with conventional UKAs [7,16,18]. However, a direct improvement of the implant positioning itself has not been demonstrated to the best of the authors' knowledge. A general limitation of navigation is that although it is a powerful visual aid, the outcome of the procedure is still dependent on the mechanical tools actually performing the procedure. The outcome of this navigation, especially in minimally invasive knee approaches, can be negatively influenced by the use of relatively imprecise devices such as the conventional oscillating saw. In addition, the image-free registration modules, which are widely used, cannot always accurately identify the complete bone surface in limited surgical approaches.

Robot-Assisted Uncompartmental Arthroplasty  Pearle et al

231

Fig. 1. Based on a preoperative CT scan, the system allows for specific planning of the femoral and tibial implant position, and overlapping of both implants in extension.

Technical developments in the 1990s included the use of robot-assisted techniques. Because of their invasive nature and automatic surgical gestures, these robots were challenging to operate and not widely used [19]. In addition, outcomes were not superior to conventional techniques and showed device-specific complications, which eventually led to an initial rejection of this technology [20,21]. Recently, there have been novel robotic systems and concepts developed to improve the clinical efficacy of this technology in UKAs. These “semiactive” systems give the surgeon active control over the robot. Based on preoperative computed tomography (CT)–based planning, active surgeon-controlled cutting becomes possible with the added benefit of control features such as robotimposed limitations on the areas that may be resected such that iatrogenic complications may be reduced. This allows for more accurate reproduction of the preoperative plan of implant placement and may improve the overall leg alignment [22-25]. However, drawbacks of these more recent robotic systems include the necessity for an invasive frame connecting the robot and patient, thus, limiting free motion throughout the surgical procedure. In addition,

frame fixation has been shown to be problematic, associated to an expanded approach, local infections, and even iatrogenic fractures. Another limitation of this type of system is an exclusive compatibility with a specific unicondylar knee prosthesis.

Fig. 2. The robot is connected to a high-speed burr while the surgeon moves the robotic arm by guiding a force-controlled tip within the defined boundaries.

232 The Journal of Arthroplasty Vol. 25 No. 2 February 2010

Fig. 4. The complete burring process is displayed on a dedicated display on the navigation system.

(McMurry, Pa) prosthesis design are integrated in the system. Based on the preoperative CT image, the system allows for preoperative planning of the femoral and tibial implant position, including the following (Fig. 1):

Fig. 3. Stable fixation of the leg in a leg holder during resection is necessary; however, no fixation from the robot to the patient is necessary.

We report our first clinical series of UKA using a new semiactive robotic system for use with an inlay unicondylar knee arthroplasty. This CT-based technology in combination with a navigation module allows for a complete computer-assisted planning of the implant positioning and further robot-assisted predefined burring of the tibial and femoral component cavities in vivo. Although fixation of the navigation-based reference arrays is required, no rigid fixation of the robot to the patient's anatomy is necessary. Our report consists of the detailed technical considerations of the MAKO Tactile Guidance System (TGS) (MAKO Surgical Corp, Fort Lauderdale, Fla), as well as an overview of our first 10 clinical cases.

• • • • •

coronal and sagittal alignment, overall leg alignment, gross anatomical deformities (cyst and vacuole), overlapping of the components in extension, geometric alignment of varus/valgus of femoral component to the varus of the tibia implant, • tibial implant positioning relative to the posterior tibial wall. Once the optimal implant position is defined by the surgeon, parameters are saved on the system. Consequently, the bone resection areas are defined automatically by the system, and boundaries for the cutting instrument are set to prevent cutting into areas outside these boundaries. Optimal implant position and leg alignment in each individual case was based on a proper

Technique and Patients Preoperative Planning Customized CT-based planning is performed before every surgery. The scan protocol requires patients to lay supine with a motion rod attached to the operative leg. Slices are taken through the hip and ankle (5-mm slices) as well as the knee joint region (1-mm slices). The scans are saved in DICOM 3 format and transferred into the TGS software (MAKO Surgical Corp). The bone surfaces are segmented in the software to produce a 3-dimensional (3-D) model, upon which the implant positioning may be planned. All femoral and tibial sizes of the StelKast

Fig. 5. A 3-D model of the knee during the burring process indicates the bone material remaining to be removed.

