The Knee 21 (2014) 428–434
Contents lists available at ScienceDirect
The Knee
Unicompartmental knee arthroplasties: Robot vs. patient specific instrumentation☆ Zahra Jaffry, Milad Masjedi ⁎, Susannah Clarke, Simon Harris, Monil Karia, Barry Andrews, Justin Cobb MSk Lab, Department of Orthopaedics, Imperial College London, Charing Cross Hospital, Fulham Palace Road, London W6 8RF, UK
a r t i c l e
i n f o
Article history: Received 15 August 2013 Received in revised form 28 October 2013 Accepted 21 November 2013 Keywords: Osteoarthritis Unicompartmental knee replacement Surgical technique Surgical accuracy Three dimensional analysis
a b s t r a c t Background: The technical reliability demonstrated by semi active robots in implant placement could render unicompartmental knee arthroplasties (UKAs) more favourable than they are currently. The relatively untested method using patient specific instrumentation (PSI), however, has the potential to match the accuracy produced by robots but without the barriers that have prevented them from being used more widely in clinical practice, namely operative time. Therefore this study took a step towards comparing the accuracy and time taken between the two technologies. Methods: Thirty-six UKAs were carried out on identical knee models, 12 with the Sculptor, 12 with PSI and 12 conventionally under timed conditions. Implant placement in these knees was then judged against that in a pre-operative plan. Results: Tibial implant orientations and femoral implant positions and orientations were significantly more accurate in the PSI group with mean errors of 6°, 2 mm and 4° respectively, than the conventional group which had means of 9°, 4 mm and 10°. There was no significant difference between the robot and PSI generally except in tibial implant orientation (mean robotic error 3°) and tibial implant position did not vary significantly across all three groups. It was also found that use of PSI and conventional methods took half the time taken by the robot (p b 0.001). Conclusions: With further development, PSI can match and possibly surpass the accuracy of the robot, as it does with the conventional method, and achieve planned surgery in less time. Clinical relevance: This work sets the foundation for clinical trials involving PSI. © 2013 Elsevier B.V. All rights reserved.
1. Introduction Surgical precision often influences clinical outcomes in technically demanding procedures such as unicompartmental knee arthroplasties (UKAs) [1,2]. Limitations of the technique make ideal implant placement difficult [3] affecting limb alignment and hence the levels of wear, loosening and other reasons for failure and revision [4,5]. Therefore, despite the facts that UKAs allow greater bone conservation, faster functional recovery and are more cost-effective [6], the alternative of a total knee arthroplasty is currently considered the more reliable approach due to higher ‘survival rates’ [7] and perceived lower revision costs [8]. The technology developed to address this problem are semi-active robotic systems, like the Sculptor Robotic Guidance Arm (Stanmore Implants Worldwide, Elstree, UK), which combine elements of both
☆ All authors are affiliated with Imperial College London. ⁎ Corresponding author. Tel.: +44 20 3313 8833; fax: +44 20 3313 4427. E-mail address: [email protected] (M. Masjedi). 0968-0160/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.knee.2013.11.017
navigation and fully autonomous systems. Using pre-operative planning, they offer surgeons constraint outside of a pre-set zone of safety rather than simply providing them with the information to determine this zone themselves [9,10]. The results of UKAs done in patients by the predecessor to the Sculptor have been promising with all tibiofemoral alignment angles being within 2° of the original plan [11]. Yet, the tool has not been widely adopted. The recent introduction of rapid prototyped cutting guides made to match the anatomy of the patient provides an alternative method of carrying out the same task in less time. Also using a Computer Tomography (CT) or Magnetic Resonance (MR) derived plan [12], these cutting guides, or Patient Specific Instrumentation (PSI), achieve results without the use of a robotic system offering a simpler embodiment of many steps involved in the procedure [13,14]. This may lower the surgical time taken but it is not clear whether the precision of this technology can contest that of the robots. This study therefore set out to investigate the Sculptor Robotic Guidance Arm as an example of a semi active robot, PSI (Embody, London, UK) and conventional instrumentation and aimed to answer the following question: which of these are able to deliver the standard of accuracy required for consistent clinical results in UKA surgery and in what time?
