Computer-assisted surgery (CAS) or navigation has for the best part of two decades been the gold standard for accuracy and precision in total knee arthroplasty (TKA) in terms of creating neutral coronal and sagittal limb alignment. Accuracy refers to how close a measurement is to a “true” value, whereas precision refers to how close a series of measurements are to each other. In terms of individual anatomy, neutral alignment of the distal femur, proximal tibia, and limb in the coronal plane generally does not represent a “true” value, as it varies significantly between normal individuals. , Similarly, a symmetrical flexion gap does not represent a “true” value for the vast majority of individuals, as flexion gaps are typically asymmetric in the normal state. , , Navigated mechanical axis (MA) TKA therefore, for the vast majority of individuals, is a case of precisely reproducing an anatomically inaccurate target.
Kinematic alignment (KA) differs from MA in this regard, as it aims to target anatomically “true” values (i.e., the precise, prearthritic, normal values in each individual patient). Thus, a precision tool such as navigation has the potential for greater accuracy and precision in terms of reproducing and achieving a target when the target is a “true” and not an artificial target. This is perhaps doubly important when chasing a nonneutral and variable limb alignment target, as compounded errors in execution stacked on top of a planned restoration of normal anatomy could be problematic, for obvious reasons.
Although the underlying principles of KA using navigation instrumentation are identical to those outlined already in previous chapters, reviewing them through the lens of navigation surgery is useful as a means to understanding how to maximize the potential benefit of this technique. This chapter discusses the rationale for using navigation instrumentation, the applicable literature on the subject, assumptions regarding the soft tissue envelope of the arthritic knee as they apply to navigation principles, navigation system requirements for optimal planning, the routine technique for planning and executing KA TKA using navigation, and techniques for dealing with the most common complex deformities when navigating a KA TKA. KA is an advanced technique, and the chapter is written with the assumption that the reader has an understanding of the generic capabilities of computer-assisted TKA.
Why Use Navigation?
The Australian Orthopaedic Association National Joint Replacement Registry has reported significantly reduced hazard ratios (HRs) for revision for aseptic loosening when navigated TKA is performed, compared with nonnavigated TKA. This applies to both younger and older patient cohorts (<65 years, hazard ratio [HR] 0.63, P < .01; > 65 years, HR 0.71, P < .001). Numerous meta-analyses have reported superior accuracy and precision of navigation over other instrumentation platforms, , , and, to a lesser extent, improved patient reported outcomes. ,
Failure because of aseptic loosening is uncommon in KA. In a cohort of 222 joints, the 10-year survival rate for aseptic loosening as an end point was 98.5%. The single case of aseptic failure reported in this series was because of a large error of tibial positioning in the sagittal plane. In a much larger cohort with a minimum 2-year follow-up, the rate of early tibial failure because of aseptic loosening was 0.3% and was related to excessive tibial slope (11 degrees in the failures vs. 5 degrees on average in controls). Femoral aseptic loosening has not been reported in a KA cohort, but sagittal plane errors of femoral positioning have been linked to patellofemoral instability (PFI) (11 degrees average component flexion in PFI vs. 5 degrees average in controls).
Understandably, much of the focus in the KA versus MA discussion relates to tibial component and limb alignment in the coronal plane, as in KA there are commonly many patients whose coronal alignment would be considered an “outlier” from the MA perspective, but within the normal distribution of human anatomy. To date, no paper has reported increased tibial failure rates or inferior clinical outcomes in KA TKA cohorts attributed to coronal plane alignment variance. The same is not true of sagittal plane malalignment that has been linked to clinical failure in KA TKA. Fortunately, poor sagittal alignment can not only be consistently avoided with navigation instrumentation but can also be fine-tuned to take what is otherwise a generically shaped and sized femoral component to match the anatomic variance of each knee before any resections.
The aforementioned accuracy and precision of navigation relate to achieving coronal and sagittal plane targets. In the axial plane, the same reliability has not been achieved in terms of femoral component internal-external rotation (IR-ER), tibial component IR-ER, or rotation of the two components relative to each other. This can be problematic, as the axial rotation of the femur affects the balance and varus-valgus alignment of the limb in all positions of the limb (except full extension) and has maximal effect when the knee is flexed 90 degrees to the distal femoral resection plane. In all instrumentation systems, surface landmarks are used to determine the rotational axis of the femur. These landmarks are variably unreliable in terms of accurate registration, , and even if accurate registration can be obtained, the relationship between these landmarks can vary enormously in any one person. The posterior condylar axis (PCA) is the most reliable of these axes in terms of the accuracy and precision of registration but has been ignored in most navigation systems in favor of either the anterior-posterior (A-P) axis (otherwise known as Whiteside’s line) or the transepicondylar axes favored by MA surgeons.