Robot-Assisted Uncompartmental Arthroplasty  Pearle et al

plan, which was informed by a navigated preoperative kinematic analyses throughout the passive range of knee motion under a valgus load applied by the surgeon. Operative Technique The TGS (MAKO Surgical Corp) consists of 3 components: robotic arm, optical camera, and operator computer cart. The robotic arm features 5 degrees of freedom with its movement limited to the incision site via the 3-D virtual boundaries set in the software. The optical camera included with the system is an optical infrared system, whereas the operator computer cart houses the software, which drives the surgical plan. The distal end of the robot is connected to a high-speed burr (Fig. 2). The surgeon moves the robotic arm by guiding the force-controlled tip within the defined boundaries. The surgeon can sense when he or she is cutting bone, and the feedback mechanism of the robot includes the active prevention of inaccurate motion out of the designated areas. The robot gives the surgeon active feedback (haptic and audio) and allows for a quick burring process of even the complex shapes of the femoral and tibial bone surface. As the surgeon approaches the predefined boundaries of the implant position, the robot provides audio feedback and eventually haptic feedback at the boundaries via active stiffness of the robotic arm. In addition, excessive pressure against the limits of the 3-D cutting volume or rapid movement of the patient's anatomy immediately stops the cutting instrument, preventing unintentional resection outside the implant area from occurring. In comparison with other active and semiactive systems, the TGS does not require rigid fixation of the robot to the patient. The surgeon uses a leg holder to hold the leg stable during resection, but is able to position the knee optimally to ensure access to the targeted surfaces (Fig. 3). Furthermore, we have noted that the incision size can be as short as 2.25 in with minimal strain on soft tissue. The Burr System The system consists of a high-speed 75.000 rpm electric burr (eMax2 from the Anspach Effort, Palm Beach Gardens, Fla), which is operated via a foot pedal. Power is controlled and ensured via linkage with the robotic system. Three burr sizes are used during surgery: a 6-mm diameter spherical burr is used for rapid removal of the major bone material and for resection for the femoral post, a 2-mm diameter spherical burr is used for fine finishing, and the corners of the cutting volume and deep milling of the mini femoral canal are performed with a 1.2 mm router. The complete burring process is displayed on a dedicated surgeon display, which also shows the 3-D model of the knee, indicating the bone material remaining to be removed (Figs. 4 and 5). Implant The StelKast Unicondylar Knee System (StelKast Corporation) consists of a nonmetal-backed polyethylene

233

tibial inlay insert, a CoCr femoral component, and a femoral implant including a mini stem. Setup and Surgical Technique Positioning of the TGS is performed before the patient's arrival in the operating room (OR). Positioning of the system is based upon the affected knee and surgeon's preference (right hand vs left hand). The line-of-sight between the robot reference array and the optical camera is approximated before surgery. Once the system is positioned, the robotic arm is secured with brakes to prevent any motion. After conventional positioning and sterile draping of the patient's leg, the robot is registered. The surgeon moves the robotic arm through a defined 3-D movement. This calibrates the robotic arm movements and sets the tool center point for the cutting instrument. Attachment of femoral and tibial reference arrays follows. Anatomical surface landmarks are registered before skin incision, and the surgeon draws the patient's leg through a full range of motion applying the appropriate valgus load on the joint. After the incision, small checkpoint pins are fixated on the tibia and femur, and the 2 bone surfaces are registered. The registration of the knee anatomy to the digitized CT model is then verified before proceeding. At this time, the surgeon may alter the implant position based upon an analysis of the implant overlap through the range of motion collected moments before. The robot assists the surgeon during defined burring of the tibial and femoral component cavities. It allows for a controlled depth and width of burring of the cavity with graphical feedback on the navigation monitor of the milled bone. It is recommended to prepare the tibial cavity before the femoral to allow for any changes to the femoral component position; however, any resection order can be chosen by the surgeon. After rough burring of both cavities (including the femoral post hole) with the 6-mm spherical burr, fine milling is performed with the 2-mm spherical burr.

Table 1. Summary of Patient-Related Preoperative and Postoperative Long Leg Alignment as Well as Planned Alignment Patient no.

Age

Sex

Preoperative Alignment

Postoperative Alignment

Planned Alignment

1 2 3 4 5 6 7 8 9 10

72 57 59 79 49 64 64 47 86 76

M M M M F M M F M F

4.5° 5° 9.8° 4.5° 2.8° 6° 7° 0.3° 8° 5.6°

2.6° 2.5° 4.3° 2.9° 0.9° 2.4° 3.2° 0.2° 3.5° 2.1°

2.5° 2.5° 4° 2.5° 1° 2.5° 3.5° 0° 4° 2.5°

M indicates male; F, female. Demographic data on all 10 patients are presented.

234 The Journal of Arthroplasty Vol. 25 No. 2 February 2010

Fig. 6. Summary of the first 5 patients' postoperative radiographs in anterior-posterior (upper column) and lateral projection (lower column).