Z. Jaffry et al. / The Knee 21 (2014) 428–434
2. Materials Dry bone models used in the study were produced by Medical Models Ltd. (Bristol, UK) in a single batch from the same mould, the ‘Imperial LB Varus Knee’. The model was based on a CT scan of a patient with medial compartment osteoarthritis. Each was composed of a femur and tibia held together by cords that simulated ligaments, a plastic patella and an outer capsule that surrounded the bones. 3. Methods Digital Imaging and Communications in Medicine (DICOM) files from a CT scan of the dry bone models were used to create a Stereolithograph (STL) file with Modeller (Stanmore Implants Worldwide, Elstree, UK) and a senior surgeon planned the operation on Planner (Stanmore Implants Worldwide, Elstree, UK). A Uniglide size 4 tibial base plate and size 3 femoral component from the mobile bearing range (Corin, Cirencester, UK) were considered appropriate to use with the bone models. The pre-operative plan was designed to restore the natural joint line with implant positions and orientations that could realistically be achieved with the conventional cutting guides and alignment rods accompanying the Uniglide set. Hence, bias against the technique was
429
avoided. This involved placing the tibial component with a posterior slope of 7° (the angle at which the Corin instrument set aims to cut the tibia), medial slope of 0° and axial rotation of − 1° in relation to tibial frames of reference. Corresponding values for the femoral component were placed at 5°, 0° and 7°. The plan was then loaded onto the Sculptor to be used by the robotic group of the study. Images of the plan, including the values mentioned previously, were also printed out for surgeons to prevent discrepancies between them and define what should be achieved with the instrumentation. Guides constituting the PSI were designed on computer aided design software (Solidworks, Waltham, Massachusetts and Rhinoceros, Barcelona, Spain) using the same 3D model generated by Modeller and the STL files from Planner. These were then produced by the Objet Eden250 3D Printing System (Objet, Rehovot, Israel) in a medical grade polymer (Objet MED610). Three experienced users of the Sculptor, having performed the procedure over 20 times, were recruited. The bone model was attached to a frame that hangs from a table to replicate the position of the knee in surgery. The timer was started once the robot had been calibrated and after the incision into the capsule surrounding the bone had been made. The tracking arm was pinned to the tibia and the computer models were registered to the physical bone using the tip of the cutting burr arm, ensuring the root mean square (RMS) error in registration was less than
Fig. 1. Photographs showing the experimental setup including the bone model (A) and use of the Sculptor (B), Patient Specific Instrumentation (C and D) and conventional instruments (E and F).
430
Z. Jaffry et al. / The Knee 21 (2014) 428–434
Fig. 2. Diagrams showing:(A) The positions of planned (grey) and achieved (blue) tibial implants; (B) The points 1–4 that were compared between the two implants by creating local frames of reference for each where the origin is the intersect of the x, y, z axes; (C) The positions of planned (grey) and achieved (blue) femoral implants; (D) The points 1–4 that were compared between the two femoral implants by creating local frames of reference for each where the origin is point 4.
Table 1 Compound translational and rotational error values in 36 unicompartmental knee arthroplasty implant placements. Compound translational error (mm)—tibial component Sculptor Mean 2 SD 1 (min–max) (1–3)
PSI
Conventional
Sculptor
PSI
Conventional
2 1 (1–3)
3 1 (1–4)
2 1 (1–5)
2 1 (1–4)
4 2 (2–8)
Compound rotational error (º)—tibial component Sculptor PSI Mean 3 SD 1 (min–max) (2–5)
Compound translational error (mm)—femoral component
Compound rotational error (º)—femoral component
Conventional Sculptor PSI
6 9 3 3 (3–11) (3–14)
3 1 (2–5)
Conventional
6 10 3 5 (3–11) (4–22)
1 mm and the maximum error less than 2 mm. Once the tibial cut was made by the high speed burr, the same sequence of steps was carried out for the femur. The timer was stopped after nibblers had been used to remove any excess over-hanging bone. Between the three participants, a total of 12 UKAs were done in this way. Each surgeon was then also asked to complete four UKAs using the PSI and four using the conventional guides and instruments provided by Corin in a random order. PSI involved drilling the guides into place for the main cuts and later steps also used conventional instruments. As previously, the timer for each procedure was started after the incision into the capsule was made. The timer was stopped after the final keel cuts. A total of 12 were done with the PSI and 12 conventionally. The experimental design for each technique is illustrated in Fig. 1. From this point all cut bones were treated in the same manner. The capsule surrounding the model was removed and the tibia and femur were separated by cutting the cords between them. The Corin implants
Table 2 p-values of differences in unicompartmental knee arthroplasty implant placements between groups. Tibial component
Femoral component
Translational error (mm) Sculptor Sculptor PSI Conventional
PSI 0.995
0.995 0.840
Conventional 0.840 0.680
0.680
Sculptor Sculptor PSI Conventional
0.991 0.004
PSI
Conventional
0.991
0.004 0.030
0.030
Rotational error (°) Sculptor Sculptor PSI Conventional
0.014 b0.001
PSI
Conventional
0.014
b0.001 0.092
0.092
Sculptor Sculptor PSI Conventional
1.000 0.002
PSI
Conventional
1.000
0.002 0.002
0.002
Z. Jaffry et al. / The Knee 21 (2014) 428–434
were coated in matt enamel paint to ensure correct measurement of the laser spot from the 3D scanner. Implants were then placed in the bones and each bone was individually loaded and scanned using the NextEngine Desktop 3D Scanner, Model 2020i (NextEngine, Santa Monica, California). The scans were saved as STL files that could be
431
matched onto the 3D STL model of the plan using 3matic software (Materialise, Leuven, Belgium). Once the bones had been matched, four identical co-ordinates were selected on the implant for both planned and achieved components. Using Matlab software (MathWorks, Natick, Massachusetts), these co-
Fig. 3. Box plots showing range of translational error for the tibial and femoral components in the lateral–medial (A), anterior–posterior (B) and distal–proximal (C) directions in Sculptor, Patient Specific Instrumentation and Conventional groups. The reference lines in each indicate an acceptable range in which the error can lie.
