In the present study, we performed a detailed comparison of the cross-sectional morphology of the femoral trochlea in native and prosthetic knees. We based our choice of cutting planes on the previous report that 20° of patellofemoral flexion occurs for every 30° of knee flexion . Thus, 0, 45 and 90° cross-sections of the trochlear groove are roughly representative of 0, 67.5 and 135° knee flexion . Importantly, as there are differences in femur geometry parameters between Asian and Caucasian populations , and our findings were obtained based on measurements in Chinese subjects, it should be kept in mind that our findings are applicable to the Chinese population, and caution should be exerted when extrapolating our conclusions to non-Asian populations.
In our definition, sulcus height (H) represented the distance between the rotating axis and the trochlear groove. Since we found no significant variation among the three cross-sections in terms of sulcus height (H) values for the native knee models, we concluded that the surface of the cylinder defined in our assessment was indeed a close fit to the trochlear groove (Fig. 2). For the prosthetic knees, sulcus height (H) was comparatively higher in the 0° cross-section (by 0.76 mm), lower in the 45° cross-section (by 1.12 mm), then again higher in the 90° cross-section (by 1.53 mm) (Table 1, Fig. 4). Therefore, the native and prosthetic knees differ in terms of the best cylinder radius that would allow the cylindrical surface to closely fit the trochlear groove (Fig. 2). The discrepancies between the native and prosthetic knees in terms of sulcus height (H) values, which are exemplified in Fig. 2, may be related to different positioning of the patella following TKA; as the position of the patella determines the lever arm of the extensor mechanism, such inappropriate design of the prosthetic components is likely to influence quadriceps efficiency as well as joint reaction forces and contact levels on the femoral trochlea or condyles . Such differences might also cause the component to become anteriorly displaced, as evident in the higher sulcus height (H) values noted for the 0 and 90° cross-sections. Richard et al.  reported that, with increasing knee flexion, patellar tilt angle in the sagittal plane was substantially greater in prosthetic than in native knees, which might be related to anterior displacement of the implant. Indeed, Mihalko et al.  reported that a 2- and 4-mm build-up in the patellofemoral compartment resulted in flexion loss of 1.8 and 4.4°, respectively. Therefore, prosthetic trochlear design should be modified to avoid irritation of the soft tissue during initial and late knee flexion.
In the native knee, we noted that the sulcus angle first decreased and then increased when moving forward through the cross-sections (i.e., 159.11°, 140.66°, 148.04° in the 0°, 45° and 90° cross-sections, respectively), indicating that the bony structure may give more freedom for the patella to engage into the groove, but may hold the patella during knee flexion. This observation should be considered in the context of the patella-femoral reaction force (PRF), which represents the resultant vector of the quadriceps tendon strain force and the patellar tendon strain force, and is oriented inward in the coronal and axial views. The inward vector of PRF, occurring on the slope of the lateral femoral trochlea, neutralizes the lateral vector produced by the Q-angle of the knee [18, 19]. Therefore, the inward vector of PRF increases with the lateral tilt angle (α), stabilizing the patella. In the 0° cross-section, the PRF is relatively small, and thus the decrease in lateral tilt angle (α) has little effect on the PRF vector or the stability of the patella. However, recent studies revealed that quadriceps forces are highest between 70 and 110° [11, 20, 21]. In the 45° cross-section (67.5° of knee flexion), the PRF is relatively high, and thus the decrease in lateral tilt angle (α) might result in a decrease in the PRF vector and subsequent patellar instability. In the 90° cross-section (135° knee flexion), even though the lateral tilt angle (α) in the MP component is higher than that noted in the native knee, the patella transits over the intercondylar notch [11, 22, 23], and it is the retinacula, rather than the bony structure, that might serve as the main factor maintaining patellar stability. As the patella tilts laterally from 0 to 75° of knee flexion , the contact area is mainly on the lateral side; however, the medial tilt (β) might affect patellar tilting during knee flexion, together with the sulcus angle, and does not represent the main factor regulating patellar motion in the 0 and 45° cross sections. Additionally, the sulcus angle was larger in the prosthetic components in all cross-sections, which represents an adverse factor for patellar restraint. Therefore, after TKA, patellar stability might be more dependent on the static and dynamic stability of soft tissues rather than on the bony structure.
The present study showed that the prosthetic trochlear design does not correspond to the morphology of the native trochlea. In mid-flexion, the sulcus height (H), sulcus angle and lateral tilt angle (α) were all significantly lower in the prosthetic components, which might cause lever arm shortening, extensor weakness and decrease in the inward vector of PRF. These changes in anatomy might provide explanations for the clinical prevalence of relative quadriceps weakness [12, 25] and potential patellar dislocation after TKA. Hence, the current prosthetic trochlea might not facilitate patellar motion and quadriceps strength in mid-knee flexion.
In a well-aligned and balanced total knee prosthesis, the resurfaced patella will present a complex 3D movement pattern, broadly similar to that noted in the native knee, as discussed above. The behavior of a particular patellar component is dependent on the surface geometry variables of the mating femoral component, as well as the extrinsic stability provided by muscle and soft tissue support. Articular surface geometries of patellar components vary greatly, and each implant design bears particular advantages, with none being ultimately superior . For example, the majority of currently available patellar components are of the all-polyethylene, dome-shaped type, which may compensate for limited degrees of patellar tilt and rotation by maintaining acceptable contact congruency. Regarding femoral components, on the one hand, the MP, Triathlon and NexGen prostheses evaluated provide deepened central femoral grooves in the 45° cross-section, and the MP component has a distal extension of the trochlear groove. For patella resurfacing or non-resurfacing, such designs would be patella-friendly and provide more stability. On the other hand, all four prostheses provide larger sulcus angles than those in the native knee in all three cross-sections examined; for this reason, we speculate that, other than the native patella, the resurfaced patella would maintain stability better than the non-surfaced patella because of improved contact congruency. By contrast, if undergoing arthroplasty, native knees with similar angles would provide more stability if the native patella were retained.
There were some limitations in this study. First, the data regarding three prosthetic systems were not complete for the 90° cross-sections, as the prosthetic groove of the posterior stabilized component was not long enough to be identified in this cross-section. Therefore, more appropriate components should be included in a future study. Second, the geometry of the cartilage surface differed from that of the bone in the trochlea, although the difference was small . Third, this study only focused on the anatomical parameters of the femoral trochlea obtained from 3D-CT images, to provide some explanations with clinical implications. However, the dynamic performance of the implant and its effect on patellar motion and ligament tension should be studied in the future. Finally, only one implant size was included for each prosthetic component, even though most components are available in several sizes. Nevertheless, the implants were sized according to well-established protocols and criteria (e.g., |APk – APc| < 2 mm) and commonly used based on our experience. Moreover, the study participants were selected from an imaging database with data regarding the lower extremities of 100 healthy Chinese individuals (50 males, 50 females). Furthermore, since implants of different sizes are manufactured in the same shape, our findings regarding shape-specific parameters are relevant even if they were obtained based on one implant size.