Transepicondylar axis accuracy in computer assisted knee surgery: A comparison of the CT-based measured axis versus the CAS-determined axis

Abstract

Rotational malalignment is recognized as one of the major reasons for knee pain after total knee arthroplasty (TKA). Although Computer Assisted Orthopaedic Surgery (CAOS) systems have been developed to enable more accurate and consistent alignment of implants, it is still unknown whether they significantly improve the accuracy of femoral rotational alignment as compared to conventional techniques.

We evaluated the accuracy of the intraoperatively determined transepicondylar axis (TEA) with that obtained from postoperative CT-based measurement in 20 navigated TKA procedures. The intraoperatively determined axis was marked with tantalum (RSA) markers. Two observers measured the posterior condylar angle (PCA) on postoperative CT scans.

The PCA measured using the intraoperatively determined axis showed an inter-observer correlation of 0.93. The intra-observer correlation, 0.96, was slightly better than when using the CT-based angle. The PCA had a range of −6° (internal rotation) to 8° (external rotation) with a mean of 3.6° for observer 1 (SD = 4.02°) and 2.8° for observer 2 (SD = 3.42°). The maximum difference between the two observers was 4°. All knees had a patellar component inserted with good patellar tracking and no anterior knee pain. The mean postoperative flexion was 113° (SD = 12.9°). The mean difference between the two epicondylar line angles was 3.1° (SD = 5.37°), with the CT-based PCA being larger.

During CT-free navigation in TKA, a systematic error of 3° arose when determining the TEA. It is emphasized that the intraoperative epicondylar axis is different from the actual CT-based epicondylar axis.

• Transepicondylar axis
• total knee arthroplasty
• posterior condylar angle
• accuracy

Introduction

The outcome of a total knee arthroplasty (TKA) depends on several factors, both patient- and surgery-related. It is known that the size of the components and, especially, their position and alignment are of great influence on the clinical outcome [1]. Primary malalignment and inadequate positioning of the femoral component in particular may lead to unsatisfactory outcomes, including patella maltracking, anterior knee pain, and flexion instability [2–4]. Malalignment is a common indication for revision surgery and can be the underlying reason for failure, PE wear, loosening and instability [5]. The revision rate attributable to malalignment may therefore be higher than stated in literature.

An external rotation of the femoral component of 3° to (a maximum of) 5° with respect to the posterior condylar line or 0° placement with respect to the transepicondylar line is thought best for optimal functionality [6]. Using the conventional and bony reference point methods, rotation of the femoral component can be determined intraoperatively by the use of the transepicondylar line, the posterior condylar line and/or Whiteside's line [7], [8].

Whilst many opinions have been expressed in the literature as to which axes are the most reliable and/or show the least intra-/inter-observer variability, none seems to be superior [9].

Several studies have shown improvement in anterior-posterior (AP) alignment using Computer Assisted Orthopaedic Surgery (CAOS) [10–12], but little is known about the attainment of better rotational alignment of the components when using CAOS [13–15].

Although these systems have been developed in an attempt to enable implants to be aligned more accurately and consistently, it is unknown whether navigation systems can improve the accuracy of femoral rotational alignment as compared to traditional techniques using mechanical guides. Since postoperative knee prosthesis problems are related to rotational malalignment, CAOS systems should reduce these errors. We studied the accuracy of intraoperative axis determination by the surgeon. To this end, the accuracy of the intraoperatively palpated and digitized TEA was compared to the postoperative CT-based epicondylar axis in 20 navigated TKA procedures.

Materials and methods

Patients

Twenty navigated TKAs in 18 patients were studied. The patients comprised 9 females and 9 males, with a mean age of 69 years (range: 46–85 years). Half of them had primary osteoarthritis; the others had secondary osteoarthritis due to rheumatoid arthritis. In all patients, the NexGen Legacy Total Knee Prosthesis (Zimmer, Warsaw, IN) was implanted with the use of cement, and the patella was resurfaced in all cases. All TKAs were performed by the same surgeon (H.M.J.vdL.). After providing informed consent, all patients participated in a prospective roentgenstereophotogrammetric (RSA) study on possible postoperative migration of the knee prostheses in CAOS TKA, including marker insertion and postoperative CT scans. To this end, tantalum (RSA) markers were inserted in the bone, and a CT scan was acquired postoperatively to measure component position.

Preoperatively, the AP leg alignment was measured on long-leg standing radiographs using the hip-knee angle (HKA) and the femoral-tibial angle (FTA). The mean preoperative HKA was 181° (SD = 4.1°) with a range of 172 to 188°; the mean FTA was 176° (SD = 7.2°) with a range of 166 to 180°. The mean extra time required for navigation during the surgery was 20 minutes.

Computer navigation

We used the VectorVision CT-free computer navigation system, software version 1.5.2 (BrainLAB, Feldkirchen, Germany). During surgery, two navigation arrays with infrared reflectors are fixed on the leg; one on the femur and one on the tibia. Identification of the anatomical landmarks, bony surfaces and axes of the knee and leg was undertaken initially using a navigated blunt pointer. The femoral localization points consisted of the medial and lateral epicondyles (Figure 1), the anterior sulcus, the femoral mechanical axis and the posterior condyles.