432
Z. Jaffry et al. / The Knee 21 (2014) 428–434
ordinates were used to construct a local frame of reference and aligned with anatomical frames. The origins of tibial and femoral components were the average of four set points on the implant and point 4 respectively (Fig. 2). The x axis points laterally, y anteriorly and the z axis is perpendicular to the x and y axes, pointing inferiorly. The position and orientation of the implant were then compared to that of the planned implant in all six degrees of freedom. To validate the use of the 3D laser scanner, an uncut tibia was scanned and matched to the 3D model from Modeller to test it against
a ‘gold’ standard scanning technique- CT. The difference between models was found by finding the average distance between two corresponding points on the two bone models in three different places. Five scans of the same bone were taken and the difference in implant position between each scan was established. This process was replicated by another investigator so both intra- and inter-observer reliability were measured. The magnitude of error (compound error) in both position and orientation of implants was found using the sum of the RMS values of
Fig. 4. Box plots showing range of rotational error for the tibial and femoral components in the flexion–extension (A), varus–valgus (B) and axial rotation (C) directions in Sculptor, Patient Specific Guide and Conventional groups. The reference lines in each indicate an acceptable range in which the error can lie.
Z. Jaffry et al. / The Knee 21 (2014) 428–434
error in the translational (lateral–medial, anterior–posterior and distal– proximal) and rotational (flexion–extension, varus–valgus, axial rotation) directions respectively for all groups. An analysis of variance (ANOVA) test was carried out using SPSS statistical analysis software (IBM, Armonk, New York) to identify the most accurate technique for carrying out a UKA. Post-hoc analyses provided further information on the data and p-values less than 0.05 were considered significant (α = 0.05, β = 0.2). Another ANOVA test was carried out to compare the average times taken to complete a procedure between groups. Finally, the Bland and Altman statistical technique was used to determine the level of agreement between repeat measurements of the implant position in the same bone and among two different observers. 4. Results The Bland–Altman plots demonstrated that all points fell within two standard deviations of the mean difference indicating the reliability of the method. Implant placement accuracy is compared across groups in terms of the mean compound translational and rotational errors in Table 1. Conventional instruments delivered substantially less accuracy than both the Sculptor and the PSI, with Sculptor and PSI groups performing comparably in all but tibial implant orientation, where the Sculptor was more accurate. There was no significant difference across groups in tibial implant position. The p-values of differences between groups are shown in Table 2. The results are expanded in Figs. 3 and 4 which show the range of all three groups in six degrees of freedom for both the tibial and femoral components. This allows an idea of the precision achieved and the reference lines indicate an acceptable range in which implant placement should be [15]. Most Sculptor and PSI plots fall within these lines while ten of the twelve conventional plots fall outside of the accepted error range; the ones to notice are those where the median falls out of this range, which is never the case for the two other groups. Conventional instruments produced a wider range in all variables, most notably in tibial axial rotation (Fig. 4C), where the range was ±9°, compared with ±5° for PSI and ±3° for the robot. The mean time taken to complete a procedure with the Sculptor (20 min and 25 s ± 5 min 31 s) was almost double that of PSI (7 min and 52 s ± 2 min 1 s) and conventional (10 min and 31 s ± 3 min 19 s) groups with an overall p value of less than 0.001. No significant difference was found between PSI and conventional groups (p = 0.232).
5. Discussion This study investigated three techniques for UKA surgery and compared the implant placement accuracy and operative time taken as a step towards improving clinical outcomes. Surgical precision has been determined by differences between the planned and achieved implant position and orientation [9]. While there is no consensus on the level of accuracy actually needed for clinically satisfactory results, there is evidence suggesting that translational or rotational error in prosthesis placement, i.e. greater than a 7° posterior slope (in flexion–extension) in the tibial component [4] or any other error resulting in angular errors of more than 5° in the knee [5], may lead to poorer outcomes [15]. The plan was made in accordance with this and reference lines in Figs. 3 and 4 indicating an ideal range in which the plan should be achieved allowed easy evaluation of results and comparison between groups. Personal communication with Corin ensured that these ranges were the same as the ones stated by the Oxford knee manual used for this study. These values have previously been adopted by studies that had similar methods in their use of plastic dry bones, which eventually lead to the Sculptor Report [16]. Reliability of the method was ensured in a number of other ways. The use of a laser scanner (rather than CTconsidered the gold standard technique) had been validated in this study using intra- and inter-observer reliability tests. Matches of 3D models from CT images and the laser scanner were also compared on 3matic showing a maximum difference of 1 mm between models for both the tibia and femur. The use of laser scanning over CT as well as the use of the Sculptor software has also been validated in other papers [17,18]. Thirty-six UKAs being carried out with three techniques and three surgeons can be argued to introduce confounding variables and past literature has provided a standard deviation with which to carry out
