In-vitro mechanical tests are commonly performed to assess the effect of implant design on the stability of hip endoprostheses pre-clinically. There is no standard protocol for these tests, and the loading conditions used vary greatly. Efforts have been made to standardize the test conditions , however, it is not clear how the abductor muscle and the anterior-posterior hip contact force influence the translational and rotational stability of the implant. The present study examined the effect of these two parameters in the in-vitro assessment of cementless hip implant primary stability.
As any biomechanical investigation this study has some limitations. Composite femurs were used instead of human femurs, and the implant motion was measured at only one location. These two limitations are discussed in detail in the following paragraphs. In addition, different load magnitudes were applied in sequence to each specimen,. To minimize this effect on subsequent migration, the study was designed such that the load magnitude was applied in increasing increments simulating postoperative rehabilitation. However, during a pilot test, the micromotion observed during simulated walking was similar whether these loads were applied before or after the stair climbing cycles.
Composite femurs were used to minimize experimental variability, as was done in other studies for the same reason [13, 23, 38]. Their structural stiffness has been shown to approximate that of natural bone, but with less variability [39, 40]. No comprehensive study comparing implant stability in composite versus cadaveric femurs was found in the literature, however, in-vitro tests with composite femurs  have yielded axial migration comparable to cadaveric femurs  for the CLS and press-fit Muller implants.
In our tests, the implant motion was measured at a single location. With the magnitude of physiological loads applied, the stem and the bone could not be considered rigid bodies; therefore the motion at other locations could not be determined from our experimental data. Some in-vitro studies have measured bone-implant motion at multiple locations, as reviewed by Britton et al , but the individual measurements are often limited to a single axis (e.g [23, 43]). Experimentally, space restrictions generally translate into having to choose between measuring three-dimensional motion at limited locations and measuring uniaxial motion at several locations. With a single axis motion measurement approach, however, rotational motions between the implant and the bone can incur large errors in translational motion measurement, which are proportional to the distance between the bone-implant interface and the sensor axis. A six-degree of freedom motion measurement device enabled us to avoid such error, however, our motion measurements were limited to one location.
Two common testing set-ups were selected for this study: the first set-up applied the hip contact force alone while the second applied the hip contact force together with the abductor force. The abductor force is often included rather than other muscle groups because the abductors were demonstrated to have the most important effect of all muscle groups on stresses and strains in the proximal femur [26, 28]. More complex set-ups have been used in the literature, but they are less common. For example, in one study several muscle forces (abductor, ilio-tibial band, tensor fascia latae, vastus lateralis and vastus medialis) were simulated with multiple independent actuators . A set-up modeling the hip contact force alone, on the other hand, is advocated for its simplicity and reproducibility. In a previous study , the use of this simpler model was justified based on the reported small effect of muscles on cement stresses in cemented constructs .
Our measured distal migration/micromotion magnitudes for the VerSys FMT stem (walking: ~100 μm/10 μm with both set-ups; and stair climbing: 191 μm/8 μm and 385 μm/16 μm with and without the abductor force, respectively) were within the range of values reported for other cementless implants tested in composite or cadaver femurs. Distal migration/micromotion in the order of 150 μm/10 μm, 70 μm/30 μm, and 400 μm/50 μm were reported in other studies [9, 10, 23] for the CLS stem, a press-fit cementless implant similarly intended for proximal fixation. Stem migration measured clinically for the CLS stem, however, is substantially larger (with an average in the order of 0.7 mm at 6 months) than the reported values from in-vitro experiments [2, 44]. This may be in part due to the limited number of gait cycles modeled in-vitro (usually 1000 or 5000 cycles) and/or the use of simpler and lower loads compared to those sometimes seen in-vivo, which may reach as high as eight times the body weight during stumbling, for example . Furthermore, adaptation of the bone, i.e. remodelling and local bone resorption, may also affect post-operative implant motion. In-vitro tests could at best simulate resorption by milling the bone interface at a predetermined location prior to testing . Nonetheless, the objective of in-vitro primary stability tests for cementless stems is not to provide an estimate of in-vivo migration, but to ensure that a favourable environment for successful bone ingrowth will be achieved post-operatively. It has been proposed that micromotion may be a better predictor than migration for the long-term performance of femoral implants , however, no clinical data was available to compare with our micromotion results.