Burring of the femoral keel canal is achieved with a 1.2-mm fluted router. Permanent graphical feedback on the navigation screen visualizes the actual achieved vs the planned cavity, specifically based on preoperative planning. Once both cavities have been milled out, femoral and tibia component trials are inserted and a complete flexion-extension arc is performed. Computerized simulation of the implants in situ shows the actual overlapping of the implant components, giving the surgeon feedback about the current leg alignment and knee gap kinematics. Finally, after acceptance of the implant position, both implant components are cemented, and a final range of motion can be taken to compare the original, trial, and final implant kinematics and knee alignment (varus/ valgus, internal rotation). Before site closure, both mini checkpoints and both bone reference arrays are removed. Patients Patients' ages varied from 47 to 86 years, with 7 male and 3 female patients included (Table 1). Eight patients had an ongoing history of medial osteoarthritis of the knee joint with an increase of pain and loss of motion within the previous 12 months. Two patients had symptomatic avascular necrotic lesions of the medial femoral condyle with collapse of the chondral surface. Preoperative mechanical leg alignment values ranged

from 0.3° varus to 9.8° varus. No relevant comorbidities were found in those patients. Patients were informed about the use of the robot system preoperatively, whereas information about the operative details was specifically included in the final clinical assessment. All operations were performed by one surgeon (A.P.) using a combination of epidural and femoral block anesthesia. A tourniquet was used in all cases. Recorded intraoperative parameters included total operation time, time of robotic use, and tourniquet time.

Results The use of the MAKO system was feasible in all 10 cases. No technical failures or equipment-specific problems occurred. The setup time for the robot averaged 41 minutes and was done by a specialized technician. This was performed before surgery while the patient had the epidural anesthesia administered. The average time for the intraoperative registration process was 7.5 minutes (6-13 minutes). The length of skin incision was on average 8 cm. Duration of time needed for robot-assisted burring was on average 34.8 minutes (18-50 minutes), whereas the first 5 cases averaged 42.8 minutes and the last 5 cases, 27.3 minutes. Mean tourniquet time was 87.4 minutes (68-113 minutes), and the overall operation time was 132 minutes (118-152 minutes). In all patients,

Fig. 7. Summary of the last 5 patients' postoperative radiographs in anterior-posterior (upper column) and lateral projection (lower column).

Robot-Assisted Uncompartmental Arthroplasty  Pearle et al

the planned and subsequent intraoperative tibiofemoral angle in the coronal plane was within 1°. Postoperative long leg axis radiographs were within 1.6° to the intraoperative measured values. The average stay in hospital was 2.2 days. At 6 weeks postoperatively, there were no complications. All patients were fully weight bearing and presented with uneventful wound healing. Values for the range of knee motion after 6 weeks were mean flex/extension: 125/3°. A summary of all the patients' postoperative radiographs are presented in Figs. 6 and 7.

Discussion We have demonstrated successful robot-assisted UKA placement in a series of our first 10 patients. It may be particularly applicable to clinical cases where minimal access is desirable and where the bone footprint for both implant components requires greater accuracy. Although it has been shown that exact implant placement affects the clinical short-term and long-term outcome of UKA, this new technique may be able to improve the positioning based on the patients individual anatomy and with regard to the planned leg alignment [11,17,26,27]. Previous studies on a comparative robotic system for UKA also showed promising initial results. Cobb et al [22] were able to demonstrate, in a prospective clinical study, a significant improvement in implant placement, and that accurate leg alignment can be achieved successfully with the aid of a semiactive robot system in UKA [25]. Although the technique and robotic system Cobb et al described may be comparable with the current authors' system, it does have some important limitations. Most importantly, the MAKO system does not depend on any permanent or rigid fixation to the patient's bony anatomy. Registration via regular navigation-based reference markers allows for a dynamic tracking of femur and tibia, whereas the robot movements are independent of the patient's positioning or movement. Therefore, there is no further rigid fixation device necessary, which reduces potential complications such as infection, iatrogenic fractures, or soft tissue injury, because of the robot's weight and movement. In addition, another drawback of rigid fixation is a potential reduction in the scope of the approach to the knee and intrusion into the surgical field. The current authors believe that avoiding these potential complications and limitations associated with rigid fixation could be a real advantage of the new system described herein. Another advantage of the MAKO system is the greater precision in burring of the bony surface compared with regular UKA cutting guides. It permits the creation of individual bony surfaces of any shape, which cannot be generated by an oscillating saw. Consequently, a press-fit cavity for the implant can be created. Thus, preservation of the remaining bone surface is possible, which can be very useful for revisions and conversions to total knee prosthesis. In addition, with reduction in “collateral