433
power calculations suggesting the use of a larger number [11]. However, less variability was expected due to the use of identical bone models rather than patients and the use of a relatively novel technique, i.e. the PSI, in addition to the robot discussed in these previous papers. Nevertheless, the methods, including the bone models used, were realistic, served the purpose of the study well and produced results with the robot that were comparable to its use elsewhere [11,16]. As expected, implant placement was less predictable in the conventional group than the other two. PSI was shown to be only less accurate in tibial implant orientation compared to the robot and with the significantly lower time spent on the procedure, the potential PSI has to surpass the Sculptor in its efficiency with further development is highlighted. Although, it is important to note at this point that the shorter operative time does come at the price of longer planning time. As a small study on dry bones, conclusions now need to be enforced with further investigations in patients. It is important to continually test efficacy to keep track of progress in the trend towards minimally invasive surgery [19] and form foundations for cost-effectiveness analyses [6]. Between the time of previously mentioned studies and this one, the robot alone has undergone many improvements including a reduction in size, cost and operating system [20]. More importantly, newer and simpler techniques, such as the guides, should be taken through to clinical trials. Especially when considering the balance between surgical time and accuracy, PSI appears to offer the prospect of a real alternative in providing clinically relevant precision for UKAs and possibly at a lower cost due to the reduced instrumentation and inventory. 6. Conflict of interest None to declare. Acknowledgments We thank the following surgeons for making the time to participate in this study: Mr. D. Nathwani, Mr. K. Davda and Mr. A. Anand. Their efforts and the efforts of the authors that also took part in carrying out the unicompartmental knee arthroplasties are greatly appreciated. The support from everyone else at the musculoskeletal lab at the Charing Cross Hospital has been invaluable and we are especially grateful for the guidance given by Mr. A. Aqil and Dr. R. Abel. References [1] Pearle AD, Kendoff D, Musahl V. Perspectives on computer-assisted orthopaedic surgery: movement toward quantitative orthopaedic surgery. J Bone Joint Surg Am 2009;91(Suppl. 1):7–12. [2] Lonner JH. Indications for unicompartmental knee arthroplasty and rationale for robotic arm-assisted technology. Am J Orthop 2009;38(2 Suppl.):3–6. [3] 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;18(7 Suppl. 1):2–8. [4] Hernigou P, Deschamps G. Posterior slope of the tibial implant and the outcome of unicompartmental knee arthroplasty. J Bone Joint Surg Am 2004;86-A(3):506–11. [5] Hernigou P, Deschamps G. Alignment influences wear in the knee after medial unicompartmental arthroplasty. Clin Orthop Relat Res 2004;423:161–5. [6] Willis-Owen CA, Brust K, Alsop H, Miraldo M, Cobb JP. Unicondylar knee arthroplasty in the UK National Health Service: an analysis of candidacy, outcome and cost efficacy. Knee 2009;16(6):473–8. [7] Amin AK, Patton JT, Cook RE, Gaston M, Brenkel IJ. Unicompartmental or total knee arthroplasty?: Results from a matched study. Clin Orthop Relat Res 2006;451:101–6. [8] Koskinen E, Eskelinen A, Paavolainen P, Pulkkinen P, Remes V. Comparison of survival and cost-effectiveness between unicondylar arthroplasty and total knee arthroplasty in patients with primary osteoarthritis: a follow-up study of 50,493 knee replacements from the Finnish Arthroplasty Register. Acta Orthop 2008;79(4):499–507. [9] Davies BL, Rodriguez y Baena FM, Barrett ARW, Gomes MPSF, Harris SJ, Jakopec M, et al. Robotic control in knee joint replacement surgery. Proc Inst Mech Eng H 2007;221(1):71–80. [10] Lang JE, Mannava S, Floyd AJ, Goddard MS, Smith BP, Mofidi A, et al. Robotic systems in orthopaedic surgery. J Bone Joint Surg Br 2011;93(10):1296–9. [11] Cobb J, Henckel J, Gomes P, Harris S, Jakopec M, Rodriguez F, et al. Hands-on robotic unicompartmental knee replacement: a prospective, randomised controlled study of the acrobot system. J Bone Joint Surg Br 2006;88(2):188–97.
434
Z. Jaffry et al. / The Knee 21 (2014) 428–434
[12] Koeck FX, Beckmann J, Luring C, Rath B, Grifka J, Basad E. Evaluation of implant position and knee alignment after patient-specific unicompartmental knee arthroplasty. Knee 2011;18(5):294–9. [13] Fitz W. Unicompartmental knee arthroplasty with use of novel patient-specific resurfacing implants and personalized jigs. J Bone Joint Surg Am 2009;91(Suppl. 1):69–76. [14] Schlueter-Brust K, Bontemps G, Sobottke R, Röllinghoff M, Michael JWP, Siewe J, et al. The future of surgical orthopaedics of the knee. Proc Inst Mech Eng H 2010;224(6):729–34. [15] Biomet. Oxford partial knee-manual of surgical technique. Available at http://www. biomet.nl/resource/5980/Oxford-Knee-Optec.pdf. [Accessed May 4, 2012]. [16] Barrett A. Plastic bone testing report-Acrobot Sculptor Version 2.0.1. London: The Acrobot Company Limited; 2010.
[17] Karia M, Masjedi M, Andrews B, Jaffry Z, Cobb J. Robotic assistance enables inexperinced surgeons to perform unicompartemental knee arthroplasties on dry bone models with accuracy superior to conventional methods. Adv Orthop 2013;2013. http://dx.doi.org/10.1155/2013/481039 [Article ID 481039]. [18] Masjedi M, Jaffry Z, Harris S, Cobb J. Protocol for evaluation of robotic technology in orthopedic surgery. Adv Orthop 2013;2013. http://dx.doi.org/10.1155/2013/194683 [Article ID 194683]. [19] Bargar WL. Robots in orthopaedic surgery: past, present, and future. Clin Orthop Relat Res 2007;463:31–6. [20] Yen PL, Davies BL. Active constraint control for image-guided robotic surgery. Proc Inst Mech Eng H 2010;224(5):623–31.