Surface mapping of the distal femur is an integral part of the registration process that produces the working model of the knee in navigated TKA. The posterior condyles are easily and reliably mapped and are an alternative to other, less consistent surface landmarks for setting femoral component rotation. The kinematic flexion-extension axis of the tibia between 20 degrees and 120 degrees is the cylindrical axis of the femoral condyles for which the PCA is an excellent surrogate. This is reliably reconstituted in KA TKA by the matched distal and posterior femoral resection algorithm that reconstitutes the prearthritic PCA and consequently eliminates much of the variation imposed by other rotational axes. There is some evidence that femoral bone loss is uncommon in osteoarthritis. Thus, only the loss of distal femoral cartilage needs to be accounted for when calculating the matched resections from the distal femur to restore the premorbid femoral joint line. Tibial bone loss, by contrast, is common and difficult to quantify. Consequently, calculating the matched bony resections necessary to restore premorbid tibial height, slope, and coronal angulation is less reliable. With conventional instruments this uncertainty can be managed by making an initial, conservative, preliminary tibial resection, assessing balance and then revising the tibial resection as required. With navigation, this can be avoided in the virtual planning stage in which virtual implant size and position can be adjusted on the registered model of the knee. First, the osteophytes are cleared and next the collateral ligament on the worn side of the joint is tensioned with the knee fully extended. A valgus stress is applied to the medial collateral ligament (MCL) for a varus knee, and valgus stress is applied to the lateral collateral ligament (LCL) for a valgus knee. The navigation application will automatically measure the hip-knee-ankle angle (HKAA). The HKAA thus achieved is the target HKAA. The target lateral distal femoral angle (LDFA) is then easily determined by placing the virtual femoral trial in the planning software, to produce the necessary matched distal and posterior femoral resections. The target medial proximal tibial angle (MPTA) is then calculated by the formula: MPTA = HKAA – LDFA. This formula is based on the “Subtraction Principle” that fundamentally states that knowing two of the variables allows the surgeon to calculate the missing third value, usually the MPTA. Having thus identified the MPTA, the angle of the tibial cut in the coronal plane can be set and the height of tibial resection can be adjusted to allow for the thinnest liner and therefore minimal bone resection. The slope of the tibial cut in the sagittal plane can also be matched to the unworn compartment in the planning software. Resection gaps are then checked virtually and small adjustments made to the planned implant position to optimize resection gaps.
When the definitive bone resections are made according to the plan, in most cases no further release or tibial recut is necessary to achieve a perfectly balanced KA TKA. If the surgeon is working with a restricted kinematic approach with defined coronal plane alignment limits, “outliers” are identified virtually and the necessary adjustments made before any bony resections are made. Soft tissue releases may be required to achieve a balanced ligamentous envelope in such cases.
Kinematic TKA alignment, although often described in terms of bony resection, is essentially a soft tissue procedure in which the goal of achieving arbitrarily neutral alignment is ignored, as so doing inevitably alters the joint line. Soft tissue releases are then often required to make each patient’s unique and complex physiologic soft tissue envelope match the new nonphysiologic joint line. In the normal knee, for example, there is greater laxity on the lateral side of the knee than the medial side at all angles. , , In full extension, these differences are very small and the normal knee can be considered symmetrically balanced when fully extended. As soon as the posterior capsule is relaxed by flexing the knee, the lateral side relaxes to facilitate a medial pivot. This asymmetry is evident in early, , , middle, and late flexion. Furthermore, overall ligamentous laxity is significantly greater in females than males , and is differentially affected by resection of the cruciate ligaments. Anterior cruciate ligament (ACL) resection increases extension laxity but has a minimal effect on flexion laxity. Posterior cruciate ligament (PCL) section has the converse effect. Navigation allows virtual preoperative fine-tuning of implant size and position so that these very complex relationships can be replicated to reproduce natural soft tissue tension with as few releases or tibial recuts as possible.