235

damage,” these bony areas can be used to introduce complete new CT-based custom-made prostheses with more complex and individual shapes including bone or soft tissue conserving designs. By virtue of permanent visual feedback on the amount of remaining bone and the specific position of the burr, less invasive approaches to the knee may be possible, which can be particularly critical in patients with complex bone defects, soft tissue problems, or coexisting metabolic bone disease. As previous studies have shown, minimally invasive UKA has resulted in higher rates of revision and more frequent aseptic loosening than conventional UKA, which is likely to be due to the increased difficulty in identifying bony landmarks [5,7,18]. We believe that, in general, CT-based planning allows for better feedback of implant position and intraoperative simulation of the implant overlap during complete knee movements. Consequently, it enables the surgeon to alter his plan and include more patient-specific data intraoperatively. The joint line and the orientation of the implant can be planned and executed more accurately than with conventional techniques. As shown by Cobb et al [22], with their Acrobot system, it was possible to reproduce a preoperative plan more accurately with robot assistance. They showed that the tibiofemoral alignment achieved with the robot was consistently within 2 degrees of the planned position, whereas this result was achieved in only 40% of those cases where conventional techniques were used. This strongly suggests that greater outcomes may be obtainable in a clinical setting. Another benefit in the use of the robot may be a shorter learning curve for the surgeon, especially those in the earlier stages of their training [27]. There may also be a reduction in surgical errors associated with minimally invasive approaches. Minimally invasive techniques for UKA placement by definition restrict the size of the surgical field and make it more challenging to identify all relevant bony landmarks. Therefore, the clinical experience of the surgeon is often recognized to be one of the most crucial factors during UKA surgery. Although the system offers some unique technical advantages over conventional as well as previously described navigation techniques, we are aware of several drawbacks. It must be mentioned that the overall costs of the system are high, excluding additional costs for CT scanning and regular maintenance of the robot. It is also necessary to use additional skilled personnel in the OR. However, that may become unnecessary once a defined and established clinical use protocol has been developed. The relatively complex setup of the robot in the OR has to be fully defined and established before sterile draping of the patients commences and to be specifically tailored to each case. Finally, CT-based systems fail to incorporate soft tissue tension into the planning. Gap kinematics, however, are tracked intraoperatively by tracking a manual flexion/extension cycle of the knee before the

236 The Journal of Arthroplasty Vol. 25 No. 2 February 2010 burring process. The implant placement can be refined based on the predicted gaps; this soft tissue balancing process is distinct from traditional practices, and its reliability is unknown. The end point of this study was at 6 weeks postsurgery, and hence, long-term results were not obtained. However, the study was not designed to be a prospective clinical study. We sought to point out the new and relevant technical features of this particular system. All 10 patients in our group were operated upon by one surgeon (AP), which is both advantageous and disadvantageous. On one hand, the approach was consistent, and operating time was seen to decrease as the surgeon became increasingly familiar and adept with the equipment. For the first 5 cases, the average operating time was around 140 minutes, whereas the average time taken to complete the surgery in the second 5 was around 120 minutes. On the other hand, we cannot make any conclusions about successful use between surgeons. It needs to be mentioned that this operating time is still relatively long compared with conventional UKA time, but we would consider it acceptable at this time because the system is being developed and validated. We envisage the operating time being further reduced as familiarity and specific skills are developed. In addition, we envisage other future applications of the system, including procedures where precise bone burring is particularly important. In conclusion, critical parameters such as precise implant placement and leg alignment can be monitored and actively controlled intraoperatively with the recently introduced robotic system for UKA. Burring of the exact cavity of interest for the specific implants allows for greater conservation of bony surfaces during arthroplasty. The likelihood of satisfactory and reproducible long-term clinical results may therefore be higher than with conventional techniques. Prospective clinical studies are needed to truly establish this. Future applications may involve procedures where precise bone burring is especially necessary.

References 1. Berger RA, Meneghini RM, Jacobs JJ, et al. Results of unicompartmental knee arthroplasty at a minimum of ten years of follow-up. J Bone Joint Surg Am 2005;875:999. 2. Berger RA, Nedeff DD, Barden RM, et al. Unicompartmental knee arthroplasty. Clinical experience at 6- to 10-year followup. Clin Orthop Relat Res 1999;367:50. 3. 2002 US markets for reconstructive devices 2002. Toronto (Ontario, Canada): Millenium Research Group; 2003. 4. Dalury DF, Dennis DA. Mini-incision total knee arthroplasty can increase risk of component malalignment. Clin Orthop Relat Res 2005;440:77. 5. Fisher DA, Watts M, Davis KE. Implant position in knee surgery: a comparison of minimally invasive, open unicompartmental, and total knee arthroplasty. J Arthroplasty 2003;187(Suppl 1):2.