Contents lists available at ScienceDirect
The Knee
Unicompartmental knee arthroplasties: Robot vs. patient specific instrumentation☆ Zahra Jaffry, Milad Masjedi ⁎, Susannah Clarke, Simon Harris, Monil Karia, Barry Andrews, Justin Cobb MSk Lab, Department of Orthopaedics, Imperial College London, Charing Cross Hospital, Fulham Palace Road, London W6 8RF, UK
a r t i c l e
i n f o
Article history: Received 15 August 2013 Received in revised form 28 October 2013 Accepted 21 November 2013 Keywords: Osteoarthritis Unicompartmental knee replacement Surgical technique Surgical accuracy Three dimensional analysis
a b s t r a c t Background: The technical reliability demonstrated by semi active robots in implant placement could render unicompartmental knee arthroplasties (UKAs) more favourable than they are currently. The relatively untested method using patient specific instrumentation (PSI), however, has the potential to match the accuracy produced by robots but without the barriers that have prevented them from being used more widely in clinical practice, namely operative time. Therefore this study took a step towards comparing the accuracy and time taken between the two technologies. Methods: Thirty-six UKAs were carried out on identical knee models, 12 with the Sculptor, 12 with PSI and 12 conventionally under timed conditions. Implant placement in these knees was then judged against that in a pre-operative plan. Results: Tibial implant orientations and femoral implant positions and orientations were significantly more accurate in the PSI group with mean errors of 6°, 2 mm and 4° respectively, than the conventional group which had means of 9°, 4 mm and 10°. There was no significant difference between the robot and PSI generally except in tibial implant orientation (mean robotic error 3°) and tibial implant position did not vary significantly across all three groups. It was also found that use of PSI and conventional methods took half the time taken by the robot (p b 0.001). Conclusions: With further development, PSI can match and possibly surpass the accuracy of the robot, as it does with the conventional method, and achieve planned surgery in less time. Clinical relevance: This work sets the foundation for clinical trials involving PSI. © 2013 Elsevier B.V. All rights reserved.
1. Introduction Surgical precision often influences clinical outcomes in technically demanding procedures such as unicompartmental knee arthroplasties (UKAs) [1,2]. Limitations of the technique make ideal implant placement difficult [3] affecting limb alignment and hence the levels of wear, loosening and other reasons for failure and revision [4,5]. Therefore, despite the facts that UKAs allow greater bone conservation, faster functional recovery and are more cost-effective [6], the alternative of a total knee arthroplasty is currently considered the more reliable approach due to higher ‘survival rates’ [7] and perceived lower revision costs [8]. The technology developed to address this problem are semi-active robotic systems, like the Sculptor Robotic Guidance Arm (Stanmore Implants Worldwide, Elstree, UK), which combine elements of both
☆ All authors are affiliated with Imperial College London. ⁎ Corresponding author. Tel.: +44 20 3313 8833; fax: +44 20 3313 4427. E-mail address: [email protected] (M. Masjedi). 0968-0160/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.knee.2013.11.017
navigation and fully autonomous systems. Using pre-operative planning, they offer surgeons constraint outside of a pre-set zone of safety rather than simply providing them with the information to determine this zone themselves [9,10]. The results of UKAs done in patients by the predecessor to the Sculptor have been promising with all tibiofemoral alignment angles being within 2° of the original plan [11]. Yet, the tool has not been widely adopted. The recent introduction of rapid prototyped cutting guides made to match the anatomy of the patient provides an alternative method of carrying out the same task in less time. Also using a Computer Tomography (CT) or Magnetic Resonance (MR) derived plan [12], these cutting guides, or Patient Specific Instrumentation (PSI), achieve results without the use of a robotic system offering a simpler embodiment of many steps involved in the procedure [13,14]. This may lower the surgical time taken but it is not clear whether the precision of this technology can contest that of the robots. This study therefore set out to investigate the Sculptor Robotic Guidance Arm as an example of a semi active robot, PSI (Embody, London, UK) and conventional instrumentation and aimed to answer the following question: which of these are able to deliver the standard of accuracy required for consistent clinical results in UKA surgery and in what time?
