Off-the-shelf knee replacements versus robotics versus customized implants: the reality versus the hype

Abstract

With traditional ‘off-the-shelf’ implants for knee replacement surgery, 15–20% of patients end up unhappy with their knee. At least part of the reason for this may be the significant differences that can exist between individuals in their distal femoral and proximal tibial geometries. This article explains how patient-specific customized knee implants can address many of the deficiencies that may exist with off-the-shelf designs. Evidence is presented of how customized knee implants can lead to better alignment, better coverage, better knee kinematics, less bone resection, improved efficiency in theatre, fewer complications, better patient-reported outcomes and increased long-term prosthetic survivorship, all without any significant overall additional costs. This article also dispels some of the myths and the hype that surround the current trend towards robotic-assisted knee replacement surgery.

What is the problem with knee replacements?

If hip replacement surgery could be described as an excellent operation, then maybe knee replacement could be said to be ‘a pretty good operation’. The problem is that an excellent outcome from knee arthroplasty is far from 100% guaranteed.

Apart from all the pain, the time off work, the extremely difficult rehab that’s required, and the not-insignificant potential risks of knee replacement surgery, there remains a percentage of patients who are just not happy with their outcome. This is true even in the absence of any specific discernible complications. It is generally accepted that about 80–85% of patients are satisfied with their outcome after knee arthroplasty, which means that 15–20% are not. In the UK alone, where the annual number of knee replacements performed is in the region of approximately 100,000, this equates a total of 15,000 to 20,000 unhappy patients every single year. If one broadly accepts the general statement that a happy patient tells maybe one or two people about their good outcome, whereas an unhappy patient tells 10 or 20, it is easy to see how knee replacement surgery has perhaps a less than ideal reputation amongst many people, including general practitioners and physiotherapists, and, therefore, potential patients.

In general, a ‘standard’ knee replacement prosthesis is designed in a range of sizes (e.g. A, B, C etc), but each model is exactly the same shape. In terms of size, if one is in the middle of the ‘size A range’ or the middle of the ‘size B range’, then you will get a knee that matches your size closely. However, if you are halfway between size A and size B, then you will end up with a knee prosthesis that is either a bit too big or a bit too small for your knee. If a prosthesis is too small, this leads to reduced coverage of the bone ends, potentially resulting in more post-operative bleeding, and it also means that the prosthesis might be resting on softer, poorer-quality bone, which may increase the potential risk of subsidence and loosening of the implant. If a prosthesis is too large, then this can result in over-hang, and over-hang by as little as just 3 mm has been shown to be associated with an increased risk of soft-tissue impingement and pain after knee arthroplasty.

Similarly, with respect to shape, with a standard knee replacement, the designers tend to look at a range of different patients’ knees, with them then creating a ‘best fit’ design that most closely matches what they perceive to be an ‘ideal’ knee. The reality, however, is that not only do people’s knees vary significantly in terms of size, but they also vary considerably in terms of shape and contours ( Figure 1 ). In some people, the size of their lateral femoral condyle might fairly closely match that of their medial femoral condyle, and hence their distal femoral condylar line may be reasonably horizontal. In others, the lateral femoral condyle may be significantly smaller than the medial side, leading to quite an oblique distal femoral condylar line, in significant valgus. Equally, the radii of curvature of the lateral femoral condyle may be wide or narrow, and hence there can be significant variation in the posterior condylar angle. Meier et al. did a study using over 24,000 CT scans and demonstrated a high variability of distal femoral geometry, in terms of relative widths of the medial versus lateral condyles, distal condylar offset and posterior condylar offset, and they showed that had a standard off-the-shelf prosthesis been used for all patients, then in approximately 25% of cases there would have been a 3 mm or more mismatch in sizing.

In addition, the size, shape and posterior slope of the proximal tibia also varies significantly between individuals. Nunley et al. analysed data from 2395 knee CTs and showed “extreme” variation in posterior tibial slope, with the angle varying from −5.6° to 16.9°. They concluded that if a unicompartmental partial knee replacement prosthesis was implanted in all cases with a standard posterior tibial prosthesis slope of 5° to 7 ^o , then in 47% of cases this would produce a slope that was significantly less than the patient’s original anatomy.

To complicate matters further, all of the various measurements of shape that can be taken of a knee seem to vary independently of each other.

