Computer assistance in total knee replacement – a critical assessment of current health care technology

Abstract

The last five years have seen the rapid development of computer assisted surgery (CAS) in total knee replacement (TKR). Many surgeons perform TKR using navigation systems, which offer user-friendly workflows and reproducible results. A number of level 1 and 2 studies, according to evidence-based medicine criteria, have demonstrated that navigation in TKR allows a more precise implantation of the prosthesis, though CAS still requires an experienced surgeon and is time consuming. Nevertheless, 30% of surgeons in Germany who perform TKR have used navigation. It is still not known whether this new technique improves the longevity of TKR, however, and we therefore performed a meta-analysis to assess the results of this new health care technology.

• Total knee replacement
• computer assisted surgery
• navigation
• health care technology

Introduction

It is quite obvious that total knee replacement (TKR) has become a standard operative procedure in recent decades, with only a 10% incidence of loosening of the prostheses within the first 10 years [1], [2].

Although implant design and the operative procedure have been optimized, aseptic loosening of the prosthesis remains a major problem in TKR. As other authors have indicated, the axial alignment of the prosthesis in TKR is one of the most important factors contributing to implant longevity and, therefore, patient satisfaction. Reconstruction of the leg axis is thus the main goal to be achieved during the operative procedure [3], [4]. Optimal reconstruction of the leg axis can be achieved via exact bony resection and optimal balancing of the periarticular structures, such as the medial and lateral collateral ligaments and the posterior capsule [5], [6]. As is known from the current literature, a correct alignment of the prosthesis is achieved in only 75% of cases, even if performed by experienced surgeons [7], [8]. It has been reported that an alignment outside the range of ± 3° varus/valgus in the anterior-posterior direction results in earlier failure of the prosthesis [3], [4]. This finding becomes even more significant as the costs of health care management explode in the industrialized states, making it imperative to avoid the need for revision surgery by performing the best possible primary implantation. This is an area of particular controversy in Germany, where health care insurers have begun developing contracts under which surgeons will, in the near future, have to guarantee the longevity of their implants, and may be required to pay for any necessary revisions out of the budget for the primary implantation. Of course, leaving aside this financial aspect, the surgeon should always do his best to optimize the implantation for the sake of his patient.

The development of navigation systems for total knee replacement has progressed considerably during the past five years, not only because of the aforementioned difficulties. The currently available navigation systems have a precision of 0.5 mm and 0.5° in clinical routine [9]. They measure the leg axis and gap size intraoperatively in real time, and some provide support for the surgeon in ligament balancing.

Krackow and colleagues published the first report describing computer assisted TKR [10]. The Optotrak® system (Northern Digital Inc., Waterloo, Ontario, Canada) used allowed intraoperative visualization of the leg axis and resection planes. Both the software and hardware have been continuously improved in recent years, and most current systems allow intraoperative measurement of the leg axis, gap size, and rotation of the components [8], [11].

Clinical routine has shown that image-free systems are most easily introduced into the operative procedure, approaching a plug-and-play workflow. During the operation, anatomic landmarks and surfaces are matched with wireless instruments, allowing the system to produce an initial model of the knee to be operated on. Using this virtual model of the knee, the system calculates the optimal position of the implant components. The cutting jigs can then be adjusted under control of the software, and cutting errors can be documented and corrected. As the authors have seen from over 1200 computer assisted TKR procedures performed to date, this technology is safe and delivers reproducible results in the hands of experienced surgeons.

Currently, new technologies have to be evaluated very carefully to avoid failure after initially promising results. In all fields of medicine, published studies differ considerably with respect to quality of design, number of patients, and performance of researchers. Therefore, the evidence-based criteria have been introduced [12]. CAS in TKR is a new technology that is increasingly spreading into operating rooms all over the world, and CAS is an increasingly important economic factor, with a proposed financial turnover of 1.5 billion US dollars in 2007. In view of this, the current level 1 and 2 studies should be reviewed critically.

Current literature

In the current literature we found six prospective and randomized clinical studies with more than 30 patients per group, which met all requirements for level 1 and 2 studies according to the criteria of evidence-based medicine [11], [13–17]. The authors of these papers compared conventional TKR and CAS TKR with respect to mechanical leg axis and component orientation (femoral and tibial). As all these authors could state, outliers with respect to the postoperative leg axis and component position could be significantly reduced by the use of CAS. However, in Table I it is apparent that not all differences are statistically significant.

Table I.  Current level 1 and 2 studies. Postoperative leg axis conventional (c) technique vs. navigated (n) technique.

			Mechanical leg axis
Author	System	No. of patients (n)	≤ 3°	> 3°	p
Mielke et al., 2001 [16]	OrthoPilot Aesculap®	60	n: 24/30 c: 20/30	n: 6/30 c: 10/30	Ø
Jenny et al., 2001 [15]	OrthoPilot Aesculap®	100	n: 47/50 c: 39/50	n: 3/50 c: 11/50	p < 0.05
Sparmann et al., 2003 [17]	Stryker Howmedica®	240	n: 120/120 c: 107/120	n: 0/120 c: 13/120	p < 0.0001
Hart et al., 2003 [14]	OrthoPilot Aesculap®	160	n: 56/60 c: 56/60	n: 4/60 c: 4/60	Ø
Bäthis et al., 2004 [8]	BrainLAB®	160	n: 78/80 c: 62/80	n: 2/80 c: 18/80	p < 0.001
Chauhan et al., 2004 [13]	Stryker Howmedica®	70	n: 30/35 c: 25/35	n: 5/35 c: 10/35	p < 0.032
Total		750	n: 355/375 c: 309/375	n: 20/375 c: 66/375	p < 0.001

If all six level 1 and 2 studies are summarized in one meta-analysis, it becomes clear that 355 of 375 total knee prostheses (94.66%) were implanted within the optimal range of ± 3° varus/valgus, compared to 309 of 375 (82.4%) prostheses in conventional TKR. According to the collected data, it is convenient to extend the limited view of the leg axis to the orientation of the femoral and tibial components in the coronal plane. Of the cited studies, only that by Mielke et al. found no statistically better results for femoral and tibial component orientation in the CAS group.