Z. Jaffry et al. / The Knee 21 (2014) 428–434
2. Materials Dry bone models used in the study were produced by Medical Models Ltd. (Bristol, UK) in a single batch from the same mould, the ‘Imperial LB Varus Knee’. The model was based on a CT scan of a patient with medial compartment osteoarthritis. Each was composed of a femur and tibia held together by cords that simulated ligaments, a plastic patella and an outer capsule that surrounded the bones. 3. Methods Digital Imaging and Communications in Medicine (DICOM) files from a CT scan of the dry bone models were used to create a Stereolithograph (STL) file with Modeller (Stanmore Implants Worldwide, Elstree, UK) and a senior surgeon planned the operation on Planner (Stanmore Implants Worldwide, Elstree, UK). A Uniglide size 4 tibial base plate and size 3 femoral component from the mobile bearing range (Corin, Cirencester, UK) were considered appropriate to use with the bone models. The pre-operative plan was designed to restore the natural joint line with implant positions and orientations that could realistically be achieved with the conventional cutting guides and alignment rods accompanying the Uniglide set. Hence, bias against the technique was
429
avoided. This involved placing the tibial component with a posterior slope of 7° (the angle at which the Corin instrument set aims to cut the tibia), medial slope of 0° and axial rotation of − 1° in relation to tibial frames of reference. Corresponding values for the femoral component were placed at 5°, 0° and 7°. The plan was then loaded onto the Sculptor to be used by the robotic group of the study. Images of the plan, including the values mentioned previously, were also printed out for surgeons to prevent discrepancies between them and define what should be achieved with the instrumentation. Guides constituting the PSI were designed on computer aided design software (Solidworks, Waltham, Massachusetts and Rhinoceros, Barcelona, Spain) using the same 3D model generated by Modeller and the STL files from Planner. These were then produced by the Objet Eden250 3D Printing System (Objet, Rehovot, Israel) in a medical grade polymer (Objet MED610). Three experienced users of the Sculptor, having performed the procedure over 20 times, were recruited. The bone model was attached to a frame that hangs from a table to replicate the position of the knee in surgery. The timer was started once the robot had been calibrated and after the incision into the capsule surrounding the bone had been made. The tracking arm was pinned to the tibia and the computer models were registered to the physical bone using the tip of the cutting burr arm, ensuring the root mean square (RMS) error in registration was less than
Fig. 1. Photographs showing the experimental setup including the bone model (A) and use of the Sculptor (B), Patient Specific Instrumentation (C and D) and conventional instruments (E and F).
430
Z. Jaffry et al. / The Knee 21 (2014) 428–434
Fig. 2. Diagrams showing:(A) The positions of planned (grey) and achieved (blue) tibial implants; (B) The points 1–4 that were compared between the two implants by creating local frames of reference for each where the origin is the intersect of the x, y, z axes; (C) The positions of planned (grey) and achieved (blue) femoral implants; (D) The points 1–4 that were compared between the two femoral implants by creating local frames of reference for each where the origin is point 4.
Table 1 Compound translational and rotational error values in 36 unicompartmental knee arthroplasty implant placements. Compound translational error (mm)—tibial component Sculptor Mean 2 SD 1 (min–max) (1–3)
PSI
Conventional
Sculptor
PSI
Conventional
2 1 (1–3)
3 1 (1–4)
2 1 (1–5)
2 1 (1–4)
4 2 (2–8)
Compound rotational error (º)—tibial component Sculptor PSI Mean 3 SD 1 (min–max) (2–5)
Compound translational error (mm)—femoral component
Compound rotational error (º)—femoral component
Conventional Sculptor PSI
6 9 3 3 (3–11) (3–14)
3 1 (2–5)
Conventional
6 10 3 5 (3–11) (4–22)
1 mm and the maximum error less than 2 mm. Once the tibial cut was made by the high speed burr, the same sequence of steps was carried out for the femur. The timer was stopped after nibblers had been used to remove any excess over-hanging bone. Between the three participants, a total of 12 UKAs were done in this way. Each surgeon was then also asked to complete four UKAs using the PSI and four using the conventional guides and instruments provided by Corin in a random order. PSI involved drilling the guides into place for the main cuts and later steps also used conventional instruments. As previously, the timer for each procedure was started after the incision into the capsule was made. The timer was stopped after the final keel cuts. A total of 12 were done with the PSI and 12 conventionally. The experimental design for each technique is illustrated in Fig. 1. From this point all cut bones were treated in the same manner. The capsule surrounding the model was removed and the tibia and femur were separated by cutting the cords between them. The Corin implants
Table 2 p-values of differences in unicompartmental knee arthroplasty implant placements between groups. Tibial component
Femoral component
Translational error (mm) Sculptor Sculptor PSI Conventional
PSI 0.995
0.995 0.840
Conventional 0.840 0.680
0.680
Sculptor Sculptor PSI Conventional
0.991 0.004
PSI
Conventional
0.991
0.004 0.030
0.030
Rotational error (°) Sculptor Sculptor PSI Conventional
0.014 b0.001
PSI
Conventional
0.014
b0.001 0.092
0.092
Sculptor Sculptor PSI Conventional
1.000 0.002
PSI
Conventional
1.000
0.002 0.002
0.002
Z. Jaffry et al. / The Knee 21 (2014) 428–434
were coated in matt enamel paint to ensure correct measurement of the laser spot from the 3D scanner. Implants were then placed in the bones and each bone was individually loaded and scanned using the NextEngine Desktop 3D Scanner, Model 2020i (NextEngine, Santa Monica, California). The scans were saved as STL files that could be
431
matched onto the 3D STL model of the plan using 3matic software (Materialise, Leuven, Belgium). Once the bones had been matched, four identical co-ordinates were selected on the implant for both planned and achieved components. Using Matlab software (MathWorks, Natick, Massachusetts), these co-
Fig. 3. Box plots showing range of translational error for the tibial and femoral components in the lateral–medial (A), anterior–posterior (B) and distal–proximal (C) directions in Sculptor, Patient Specific Instrumentation and Conventional groups. The reference lines in each indicate an acceptable range in which the error can lie.