What does this all mean? This means that if a knee prosthesis closely matches the size, shape and contours of your knee, then you are simply lucky. However, if, by random chance, your knee varies from the fixed design of whatever prosthesis your surgeon has decided to use, then you might be unlucky, and there might be a significant mismatch between the anatomy of your original knee and the characteristics of the prosthesis that you receive. This might well potentially explain at least a proportion of the unhappy patients that are seen after knee replacement surgery.

Is ‘kinematic alignment’ the answer?

As a result of the inadequacies of existing off-the-shelf knee implants, surgeons have advocated newer ‘alternative’ alignment strategies in an attempt to better recreate pre-disease anatomy. A careful review of the origins of the literature on ‘kinematic alignment’ of knee replacement prostheses reveals that this technique was originally introduced to try to solve the problems specifically caused by the use of the OtisMed knee prosthesis system. With this, the femoral prosthesis had symmetrical single-radius-of-curvature medial and lateral femoral condyles. To try to correct for this anatomical aberrance, the lead surgeon promoting this prosthesis advocated placement of the tibial prosthesis into varus, to avoid over-stuffing the lateral compartment. It could be said that this technique was therefore spawned as a ‘fudge factor’ simply to try and compensate for an inappropriately designed prosthesis.

The solution: custom-made knees?

Since 2011 in the USA, and 2012 in the UK, customized individually-manufactured knee replacement prostheses have been available. These customized knee prostheses were introduced by a company called Conformis, which has, more recently, been taken over by another American company called ‘restor3d’. The aim of the customized restor3d/Conformis prostheses is to provide an implant that specifically matches the size, shape and contours of the individual patient’s own native knee before it became diseased, in an attempt to improve upon outcomes.

Alignment and coverage

It is unclear as to the exact extent that malalignment of a knee prosthesis might affect outcomes. In terms of patient-reported outcome measures (PROMs), it would appear that good clinical results can be obtained even in the presence of what may, radiographically, appear to be quite a poorly implanted knee. Malalignment can, however, affect knee kinematics, and it can also affect the amount of loading through each compartment of a knee, which may have negative effects in terms of wear and longer-term prosthetic survivorship. Surgeons have traditionally set a target to implant the device in neutral mechanical alignment, and generally within ± 3 ^o from a neutral mechanical axis is considered acceptable. There are newer alignment techniques (kinematic alignment, reverse kinematic alignment, etc) and ongoing debate as to what the target alignment should be. Regardless, there are a number of potential approaches available for trying to ensure that the targeted prosthetic alignment is achieved consistently and reliably.

Computer navigation for knee replacement surgery was first introduced in 1997, and it consists of an optical localizer and arrays mounted with light-emitting diodes, linked to a computer. Although the technology has evolved significantly, the basic principles remain the same, with computer navigation allowing for accurate placement of cutting blocks, which are utilized to help guide the bone cuts. A significant evolution from computer navigation was the introduction, approximately 10 years ago now, of robotic-assisted knee replacement, where robotic arms are used to make the bony cuts. Both of these techniques have been shown to improve accuracy of implant alignment; however, both of these approaches carry a not-insignificant list of potential negatives, including:

•

a learning curve for surgeons,
•

significant additional up-front costs,
•

the use of diaphyseal pins, with the propensity for pin site complications (e.g. fractures or infections) and
•

prolonged operating times.

The use of patient-specific instrumentation (PSI) has been developed as a tool for improving prosthesis placement without all of the various negatives of computer navigation or robotics.

With PSI, 3D-printed nylon patient-specific cutting blocks are manufactured based on preoperative CT scan data. The advantages of PSI are that it can be used as a highly accurate aid to guide the bone cuts required for knee arthroplasty surgery, and it can reduce operating times without the need for additional equipment (and very significant additional costs) in the operating theatre ( Figure 2 ).

In a study of 232 patients undergoing total knee replacement (TKR) surgery, 125 patients received a patient-specific prosthesis with PSI (iTotal-CR, Conformis), and these were compared to 107 patients receiving a conventional TKR with standard cutting blocks (Triathlon, Stryker). Pre- and postoperative alignment was measured on long-leg weight-bearing radiographs, which showed that the rate of ± 3° outliers from neutral limb axis was 16% in the patient-specific TKR group, but 26% in the conventional TKR group. In a similar study, Ivie et al. found that patients undergoing TKR with PSI were 1.8 times more likely to end up with alignment within 3 ^o of neutral mechanical alignment compared to those undergoing TKR with intramedullary instrumentation.

In addition to alignment in the coronal plane, appropriate placement in the sagittal plane and correct rotational alignment are also important factors.

