The most important finding in this study was that variance in the TAL was smaller than that of the FAT, which means that it is more appropriate to use the TAL rather than the FAT as an alternate femoral rotational axis. In addition, no gender differences were found in the TAL, meaning that it could be used as a femoral rotational axis regardless of gender.
Owing to the advances in the development of the TKA instruments, the frequency of placements of femoral components with an inappropriate rotation has decreased . However, the rotation provided by such instruments may not be ideal, and thus surgeons should carefully determine relevant anatomical parameters for correct implant positioning . Despite the availability of radiographic images for correct implant positioning, unsatisfactory clinical outcomes occur for some patients. In consideration of these unfavorable and unexpected situations, it has been proposed that such outcomes could be ascribed to constitutional alterations specific to a certain race or gender . This should be verified by analyzing anatomical differences between populations [4, 5].
In this context, some studies have shown rotational differences between female and male patients. Previous studies have shown that the TEA and the AP axis are reliable axes for determining femoral rotation and that the TEA was the most reproducible landmark leading to the best balance , whereas the posterior femoral condyles were less reproducible for determining femoral rotation.
Various bony landmarks may not always be identifiable on x-ray images or during surgery. The clinical TEA can be easily obtained in CT images, whereas the surgical TEA is not always readily determined. Previous studies have shown that the identification rate of the medial sulcus using CT images ranges from 20% to 74% [2, 22, 26], which may further decline in knees with severe osteoarthritis. The accuracy of TEA measurements may be hampered by soft tissue coverage or difficulty in accessing the lateral condyle as well as by other surgical factors [7, 23]. The FAT lines obtained in a previous study exhibited relatively invariable internal rotation with respect to the clinical TEA . Twelve degrees of external rotation with respect to the FAT line during TKA would properly approximate the femoral rotational alignment to the TEA . In addition, the FAT line, as well as the AP axis, were not affected by varus–valgus deformities in osteoarthritic knees . However, their study did not include the angle between the surgical and clinical TEAs. They considered the angle between the PCA and the surgical TEA to be approximately 3°, whereas the angle between the FAT and the TEA should be 9.2° ± 3.6° . Nevertheless, the FAT internal rotation angle determined in this study is smaller than that indicated in previous studies [8, 23]. However, these researchers have recently measured the FAT intraoperatively, and they found that it was internally rotated by 7.3° ± 4.0° with respect to the TEA . This is very close to the value we obtained in this study, which was 6.8° ± 6.1°. It should be nonetheless noted that anthropometric characteristics are related to genetic, environmental, and sociocultural conditions and to lifestyle, health, and functional status . These variations make it challenging to provide a standard interpretation of their respective values .
Talbot and Bartlett showed that the TAL has a close correlation to the TEA and, for the first time, established an alternative to the direct visualization of the TEA . The TAL was internally rotated by 7.3° ± 1.8° with respect to the TEA in healthy knee joints, which is approximately 1° more internally rotated than in our study . However, in patients with arthritis, this value has been reported to be 5.6° ± 2.3°  and 6.1° ± 2.5° , which is comparable to our results.
After comparing the variances between the FAT and the TAL, the latter exhibited a more consistent distribution. The anterior protrusion of the lateral condyle is more anterior and larger than that of the medial condyle, and thus the distribution of the TAL with respect to the TEA is uniform. Such a trend was also found in a previous study .
A cadaveric study showed that the median surface of the cortical bone may be depressed or protruded, leading to negative values of the FAT with respect to the TEA . This could be the reason that the variance of the FAT with respect to the TEA tends to be larger than that of the TAL with respect to the TEA, implying that the TAL is a more reliable indicator of rotational alignment . However, another recent study showed that the posterior cortical bone is a better landmark than the anterior cortical bone of the distal femur for indirectly determining the surgical TEA . Data showed that the anterior cortical bone has a greater variance in relation to the TEA than the posterior cortical bone. This is attributable to the greater variability of the geometry of the distal femur in the anterior aspect compared with the posterior aspect .
Although we did not measure the posterior cortical bone line, the fact that the anterior cortical bone exhibited a greater variance in our study is in line with previous findings. Our results showed no gender differences in the TAL, and its variance was smaller than that of the FAT. Therefore, the TAL can be considered as a useful landmark.
There were three main limitations in this study. First, MRIs were used to construct 3D representations of the distal femur, Nevertheless, the MRIs allowed us to reconstruct soft tissues, such as the articular cartilage, and the inaccuracy in 3D reconstruction could be reduced using a protocol described in a previous study . Secondly, the population under study lacked ethnic diversity, and the results might differ for other populations. Thirdly, this study does not provide postoperative clinical outcomes because we did not investigate patients who underwent TKA. Nevertheless, this study provides valuable information of alternative anatomical references for surgeons when the posterior condylar surface and the trochlear groove are worn and distorted.