6. Hamilton WG, Collier MB, Tarabee E, et al. Incidence and reasons for reoperation after minimally invasive unicompartmental knee arthroplasty. J Arthroplasty 2006;216 (Suppl 2):98. 7. Keene G, Simpson D, Kalairajah Y. Limb alignment in computer-assisted minimally-invasive unicompartmental knee replacement. J Bone Joint Surg Br 2006;881:44. 8. Ridgeway SR, McAuley JP, Ammeen DJ, et al. The effect of alignment of the knee on the outcome of unicompartmental knee replacement. J Bone Joint Surg Br 2002;843: 351. 9. Collier MB, Engh Jr CA, McAuley JP, et al. Factors associated with the loss of thickness of polyethylene tibial bearings after knee arthroplasty. J Bone Joint Surg Am 2007;896:1306. 10. Hernigou P, Deschamps G. Alignment influences wear in the knee after medial unicompartmental arthroplasty. Clin Orthop Relat Res 2004;423:161. 11. Hernigou P, Deschamps G. Posterior slope of the tibial implant and the outcome of unicompartmental knee arthroplasty. J Bone Joint Surg Am 2004;86-A3:506. 12. Eickmann TH, Collier MB, Sukezaki F, et al. Survival of medial unicondylar arthroplasties placed by one surgeon 1984-1998. Clin Orthop Relat Res 2006;452:143. 13. Register SKA. The Swedish Knee Arthroplasty Register: Annual Report 2006. Available at: http://www.knee.nko. se/english/online/uploadedFiles/107_SKAR2006_Eng/v2. pdf 2005. 14. Emerson RH, Head WC. Failure mechanisms of unicompartmental knee replacement: the impact of changes in operative technique and component design. Semin Arthroplasty 1991;21:23. 15. Emerson Jr RH, Higgins LL. Unicompartmental knee arthroplasty with the oxford prosthesis in patients with medial compartment arthritis. J Bone Joint Surg Am 2008; 901:118. 16. Jenny JY, Boeri C. Unicompartmental knee prosthesis implantation with a non–image-based navigation system: rationale, technique, case-control comparative study with a conventional instrumented implantation. Knee Surg Sports Traumatol Arthrosc 2003;111:40. 17. Mariani EM, Bourne MH, Jackson RT, et al. Early failure of unicompartmental knee arthroplasty. J Arthroplasty 2007; 226(Suppl 2):81. 18. Jenny JY, Ciobanu E, Boeri C. The rationale for navigated minimally invasive unicompartmental knee replacement. Clin Orthop Relat Res 2007;463:58. 19. Siebert W, Mai S, Kober R, et al. Technique and first clinical results of robot-assisted total knee replacement. Knee 2002; 93:173. 20. Honl M, Dierk O, Gauck C, et al. Comparison of roboticassisted and manual implantation of a primary total hip replacement. A prospective study. J Bone Joint Surg Am 2003;85-A8:1470. 21. Siebel T, Kafer W. Clinical outcome following robotic assisted versus conventional total hip arthroplasty: a controlled and prospective study of seventy-one patients. Z Orthop Ihre Grenzgeb 2005;1434:391. 22. Cobb J, Henckel J, Gomes P, et al. Hands-on robotic unicompartmental knee replacement: a prospective, randomised controlled study of the acrobot system. J Bone Joint Surg Br 2006;882:188.

Robot-Assisted Uncompartmental Arthroplasty  Pearle et al 23. Cobb JP, Kannan V, Brust K, et al. Navigation reduces the learning curve in resurfacing total hip arthroplasty. Clin Orthop Relat Res 2007;463:90. 24. Davies BL, Rodriguez y Baena FM, Barrett AR, et al. Robotic control in knee joint replacement surgery. Proc Inst Mech Eng [H] 2007;2211:71. 25. Rodriguez F, Harris S, Jakopec M, et al. Robotic clinical trials of uni-condylar arthroplasty. Int J Med Robot 2005;14:20.

237

26. Assor M, Aubaniac JM. Influence of rotatory malposition of femoral implant in failure of unicompartimental medial knee prosthesis. Rev Chir Orthop Reparatrice Appar Mot 2006;925:473. 27. Rees JL, Price AJ, Beard DJ, et al. Minimally invasive oxford unicompartmental knee arthroplasty: functional results at 1 year and the effect of surgical inexperience. Knee 2004; 115:363.