Figure 1. Screen of the navigation system during registration of the epicondyles of the femur. [Color version available online.]

Before identification of the bone and rotational centers of the leg and knee, the surgeon chose which reference axis was to be used for determining the correct position (i.e., rotation) of the femoral component. These reference axes in the BrainLAB system are the epicondylar line, the posterior condylar line or Whiteside's line [16]. After the digitization of bony landmarks is complete, the software calculates the ideal position of the femoral and tibial components based on the selected axes and surfaces as indicated with the pointer. With regard to rotation, the system uses the chosen rotational axes and but does not take all three into account. Hence, it shows the displacement of the component relative to all three axes (Figure 2).

Figure 2. The position of the femoral component, showing the calculated position relative to the rotation axis. [Color version available online.]

The selected rotational reference line in all cases was the epicondylar axis. To be able to identify the selected and registered epicondylar points on CT postoperatively, the digitized lateral and medial points on the epicondyles were marked with tantalum markers 1 mm in diameter. These markers can be assessed with high accuracy on CT scans and radiographs.

CT scanning

Postoperatively, prosthesis placement was checked using multi-slice CT. Based on availability, either a 16-slice (9 patients) or 64-slice (9 patients) machine was used (Aquilion, Toshiba, Otawara, Japan). CT protocols were developed based on recommendations by the BrainLAB company. For a 16-slice CT, scanning parameters were beam collimation 16 × 1 mm and pitch 0.938; images were reconstructed using a medium-smooth kernel with 1-mm slice thickness and 1-mm reconstruction index. For a 64-slice CT, scanning parameters were beam collimation 64 × 0.5 mm and pitch 0.828; images were reconstructed using a standard kernel with 1-mm slice thickness and 1-mm reconstruction index.

Images were interactively viewed on a workstation (Vitrea2, Vital Images, Minnetonka, MN) using an extended window scale (16-bit deep, up to a window width and level of 65,500). Therefore, no dedicated metal-artefact reduction filtering techniques had to be employed.

After aligning the markers in a single plane by thin-slice multiplanar reconstruction (thin MPR), thin-slice images (1–2 mm in thickness) of the distal femur were used to measure the postoperative rotational axes (Figure 3). If necessary, thick MPR may be employed to help visualize both tantalum markers at the same time.

Figure 3. Example of a CT slice (1 mm thick) of the distal femur (blue dotted line = CT-based transepicondylar line; red dashed line = line pointed and marked by tantalum markers; purple solid line = posterior condylar line). The angles between the CT-based and pointed lines and the posterior condylar line were measured. [Color version available online.]

On the postoperative CT scan, the most prominent part of the epicondyles was used to draw a line, the CT-based transepicondylar line (CTB-TEL). The other reference line was drawn between the tantalum markers–the so-called marker-based transepicondylar line (MB-TEL). The reference posterior condylar line (PCL) was drawn following the inner border of the posterior part of the femoral component, being the posterior condylar femoral osteotomy. We then measured the posterior condylar angle (PCA): this is the angle between the PCL and the transepicondylar line (TEL) (Figure 3) [17]. This was done for the CTB-TEL and the MB-TEL separately, giving the CT-based angle (CTBA) and the marker-based angle (MBA), respectively. In both instances, the same PCL was used. The CTBA and the MBA were measured twice by observer 1 (H.M.J.vdL.) and by observer 2 (R.G.H.H.N.) separately.

Since the true TEL is not known, the mean of the two PCAs (CTBA and MBA) can be used as the best estimate (limits of agreement). The difference in the two measurements for each observer of the PCA was statistically evaluated by the method of Bland and Altman [18], a non-parametric approach for comparing two methods of clinical measurement. Cohen's Kappa is calculated to assess the agreement between the two observers, where kappa = 1.0 implies perfect agreement and kappa = 0 suggests that the agreement is no better than that which would be obtained by chance.

Results

The mean measured CTBA was 3.6° for observer 1 (95% confidence interval between 1.72 and 5.48) and 2.8° for observer 2 (95% confidence interval between 1.21 and 1.59).

The mean measured MBA was 0.55° for observer 1 (95% confidence interval between −1.18 and 2.28) and 0.95° for observer 2 (95% confidence interval between −0.76 and 2.66). Thus, overall, a larger PCA was found using the CTB-TEL than with the MB-TEL (Figure 4).

Figure 4. A plot of the differences in measurements obtained with the two PCA methods (red line = average difference of 3.18°; upper red dashed line = +2 standard deviation; lower red dashed line = −2 standard deviation [SD = 3.67°]). [Color version available online.]

The inter-observer relationship for measurement of the CTBA by observers 1 and 2 was calculated and showed a linear pattern with a correlation coefficient of 0.95. The intra-observer correlation was kappa = 0.93 for the CTBA and kappa = 0.96 for the MBA (Cohen's Kappa is good if >0.80).

The mean difference between the two epicondylar line measurement methods was 3.1° (range: 0.5 to 8°; SD = 5.37°) (Figure 4).