The high torsional Fap loads experienced by the proximal femur during stair climbing are well documented and have been shown to occur during other activities such as jogging, fast walking, and rising from a chair [32, 48, 49]. Concerns have been raised that these forces may exceed the stem's torsional fixation strength . However, these concerns were based on comparisons with in-vitro torsional strength assessments obtained without cranical-caudal loading on the implant [29–31], which may have underestimated the torsional strength under more physiological loading. Torsional loading has been said to affect the rotational motion of femoral hip implant [24, 50]. One of these studies, however, did not apply a cranial-caudal load or measure the translational motion , while the other varied not only the torsional load applied, but also the muscle loads . Our results indicate that for a collarless, cementless implant, increasing Fap not only increases the axial rotation of the implant but that the motion increases in other directions as well, particularly distally. A similar finding was reported in another study, in which stair climbing loads generated approximately 150 μm of distal migration, compared to 30 μm of proximal migration when simulating walking loads for the CLS implant . In their study, however, the Fap (~200N, i.e. ~0.3 BW for a 70 kg individual) was smaller than the values reported for stair climbing in-vivo, i.e. 0.6 BW  and muscle forces also varied between their walking and stair climbing set-ups . Moreover, proximal migration was observed under walking loads, which the authors attributed to errors inherent in their motion measurement system. The current study, on the other hand, looked at the effect of Fap in isolation from other parameters. Increasing Fap from 0 to 0.3 BW did not have a significant effect on implant motion, but a significant increase in migration (mainly in the distal direction) was observed when increasing Fap from 0.3 BW (walking) to 0.6 BW (stair climbing) - this effect was largest without the abductor. The micromotion also increased with increasing Fap (mainly in the anterior direction), but this effect was only seen without the abductor. Rotation was primarily in the transverse plane, i.e. about the implant long axis; without the abductor stair climbing produced on average 10 times higher rotational micromotion (Table 3) and 62 times higher rotational migration (Table 2) about this axis compared to walking loads. Our results therefore support our first hypothesis: the higher Fap loads observed during stair climbing result in greater implant-bone micromotion and migration compared with walking.
We found that inclusion of the abductor muscle force stabilized the implant both in translation and rotation, particularly when simulating stair climbing. This does not support our second hypothesis. This observation, however, is similar to another study in which inclusion of muscles (abductor, tensor fascia latae and vastus lateralis) resulted in less migration than did the hip contact force alone for a cemented implant . Nonetheless, there are seemingly conflicting results in the literature; another study reported that including muscle forces (abductor, tensor fascia latae, vastus lateralis, and vastus medialis) resulted in much greater motion than did the hip contact force alone for the CLS cementless implant . Although related debates, there is no clear explanation on this conflicting result. We suspect that these differing observations may be related to differences in medial-lateral bending moments in the femur, which are. not only affected by the abductors, but also in great part by the orientation of the hip contact force. In the study by Kassi et al. , the hip contact force was applied at a 20° angle from the long axis of the femur in the frontal plane, whereas in the current study and that of Britton el al.  it was applied at 13°. These two angles are within the range reported from in vivo measurements [15, 44, 52], yet they generate different bending moment distributions. At 13° from the femur axis , the hip contact force generates medial bending in the femur, which tapers to roughly neutral bending around the implant tip, whereas at an angle of 20°  it generates medial bending in the femur around the proximal stem, but substantial lateral bending at the implant tip. The abductor load generates an additional medial bending moment, which, when superposed with the effect of the hip contact force, results in a more pronounced medial moment when the hip contact force is applied at 13° compared with when the force is applied at an angle of 20°. Differences in implant-bone interface contact stresses from the resulting bending moments may explain why the muscle forces affected implant motion differently between these studies. If this is the case, the orientation of the hip contact force may be more important than whether or not the abductor force is included in in-vitro primary stability studies. Nonetheless, it is also possible that the effect of muscles on implant motion is sensitive to the implant design.
The muscle attachment technique may also have affected the implant motion. In one study  the femurs were machined at the muscle insertion site which may have artificially weakened the femur, possibly increasing in the bone-implant motion. In the current study, the abductor attachment was done through a polymethyl-methacrylate that was fitted onto the greater trochanter, and which may have reduced the motion by stiffening the bone locally. Britton et al., however, also observed a reduction in implant motion when adding muscle forces with woven polyethylene straps glued to the greater trochanter, which is unlikely to have stiffened the bone .
Whether it is better to include or exclude the abductor and/or other muscles during pre-clinical testing is debatable. It can reasonably be argued that including all muscles provides a more physiologically representative loading scenario. However, the question of how much bending occurs physiologically is still being argued, e.g. . Inclusion of muscle forces also introduces a potential source of inter-specimen variability which could overshadow the effect of the variable being studied. Since migration measured in-vitro is typically lower than reported clinically, a set-up yielding higher bone-implant motion could be considered as favourable for pre-clinical testing. Based on our results, with the hip contact force applied at 13° from the femur axis in the frontal plane, maximum implant motion was observed when simulating stair climbing without the abductor force.