Literature Review of Navigated Kinematically Aligned Total Knee Arthroplasty
This section reviews the literature supporting navigated KA TKA. Five studies have been published that employed a navigated KA technique. All used an image-free system, a cemented tibial technique, and cruciate-retaining components. In all these studies, a restricted KA technique was used with defined limits on some or all of the intraoperative coronal plane prosthetic and limb angles ( Table 7.1 ). These limits are listed below and some would argue that the restrictions placed on coronal alignment limit the applicability of these studies to true, unrestricted KA. Four of these studies used the same prosthesis (Stryker Triathlon, Stryker, Mahwah, New Jersey). , and the Stryker navigation platform. In two studies, , an abbreviated navigation platform (Orthomap ASM, Stryker, Michigan) was used. Intraoperative HKA limits were set at ±3 degrees in both studies, but at 1 year, the standing HKA ranged from 7.4 degrees varus to 6.3 degrees valgus in one study and 5.2 degrees varus to 7.6 degrees valgus in the other. In the two studies by McEwen, , a navigation platform with the capability for soft tissue integrated KA planning was used (Precision CAS, eNact Knee Navigation System v4.0 software, Stryker Leibinger, Freiburg, Germany). Intraoperative component and HKA limits were set at ±6 degrees, and at 2 years or greater the standing HKAs ranged from –9 degrees varus to 5 degrees valgus in one study and from –6 degrees varus to 4 degrees valgus in the other. In one of these studies (randomized controlled trial [RCT] of bilateral KA vs. MA), the intraoperative navigation-based and standing HKA frequency distributions were compared. In the KA cohort, the intraoperative measures were statistically maintained at 2 years on standing 3 foot radiographs. In the MA cohort, however, the tight intraoperative frequency distribution was not maintained at 2 years and in fact was statistically similar to that of the KA group. Furthermore, the position that the MA knees would have been navigated to had they been done with a KA technique (virtual KA) matched where they ended up at 2 years. The argument can therefore be made that the replaced knee will gravitate toward its prearthritic alignment over time, regardless of the alignment technique used at the time of surgery, in which case a KA technique that targets the prearthritic alignment and does not require releases makes a great deal of sense.
|Author||Study Type||n (Kinematic Alignment)||Length of Follow-Up Years||pLDFA Limit (Degrees)||pMPTA Limit (Degrees)||HKA Limit (Degrees)||Standing HKA (Degrees)|
|Hutt J||Case series||100||2.4 (1–3.7)||±5||±5||±3||–0.5 (–7.4 to 6.3)|
|Hutt J||Case series||55||Short—not specified||±5||±5||±3||–0.1 (–5.2 to 7.6)|
|Matsumoto T||RCT||30||1||No limit||All set at –3||No specified limit||–1.8 (–6 to 1.5)|
|McEwen P||Case series||192||3.5 (2–5)||±6||±6||±6||–0.7 (–9 to 5)|
|McEwen P||RCT||45||2–4||±6||±6||±6||–1.0 (–6 to 4)|
Based on Bellemans’ work, an accurate unrestricted KA technique would result in approximately 24% of limbs having an HKA outside ±3 degrees from neutral. This is consistent with the two navigated studies with the widest target HKA window (±6 degrees) in which the percentages outside ±3 degrees were 36% and 24%. In both these studies, there were significantly more male patients (Chi-square test, P = .02 28 and P = .004 24 ), thus increasing the number of cases with an HKA more varus than 3 degrees. However, numerous meta-analyses of conventional versus navigated MA TKA have reported HKA percentages outside ±3 degrees of between 25% and 30% in conventionally instrumented cohorts. In summary, therefore, a navigated KA technique produces a similar percentage of HKAs outside ±3 degrees as would be expected based on prearthritic anatomy, and a similar percentage of alignment outside ±3 degrees as would be produced using a conventional MA technique. The KA knee, however, will be balanced with few releases, whereas the MA knee must by definition be unbalanced despite more releases.
As mentioned earlier, no KA paper has reported increased aseptic tibial failure rates or inferior clinical outcomes related to an HKA outside ±3 degrees. No aseptic tibial failures have been reported in the short-term navigated KA studies. In terms of short-term data predicting long-term stability, radiostereometric analysis (RSA) is the gold standard. Migration below established thresholds at 2 years is predictive of long-term stability. A recently published RSA study compared cemented Triathlon tibial component migration in KA and MA cohorts. The KA cohort had an HKA range from 9 degrees varus to 1 degree valgus and a prosthetic MPTA (pMPTA) from 8 degrees varus to 1 degree varus. There were no differences between groups for tibial component migration or inducible displacement, both of which were below acceptable thresholds. Similarly, there was also no correlation between alignment and tibial component migration or alignment and inducible displacement. These results cannot necessarily be extrapolated to other prostheses, and ultimately, long-term data will be required to definitively answer the question of alignment and aseptic tibial failure.
Excessive tibial slope has been linked to aseptic tibial failure, , and excessive femoral flexion to patellofemoral instability. Three navigated KA studies report on sagittal plane implant positions. In one study, tibial slope was arbitrarily set at 7 degrees, in keeping with the manufacturers’ instructions. In the other two studies, , tibial component slope was matched to the slope of the unworn side of the tibia and femoral component flexion was adjusted to optimize sizing while staying within manufacturer-defined limits for combined flexion slope. In each of these studies, the mean and maximum femoral flexion angles were 3 degrees and 8 degrees, respectively, as were the mean and maximum tibial slope angles. This eliminates sagittal plane malposition as a mode of failure and is one of the more compelling arguments for using a navigation system.