All knees had a patellar component with good patellar tracking and no anterior knee pain. The mean postoperative maximum flexion was 113° (SD = 12.9°).

Discussion

Determination of the transepicondylar axis (TEA) during surgery is reproducible; however, comparison of the intraoperatively determined axis with a postoperative CT scan showed a systematic error of 3°. In general, determining the accurate rotation of the femoral and tibial components is difficult. However, correct component rotation is very important in TKA in order to optimize patellofemoral and tibiofemoral kinematics. We studied the accuracy of intraoperative axis determination by the surgeon using CAOS and found an inaccuracy of 3°.

There are three methods for determining femoral rotation on the basis of bony landmarks. These are based on (1) the posterior condyles with 3° of external rotation; (2) the anterior-posterior axis according to Whiteside; and (3) the TEA.

The TEA approximates the flexion axis of the knee. According to Miller et al. [19], alignment of the femoral component parallel to the epicondylar axis results in the most normal patellar tracking and minimizes patellofemoral shear forces early on. However, Kinzel et al. [20] stated that, even in experienced hands, clinical estimation of the epicondylar axis is inaccurate and should not be relied upon as the sole determinant of femoral rotation.

The goal of CAOS in TKA is to assist the surgeon in determining the optimal rotational position of the components. Although accuracy in the coronal (AP) alignment is improved by the use of CAOS [10–12], less is known about the effect on (or improvement in) rotational alignment as compared to that achieved with more traditional techniques involving mechanical guides [13–15].

Identification of the transepicondylar line during navigation is performed using a blunt pointer that the surgeon places on the palpated medial and lateral epicondyle(s) [21]. However, the shape and the soft tissue coverage of the epicondyles make these points difficult to assess, especially given the different shape of the medial and lateral epicondyles. The most prominent point of the medial epicondyle appears to be more easily detectable than the medial sulcus [21], [22]. Since the most prominent point medially and the center of the sulcus are on a line diverging by 2°, an error may be introduced, thus explaining the systematic error found in this study between the CT- and marker-based measurements. RSA markers have not been used previously, although Jerosch et al. [23] used digital analysis by video registration.

Intra-observer error in obtaining the TEA has been found to be considerable [24], [25]. The CT-free navigation software does not take into account the difference in shape of the epicondyles, and users tend to choose the most prominent and thus most easily palpable point in order to identify the landmarks. In developing computer assisted surgical techniques, one must be certain of the validity of measurements, inter-rater reliability, and reproducibility. The current method of localization of the epicondyles is therefore not ideal.

The PCA can best be measured on CT scans [26]. This “gold standard” was compared with the intraoperatively determined angle. The reproducibility of this measurement and the observer agreement between the PCAs determined using CTB-TEL and MB-TEL is very good. Furthermore, reproducibility was evaluated for observer 1 and showed an equally good result (0.93 versus 0.96).

We found that, overall, a larger PCA is measured using the CTB-TEL of 3°. Thus, the current localization procedure for the epicondyles in CAOS could lead to less external rotation of the femoral component when based on the epicondylar line. One should be aware of this difference and the possible relative internal rotation.

In general, when using CAOS, besides trying to achieve an adequate position for the sawing block, one must be aware of errors made while cutting or cementing. Partially sclerotic bone in an arthritic knee can cause the saw to divert and thus change the direction of the surface. Therefore, after cutting the bone, the surface must be checked. However, by using a computer assisted technique, the surgeon becomes aware of such cutting errors and is therefore able to correct them [27].

When using the current software for CAOS in TKA, one should check the rotational alignment of the components using the “conventional” techniques, using ligament balancing. A combination of Whiteside's line and the PCA provides a visual rotational alignment check during primary arthroplasty [7].

Using only the posterior condylar line is not reliable either. Hypoplasia and/or distortion of the lateral condyle have been described in the valgus knee [28], thereby influencing the PCA [29]. There is also a tendency for the PCA to increase with age, causing a variation in the posterior condylar angle in knees [30]. Hence, the posterior condyles are potentially unreliable reference points for femoral component rotation in some knees [31], with wide inter-individual variability of the PCA [32].

Lastly, all three bony landmarks have the disadvantage that they will not create a symmetric flexion gap in all cases. The balanced flexion gap method has the disadvantage that the femoral component may not be aligned parallel to the epicondylar axis in some cases; however, Olcott and Scott [33] stated that the TEA most consistently recreated a balanced flexion space. It is not known which of the two methods will produce better clinical results.

Conclusion

During navigation in total knee arthroplasty using the CT-free BrainLAB system, a systematic error is present between the intraoperatively determined transepicondylar axis and the CT-based bony axis.

We believe there is a need for a more accurate method to determine the epicondyles/rotation axes, thereby improving the positioning of the femoral component. It is necessary to be aware of a systematic error whilst using a navigation system. Determination of the best-fit axis may require that a combination of all rotational axes or a cloud of points at the epicondyles be used in the software to improve the accuracy of rotation.

The operating surgeon should be aware that the computer is only providing information based on the software flow of the program. Thus, “expecting the computer to recognize the epicondylar axis when we have no ‘iron clad’ way ourselves exposes the true limitations of any computer assisted surgery” [34].