432
Z. Jaffry et al. / The Knee 21 (2014) 428–434
ordinates were used to construct a local frame of reference and aligned with anatomical frames. The origins of tibial and femoral components were the average of four set points on the implant and point 4 respectively (Fig. 2). The x axis points laterally, y anteriorly and the z axis is perpendicular to the x and y axes, pointing inferiorly. The position and orientation of the implant were then compared to that of the planned implant in all six degrees of freedom. To validate the use of the 3D laser scanner, an uncut tibia was scanned and matched to the 3D model from Modeller to test it against
a ‘gold’ standard scanning technique- CT. The difference between models was found by finding the average distance between two corresponding points on the two bone models in three different places. Five scans of the same bone were taken and the difference in implant position between each scan was established. This process was replicated by another investigator so both intra- and inter-observer reliability were measured. The magnitude of error (compound error) in both position and orientation of implants was found using the sum of the RMS values of
Fig. 4. Box plots showing range of rotational error for the tibial and femoral components in the flexion–extension (A), varus–valgus (B) and axial rotation (C) directions in Sculptor, Patient Specific Guide and Conventional groups. The reference lines in each indicate an acceptable range in which the error can lie.
Z. Jaffry et al. / The Knee 21 (2014) 428–434
error in the translational (lateral–medial, anterior–posterior and distal– proximal) and rotational (flexion–extension, varus–valgus, axial rotation) directions respectively for all groups. An analysis of variance (ANOVA) test was carried out using SPSS statistical analysis software (IBM, Armonk, New York) to identify the most accurate technique for carrying out a UKA. Post-hoc analyses provided further information on the data and p-values less than 0.05 were considered significant (α = 0.05, β = 0.2). Another ANOVA test was carried out to compare the average times taken to complete a procedure between groups. Finally, the Bland and Altman statistical technique was used to determine the level of agreement between repeat measurements of the implant position in the same bone and among two different observers. 4. Results The Bland–Altman plots demonstrated that all points fell within two standard deviations of the mean difference indicating the reliability of the method. Implant placement accuracy is compared across groups in terms of the mean compound translational and rotational errors in Table 1. Conventional instruments delivered substantially less accuracy than both the Sculptor and the PSI, with Sculptor and PSI groups performing comparably in all but tibial implant orientation, where the Sculptor was more accurate. There was no significant difference across groups in tibial implant position. The p-values of differences between groups are shown in Table 2. The results are expanded in Figs. 3 and 4 which show the range of all three groups in six degrees of freedom for both the tibial and femoral components. This allows an idea of the precision achieved and the reference lines indicate an acceptable range in which implant placement should be [15]. Most Sculptor and PSI plots fall within these lines while ten of the twelve conventional plots fall outside of the accepted error range; the ones to notice are those where the median falls out of this range, which is never the case for the two other groups. Conventional instruments produced a wider range in all variables, most notably in tibial axial rotation (Fig. 4C), where the range was ±9°, compared with ±5° for PSI and ±3° for the robot. The mean time taken to complete a procedure with the Sculptor (20 min and 25 s ± 5 min 31 s) was almost double that of PSI (7 min and 52 s ± 2 min 1 s) and conventional (10 min and 31 s ± 3 min 19 s) groups with an overall p value of less than 0.001. No significant difference was found between PSI and conventional groups (p = 0.232).
5. Discussion This study investigated three techniques for UKA surgery and compared the implant placement accuracy and operative time taken as a step towards improving clinical outcomes. Surgical precision has been determined by differences between the planned and achieved implant position and orientation [9]. While there is no consensus on the level of accuracy actually needed for clinically satisfactory results, there is evidence suggesting that translational or rotational error in prosthesis placement, i.e. greater than a 7° posterior slope (in flexion–extension) in the tibial component [4] or any other error resulting in angular errors of more than 5° in the knee [5], may lead to poorer outcomes [15]. The plan was made in accordance with this and reference lines in Figs. 3 and 4 indicating an ideal range in which the plan should be achieved allowed easy evaluation of results and comparison between groups. Personal communication with Corin ensured that these ranges were the same as the ones stated by the Oxford knee manual used for this study. These values have previously been adopted by studies that had similar methods in their use of plastic dry bones, which eventually lead to the Sculptor Report [16]. Reliability of the method was ensured in a number of other ways. The use of a laser scanner (rather than CTconsidered the gold standard technique) had been validated in this study using intra- and inter-observer reliability tests. Matches of 3D models from CT images and the laser scanner were also compared on 3matic showing a maximum difference of 1 mm between models for both the tibia and femur. The use of laser scanning over CT as well as the use of the Sculptor software has also been validated in other papers [17,18]. Thirty-six UKAs being carried out with three techniques and three surgeons can be argued to introduce confounding variables and past literature has provided a standard deviation with which to carry out