As already noted above, there can be considerable variation in distal femoral geometry between individuals, and there can also be ‘extreme’ variability in the posterior tibial slope. During knee arthroplasty surgery with an off-the-shelf prosthesis, it can be difficult to achieve good fit and coverage of the tibia whilst at the same time maintaining good rotational alignment. In addition to the risk of prosthetic overhang causing soft tissue impingement and, hence, pain, under-hang also results in a prosthesis resting on bone that may have suboptimal strength, and leaving exposed areas of cut bone may lead to increased postoperative bleeding in a knee.

In a study by Schroeder and Martin, tibial coverage and component rotation were assessed intraoperatively in 44 TKR cases, comparing three different off-the-shelf prostheses (Biomet Vanguard, Zimmer NexGen and DePuy Sigma) to a customized implant design (iTotal, Conformis). The authors showed that in 18% of cases with an off-the-shelf tibial prosthesis there was component overhang of 3 mm or more, compared to zero cases with the customized implant design. The off-the-shelf implants also showed significantly greater underhang of 3 mm or more, compared to the customized implant design (37% vs 18%). Furthermore, 45% of the off-the-shelf prostheses showed a rotational deviation of more than 5 ^o , and 4% more than 10 ^o , compared to no deviation with any of the customized implants ( Figure 3 ).

Kinematics

In tandem with one aiming for appropriate alignment of a knee prosthesis, a further aim of knee replacement surgery should be to try to replicate, as closely as possible, the natural kinematics of the knee. In a cadaveric study by Patil et al., the kinematics of the knee were studied comparing a knee replaced with a customized prosthesis implanted with PSI versus a standard off-the-shelf prosthesis implanted with standard cutting jigs. The authors showed that the deviation from normal was significantly less with the customized prosthesis for active femoral rollback, active tibiofemoral adduction and for passive varus/valgus laxity.

The results of the cadaveric studies have been borne out in in vivo studies using mobile dynamic fluoroscopy to assess knee kinematics with a weight-bearing deep knee bend and with rising from a seated position, evaluating weight-bearing range of motion, femorotibial translation, femorotibial axial rotation and condylar lift-off occurrence. Patients having customized patient-specific implants were compared to those receiving a standard off-the-shelf prosthesis. With this, it has been shown that the customized implants gave a better range of weight-bearing knee flexion and consistently more posterior femoral rollback compared to the standard knees. The customized implants also gave greater axial rotation, and only the customized implants demonstrated femoral internal rotation at full extension, as in a normal knee. The customized implants also showed better stability in early to mid-flexion, and in conclusion, the customized implants were demonstrated to exhibit kinematics more similar to a normal knee.

Volume of bone resected

The more bone that is resected at the time of primary arthroplasty surgery, the less bone will be left in the future for any potential revision surgery, and bone loss is one of the key factors negatively impacting the ease of and the outcomes from revision knee replacement surgery. Furthermore, the more bone is preserved, the greater the bone-cement interface will be, and hence the better the fixation of the prosthesis is likely to be.

In a study by Kurtz and Slamin, the volume of bone resected at the time of knee replacement surgery was significantly less with the use of a customized implant compared to a standard prosthesis, with a 12–49% greater volume of bone being resected when a standard off-the-shelf prosthesis was used.

Efficiency in the OR

Instrument stock and preparation of instrumentation can be a significant burden in an orthopaedic theatre department. With standard knee replacements from most manufacturers, five to seven separate trays of instruments are required, with multiple different sized trials and prostheses having to be available, all of which have to be stored and sterilized, and then cleaned, re-sterilized and re-stored after each case. It has been estimated that the cost of this alone is in the region of about £800 per case, which is a hidden cost that few hospitals have the ability to measure accurately. As an approximate estimate, in a standard knee replacement case over 350 instruments are provided, of which only about 50 may actually be used.

By comparison, with the system used for customized implants from Conformis, the surgeon is provided with just one single set of prostheses (the prosthesis that is already known, in advance, will actually fit and that which is actually right for that patient), and a set of custom-made 3D-printed nylon cutting blocks that, again, are designed and made for that specific individual patient, and which are disposable. This all comes in just one single box, and the surgery can be performed with the scrub staff having just two trolleys: one for the custom-made cutting blocks and one for general orthopaedic instrumentation. This very significantly reduces the amount of equipment needed per case ( Figure 4 ), and significantly improves overall efficiency in the operating theatre.