Computer assisted knee replacements

In the last decade strides have been made to improve the reproducibility of the cuts involved in knee replacement surgery. The idea here was to use a computer to register all the main landmarks in a limb and then recommend to the surgeon the best cuts to be made. The concept, while laudable, failed to incorporate various other factors that are important in knee replacement surgery chiefly the experience of the surgeon and ultimately was not shown to be superior to a knee replacement surgeon with experience. Since that time the technology has branched into various other fields like robotic surgery and most recently partial knee replacement surgery (eg. MAKOplasty). While none of these have the track record of traditional knee replacement surgery, they're introduction has resulted in a better understanding of the patho-mechanics of the knee afflicted with arthritis.

At the very core of the understanding of computer-aided knee replacement surgery is the understanding of the various axes. These were all based on a mostly white population of normal patients examined in the West and increasingly knee replacement surgeons are beginning to appreciate the differences that patients in the Asia Pacific region pose (this includes the patients of the Australasian regions). Nevertheless, these indices are well known to any orthopaedic surgeon and useful for reference:
  1. The mechanical axis (Figure 1): A line from the centre of the head of the femur, to just medial of the knee to the center of the ankle is the weight bearing axis of the lower limb. It is 3 degrees out from the line of drop of the centre of gravity
  2. The anatomical axis of the femur (Figure 2) is 6 degrees out from the mechanical axis and 9 degrees from the line of the center of gravity and is measured from the piriform fossa to the just medial of the centre of the knee.. The mechanical axis of the tibia should be in line with its anatomical axis.
Navigation then attempts to make cuts perpendicular to the the mechanical axis.

Figure 1. In navigation the critical measurement is alignment in the frontal plane.This is generally made in relation to the mechanical axis of the lower limb. Depending on the shape of the femur however (which varies from person to person) the cut which is made in relation to a mechanical axis can be in too much valgus (as on the right).

A failure to understand the normal anatomy of the individual patient can result in suboptimal results. This has been shown in a recent research paper published by my group (G SIngh et al Bone Joint J 2013). We found that Asian patients have a posterior slope on their tibiae that is greater than the Western population. Other groups have shown other variations in Asian patients in the frontal (coronal) plane. Following data provided by navigation protocols can therefore result in knee replacements that are knock-kneed or valgus (Figure 2). In such circumstances it is important for an experienced surgeon to override these readings as seen on navigation (Figure 1).

Figure 2. In the Asian patient, the femur tends to be laterally bowed. Therefore a cut as dictated by the computer which is perpendicular to the mechanical axis will be true to the weight bearing axis but by being too valgus will supersede the limits of tolerance of the knee prosthesis.

The sagittal plane is similarly problematic and is poorly accounted for in navigation protocols increasing the risk of notching and fractures(Figure 3).

An important aspect to the success of any knee replacement is the position of rotation of the femoral implant. The only way to establish this for certain is to scan the entire lower limb. This comes at a cost of nearly 20% over the expected cost of of knee replacement surgery in performing an additional MRI. CT scans can also be done and are cheaper but the additional radiation is difficult to justify. At present all navigation protocols have poor ability to reproduce correct rotation of implants (Figure 4).

Figure 3. The side plane or sagittal plane is poorly understood in navigation. Blindly following the instructions of the computer can result in notching or taking too much off the front of the femur.

Figure 4. The rotational axis of the the femur is poorly reproduced in traditional navigational protocols.

Patients in Asia have a strong tendency to develop bowing of the tibia or tibia vara. This odd deformity results in the axis of the tibia exiting on the outside of the knee relative to other 'normal' knees that should exit at the centre. It requires the surgeon to circumvent this problem by modifying the templating (registration) at which point the rotational and sagittal data becomes unusable in my experience. 

Figure 5. Again in Asian patients, the tibia tends to be bowed. This results in the process of navigation being misled (yellow line) and causes implants to be put in the wrong position.

With all these caveats it has been shown that with appropriate modification, the axes are best achieved with navigation. Unfortunately it is also true that this has no been shown to make a difference clinically. There are other problems with navigation including fractures due to pin placement.



Desirable qualities of a good knee replacement

Does navigation provide this quality?



Reduced blood loss


Reduced cost

No (irony is that it is less cost effective in the smaller 

unit that needs it)

Reduced complications

Probably makes little difference but adds new 

complications to the list

Good looking x-rays

Only in the coronal plane

Improved performance


The ability to handle more complex cases

Probably but stemmed implants have inherent alignment and are independent of navigation


Variations on the theme
Over the last decade a number of well conducted studies have shown that navigation has an inherent problem of validity. The very criterion of success being a final correction of within 3 degrees of a neutral (0 degree) tibio femoral axis has been found to be clinically irrelevant. Hence navigation to this end has not be found to be useful. There is no question that the data generated from this initiative has been very useful in understanding the patho-mechanics of the arthritic knee. It has subsequently spun off into the use of robotics in performing knee replacement surgery. This is remarkable and probably of some use in centers where thousands of knee replacements are presided over by a specific surgeon. In these procedures, the knee is opened by the surgeon, anatomical landmarks registered into the computer and the procedure handed over to the robot until the cuts have been made and the knee is ready to be cemented in place. While intriguing, it follows along the same lines of navigation mentioned above and therefore subject to the same pitfalls. For the vast majority of joint replacement surgeons this technique has not found routine utility.

Minimally invasive options
The combination of partial knee replacements with navigation seems to be an enticing idea. In the past, partial compartment replacements were known to have poor results and in general not to be recommended. This is probably partially related to the implants being put in suboptimally and partially because the knee had more advanced disease that was originally supposed. Certainly in our experience, most patients with knee arthritis needing replacement surgery have more than one compartment involved and therefore should only cautiously be offered partial replacements. Novel methods like Makoplasty have improved the short term results of partial replacements 10-fold. Nevertheless, the issues of long term failure due to advancing (or already advanced) disease have not been dealt with and they are not likely to outperform traditional knee replacements. There is also the problem of a relatively new implant being compared to the gold standards that have been around for nearly 30 years and have a much better track record. We cover this option elsewhere on this site.