433
power calculations suggesting the use of a larger number [11]. However, less variability was expected due to the use of identical bone models rather than patients and the use of a relatively novel technique, i.e. the PSI, in addition to the robot discussed in these previous papers. Nevertheless, the methods, including the bone models used, were realistic, served the purpose of the study well and produced results with the robot that were comparable to its use elsewhere [11,16]. As expected, implant placement was less predictable in the conventional group than the other two. PSI was shown to be only less accurate in tibial implant orientation compared to the robot and with the significantly lower time spent on the procedure, the potential PSI has to surpass the Sculptor in its efficiency with further development is highlighted. Although, it is important to note at this point that the shorter operative time does come at the price of longer planning time. As a small study on dry bones, conclusions now need to be enforced with further investigations in patients. It is important to continually test efficacy to keep track of progress in the trend towards minimally invasive surgery [19] and form foundations for cost-effectiveness analyses [6]. Between the time of previously mentioned studies and this one, the robot alone has undergone many improvements including a reduction in size, cost and operating system [20]. More importantly, newer and simpler techniques, such as the guides, should be taken through to clinical trials. Especially when considering the balance between surgical time and accuracy, PSI appears to offer the prospect of a real alternative in providing clinically relevant precision for UKAs and possibly at a lower cost due to the reduced instrumentation and inventory. 6. Conflict of interest None to declare. Acknowledgments We thank the following surgeons for making the time to participate in this study: Mr. D. Nathwani, Mr. K. Davda and Mr. A. Anand. Their efforts and the efforts of the authors that also took part in carrying out the unicompartmental knee arthroplasties are greatly appreciated. The support from everyone else at the musculoskeletal lab at the Charing Cross Hospital has been invaluable and we are especially grateful for the guidance given by Mr. A. Aqil and Dr. R. Abel. References [1] Pearle AD, Kendoff D, Musahl V. Perspectives on computer-assisted orthopaedic surgery: movement toward quantitative orthopaedic surgery. J Bone Joint Surg Am 2009;91(Suppl. 1):7–12. [2] Lonner JH. Indications for unicompartmental knee arthroplasty and rationale for robotic arm-assisted technology. Am J Orthop 2009;38(2 Suppl.):3–6. [3] 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;18(7 Suppl. 1):2–8. [4] Hernigou P, Deschamps G. Posterior slope of the tibial implant and the outcome of unicompartmental knee arthroplasty. J Bone Joint Surg Am 2004;86-A(3):506–11. [5] Hernigou P, Deschamps G. Alignment influences wear in the knee after medial unicompartmental arthroplasty. Clin Orthop Relat Res 2004;423:161–5. [6] Willis-Owen CA, Brust K, Alsop H, Miraldo M, Cobb JP. Unicondylar knee arthroplasty in the UK National Health Service: an analysis of candidacy, outcome and cost efficacy. Knee 2009;16(6):473–8. [7] Amin AK, Patton JT, Cook RE, Gaston M, Brenkel IJ. Unicompartmental or total knee arthroplasty?: Results from a matched study. Clin Orthop Relat Res 2006;451:101–6. [8] Koskinen E, Eskelinen A, Paavolainen P, Pulkkinen P, Remes V. Comparison of survival and cost-effectiveness between unicondylar arthroplasty and total knee arthroplasty in patients with primary osteoarthritis: a follow-up study of 50,493 knee replacements from the Finnish Arthroplasty Register. Acta Orthop 2008;79(4):499–507. [9] Davies BL, Rodriguez y Baena FM, Barrett ARW, Gomes MPSF, Harris SJ, Jakopec M, et al. Robotic control in knee joint replacement surgery. Proc Inst Mech Eng H 2007;221(1):71–80. [10] Lang JE, Mannava S, Floyd AJ, Goddard MS, Smith BP, Mofidi A, et al. Robotic systems in orthopaedic surgery. J Bone Joint Surg Br 2011;93(10):1296–9. [11] Cobb J, Henckel J, Gomes P, Harris S, Jakopec M, Rodriguez F, et al. Hands-on robotic unicompartmental knee replacement: a prospective, randomised controlled study of the acrobot system. J Bone Joint Surg Br 2006;88(2):188–97.
434
Z. Jaffry et al. / The Knee 21 (2014) 428–434
[12] Koeck FX, Beckmann J, Luring C, Rath B, Grifka J, Basad E. Evaluation of implant position and knee alignment after patient-specific unicompartmental knee arthroplasty. Knee 2011;18(5):294–9. [13] Fitz W. Unicompartmental knee arthroplasty with use of novel patient-specific resurfacing implants and personalized jigs. J Bone Joint Surg Am 2009;91(Suppl. 1):69–76. [14] Schlueter-Brust K, Bontemps G, Sobottke R, Röllinghoff M, Michael JWP, Siewe J, et al. The future of surgical orthopaedics of the knee. Proc Inst Mech Eng H 2010;224(6):729–34. [15] Biomet. Oxford partial knee-manual of surgical technique. Available at http://www. biomet.nl/resource/5980/Oxford-Knee-Optec.pdf. [Accessed May 4, 2012]. [16] Barrett A. Plastic bone testing report-Acrobot Sculptor Version 2.0.1. London: The Acrobot Company Limited; 2010.
[17] Karia M, Masjedi M, Andrews B, Jaffry Z, Cobb J. Robotic assistance enables inexperinced surgeons to perform unicompartemental knee arthroplasties on dry bone models with accuracy superior to conventional methods. Adv Orthop 2013;2013. http://dx.doi.org/10.1155/2013/481039 [Article ID 481039]. [18] Masjedi M, Jaffry Z, Harris S, Cobb J. Protocol for evaluation of robotic technology in orthopedic surgery. Adv Orthop 2013;2013. http://dx.doi.org/10.1155/2013/194683 [Article ID 194683]. [19] Bargar WL. Robots in orthopaedic surgery: past, present, and future. Clin Orthop Relat Res 2007;463:31–6. [20] Yen PL, Davies BL. Active constraint control for image-guided robotic surgery. Proc Inst Mech Eng H 2010;224(5):623–31.
