Purpose

Internal-external (I-E) malrotation of the tibial and femoral components is associated with poor function after total knee arthroplasty (TKA). Kinematically aligned (KA) TKA uses functionally defined flexion-extension (F-E) tibial and femoral reference lines, which are parallel to the F-E plane of the extended knee to set I-E rotation of the tibial and femoral components. We determined whether any of five anatomically defined tibial reference lines and three anatomically defined femoral reference lines used in mechanically aligned (MA) TKA was useful as an F-E reference line.

Locating the true F-E axis of the knee can play an important role in component placement in TKA. It has been shown that the true F-E axis of the knee can be approximated by the visual-fit cylindrical axis. We determined whether a visual or computational-fit spherical axis is an accurate and reproducible substitute for the visual-fit cylindrical axis of the knee.

The target alignment for the femoral component in KA TKA is in more internal, valgus, and flexion rotation than in MA TKA, which might reduce the proximal and lateral reach of the trochlea and therefore cause abnormal patellar tracking. The present study determined the reduction in proximal and lateral reach of the trochlea for a kinematically aligned femoral component and for flexion of a MA and KA femoral component.

Methods

Sixty-two, three-dimensional bone models of normal knees were analyzed. We computed the bias as the mean and the imprecision as the standard deviation of the angle between the Hfive tibial and three femoral reference lines and the F-E reference line (+ external/ - internal). The bias was tested for statistical significance.

Forty three-dimensional bone models of native (i.e. healthy) knees constructed from magnetic resonance images (MRI) from the Osteoarthritis Initiative database (www.oai.ucsf.edu) were studied. Visual-fit and computational-fit methods were developed to locate the spherical axis. The visual-fit and computational-fit spherical axes were compared with the visual-fit cylindrical axis, which is the axis of a cylinder best fit to the tibial articular surface of each femoral condyle. Differences were computed with use of the dot product. Reproducibility was computed by having two observers perform measurements on fifteen knees, computing a simple linear regression, and testing the intercept of the regression equation for 0 and the slope for 1.

We simulated the alignment of the femoral component on 20 three-dimensional bone models of normal femurs. The reference was the most proximal point of the trochlea of the target alignment of a MA femoral component. The MA and KA femoral components were aligned in 0°, 5°, 10°, and 15° of flexion and reduced in size until the flange contacted the anterior femur. The reductions in proximal and lateral reach were computed relative to the target alignment for the MA femoral component.

Results

There was statistically significant bias between the tibial reference lines connecting the medial border (-2 °± 6°, p = 0.005), medial 1/3rd (6° ± 6°, p < 0.001), and most anterior point of the tibial tubercle (9° ± 4°, p < 0.001) with the center of the posterior cruciate ligament and the tibial reference line perpendicular to the posterior condylar axis of the tibia (-3° ± 4°, p < 0.001) and the F-E tibial reference line. There was minimal bias between the tibial reference line perpendicular to a line connecting the centers of the medial and lateral tibial condyles (1° ± 4°, p = 0.173) and the F-E tibial reference line.

There was statistically significant bias between the femoral reference lines perpendicular to a line 3° externally rotated from the posterior condylar axis of the femur (3° ± 0°, p < 0.001), perpendicular to the transepicondylar axis (3° ± 2°, p < 0.001), and parallel to the anteroposterior axis of the trochlear groove (5° ± 3°, p < 0.001) and the F-E femoral reference line.

The visual-fit and computational-fit spherical axes were found to be irreproducible. The difference between the visual-fit cylindrical axis and the visual-fit spherical axis averaged 1.1° ± 0.6° and was significant (p < 0.0001). The difference between the visual-fit cylindrical axis and the computational-fit spherical axis averaged 1.3° ± 0.7° and was significant (p < 0.0001).

The target alignment of a KA femoral component did not reduce the proximal reach, but did reduce the lateral reach by an average of 3 mm. Flexion of the MA and KA femoral components reduced the proximal reach by an average of approximately 0.8 mm/° of flexion, reduced the lateral reach by an average of approximately 1 mm generally, and reduced the size of the femoral component by 0.6 and 0.8 of a size/5° flexion for MA and KA respectively.

Conclusions

Surgeons that perform KA TKA should be aware of the bias and imprecision of these five tibial and three femoral reference lines from the F-E plane when setting the I-E rotation of the tibial and femoral components.

Because of the irreproducibility of finding the visual-fit and computational-fit spherical axes and their differences from the visual-fit cylindrical axis, the visual-fit cylindrical axis is the preferred axis for locating the true F-E axis when analyzing component placement and kinematics of the native knee.

The target alignment of a KA femoral component designed for use with MA TKA reduces the lateral reach by a clinically relevant amount of 3 mm on average. Flexion reduces the proximal and lateral reach and size of the femoral component. Limiting flexion of the MA and KA femoral components and a kinematically designed femoral component that increases the lateral reach might promote early patella engagement, more normal patellar tracking, and reduce anterior knee pain.