The joint line orientation angle (JLOA) represents the coronal angle between the tibial plateau or component and the ground, on a standing radiograph. In bipedal stance in nonarthritic knees, the mean JLOA angle is varus but almost parallel to the ground, , although the range of values covers a spread of around 12 degrees. This is the case with both constitutionally neutral and varus limbs. This has consistently been reproduced in navigated KA studies that report mean JLOAs of –1 degrees, –1.1 degrees, –1.3 degrees, and –0.9 degrees. MA knees, by contrast, have valgus mean JLOAs in bipedal stance that go further into valgus in single-leg stance. The more valgus the JLOA of a TKA, the more the tibia is driven into varus as the knee flexes. If this is combined with the external rotation of the femoral component, as is commonly done in MA TKA, the effect becomes more pronounced.
Navigation provides the ideal platform with which to quantify and control soft tissue balance. If a kinematic alignment technique is used, balance should reflect the symmetrical or nearly symmetrical extension gap and the usually asymmetrical flexion gap of the normal knee. The rectangular flexion gap sought in MA TKA is nonphysiologic , , and is achieved by overresecting the posterior medial condyle and underresecting the posterior lateral condyle. This overconstraint of the lateral side of the knee in flexion is compounded by reversing the native sagittal convexity of the lateral tibial plateau to the concavity of a polyethylene insert that forces the femur to roll up a slope rather than down one during deep flexion. Lateral flexion gap laxity in cruciate-retaining MA TKA has been reported to improve tibial internal rotation as the knee flexes, and postoperative flexion range. , Using a navigated KA technique, McEwen reported a mean lateral flexion gap laxity of 4.5 mm and a maximum 8 mm. The mean medial to lateral laxity differential was 2.3 mm. Increasing lateral flexion gap laxity and the “medial to lateral” laxity differential was positively associated with multiple patient-reported outcome measures.
Only two RCTs comparing MA with navigation to KA with navigation have been conducted. Matsumoto reported better flexion range and net gain of flexion, as well as better scores in two KSS subsets in the KA group, at 1 year. In a trial of bilateral surgery, McEwen reported near identical outcome scores but a significant patient preference for the KA knee that was associated with the absence of a ligament release and a more physiologic JLOA.
The Soft Tissue Envelope of the Osteoarthritic Knee
KA TKA essentially adjusts prosthetic alignment to accommodate the collateral ligaments and the PCL. One of the assumptions is that the ligamentous envelope is normal and that KA will restore native premorbid alignment without release. There is, however, evidence with larger coronal deformities that the envelope is not universally normal. This section summarizes the literature so that the assumptions that are made in the technical sections that follow are clear. In the majority of varus osteoarthritic knees, the MCL does not contract, although there is some conflict in the evidence base on this subject. Bellemans and McAuliffe both reported reduced medial laxity at low flexion angles in knees with greater than 10-degrees varus deformity. In Bellemans’s study, osteophytes were removed and any inability to correct coronal alignment to neutral by filling the medial extension gap with a spacer was interpreted as medial contracture. McAuliffe left osteophytes in place and defined neutral as the position in which a virtual Macquet’s line passed through the middle of the knee and measured laxity between a varus loaded and neutral position. Both studies fail to account for the constitutionally varus knee, that is, the fact that “neutral” is not the same as “normal,” and that a wide range of individual variation exists, let alone the effect of retained osteophytes.
In contrast to these findings, Okamoto showed no medial contracture in extension regardless of varus deformity in excess of 20 degrees, and McAuliffe reached the same conclusion with respect to arthritic knees in the 90-degree flexed position. In anteromedial arthritis, the MCL will reliably maintain its length, as intact posteromedial femoral cartilage will contact intact posteromedial tibial cartilage in deeper flexion. This is one of the underlying principles of medial unicompartmental knee replacement in which an MCL release should never be needed or executed. Chronic ACL deficiency, however, can lead to a posteromedial wear pattern, fixed anterior tibial subluxation, and subsequent PCL contracture ( Fig. 7.1 ). This latter scenario is essentially the only one in which the MCL (and the PCL) cannot always be assumed to be normal in length (even though it often will be).
In contrast, there is reasonably consistent evidence that lateral extension gap laxity can increase with increasing osteoarthritic varus deformity, , , with 10 degrees of varus appearing to be the tipping point for increasing lateral laxity. In such patients, caliper-based KA resection techniques may result in a widened lateral extension gap. In such situations, a limited release of the MCL from the tibia may be required to achieve balance, rather than further varus bony resection from the tibia. The literature reviewed above and the core assumptions regarding the ligamentous envelope of the varus knee are summarized in Box 7.1 .