For the sagittal plane, only Bäthis et al. and Sparmann et al. found a statistically better alignment for the femoral component [8], [17]. On the tibial side, all groups except Sparmann et al. and Mielke et al. [16], [17] saw a statistically significant superiority of CAS TKR compared with the conventional technique.

In conclusion, our analysis of 375 navigated versus 375 conventional TKRs reveals that CAS improves component orientation and postoperative leg axis, minimizes outliers, and therefore helps the surgeon to improve the postoperative result following TKR.

These results are promising, but do not tell us anything about the difficulties of CAS TKR. Intraoperative determination of bony surfaces and anatomical landmarks takes time, depending on the software and system. The authors of the cited studies reported an additional time of 15-40 min per operation, with a mean of 20 min [8], [11], [16], [17]. Based on our experiences in over 1200 CAS TKR procedures, we believe that a mean of 15 min per procedure is quite reasonable if the surgeon is familiar with the system and has passed the learning curve.

One possible major complication that is often discussed is fracture of the tibia or femur due to the fixation of the reference arrays. So far, however, we have not seen this complication in any of our patients. Another possible complication of CAS is an increased incidence of deep infection due to a longer exposure time. This issue must be addressed in future studies.

It is interesting that Chauhan et al. [13] reported reduced blood loss after navigated TKR, apparently because the use of an intramedullary femoral rod is obviated. However, we found no similar observations elsewhere in the literature, so this particular aspect, which could result in faster patient recovery, also requires investigation in more detail.

Discussion

Computer assisted surgery has become a widespread procedure in total knee replacement. Since a postoperative leg alignment within the range of ± 3° varus/valgus in the anterior-posterior plane is a critical factor in determining the longevity of a knee prosthesis, and the number of implanted prostheses worldwide is approximately 500,000 per year, it is reasonable to expect that computer assisted surgery will become even more interesting to various companies in the future. Furthermore, many studies have demonstrated that a lack of accuracy in implantation leads to earlier aseptic loosening of the prosthesis, thus resulting in a revision operation for the patient and increased costs for the insurance companies.

Although knee replacement has become a standard procedure, it is still a highly demanding operation with many difficulties such as the anatomic approach, bony cuts, and ligamentous stability [18–20]. As all these difficulties can result in an early aseptic loosening of the prosthesis, a new technology such as CAS must support the surgeon in all these aspects to optimize the postoperative result.

As demonstrated in our review of the literature, five out of six current studies showed that outliers with respect to the postoperative leg axis and component orientation could be reduced by the use of navigation [8], [13],[15–17]. Nevertheless, it should be noted that for the determination of the postoperative leg axis and component orientation all authors used conventional weight-bearing long leg X-rays, which can have a calculation error of ± 2° to ± 4° [15], [20], [21]. Therefore, all values should be judged critically. So far, no study has calculated the postoperative leg axis using a standardized CT scan.

It is commonly accepted that optimal ligament balancing is a critical factor in TKR and can result in a knee that is stable over the full range of movement. The surgeon must be aware of the different effects of ligamentous releases during the operation. A stable knee in extension might be unstable in flexion, if, for example, the posterior cruciate ligament is insufficient. Therefore, the software of most currently available navigation systems has been extended with a special ligament-balancing mode wherein the system visualizes the medio-lateral gap size in extension and flexion, while the leg axis is also shown on the same screen (Figure 1). Using this procedure, the surgeon can observe the effect of necessary release steps immediately, and can carefully optimize the ligamentous stability by sequential release steps [22].

Figure 1. Ligament balancing mode with visualization of leg axis, gap size and medio-lateral instability (BrainLAB®). [Color version available online]

The implementation of the ligament-balancing mode in our experimental setup led to studies in which the anatomic approach was shown to influence the intraoperatively measured leg axis, and might therefore falsify the postoperative result [23]. Moreover, it became obvious that the position of the patella (everted versus subluxed) also influences the leg axis during ligament balancing [24].

So far, however, there have been no published studies showing significant differences between the conventional and navigated techniques with respect to clinical and functional parameters. We performed a matched pairs analysis (unpublished data) with 100 patients (conventional versus navigated technique) two years after TKR, which found no differences with regard to patient satisfaction, WOMAC Score, or Knee Society Score, but a superiority in postoperative collateral stability in patients operated with the navigated technique. Nevertheless, to prove this new technology it is necessary to perform more studies that focus on the postoperative mid- and long-term results.

Conclusion

Over the past five years, navigation in total knee replacement has attracted increasing scientific interest and has developed rapidly. As many surgeons can testify, image-free systems are particularly well established in clinical routine, since they are user-friendly for the surgeons and operating room staff. This review of all currently available studies meeting criteria for level 1 and 2 studies according to evidence-based medicine shows that CAS in TKR improves component orientation and postoperative leg alignment when compared with the conventional technique.

There are still many unanswered questions: We only know a little about navigation of the patella and the rotational alignment of the tibial component. In addition, there are no studies in the current literature that focus on short- or mid-term results with regard to early rehabilitation, patient satisfaction, and ligamentous stability. However, the most important question - whether a better leg alignment leads to increased longevity of the prosthesis – will only be answerable in approximately ten years. Still, CAS for TKR has been established in many clinics and can be recommended for use in severe preoperative leg-axis deformities, since available data indicate that the rate of complications is very low.