Interface Micromotion of Cementless Hip Stems in Simulated Hip Arthroplasty

Problem statement: The design of hip prostheses has evolved over time due to various complications found after hip replacement surgery. The currently commercially available cementless femoral stems can be categorized into one of three major types, straight cylindrical, tapered rectangu lar and anatomical. Each type proposes a unique concept to achieve primary stability-a major requirement for bone healing process. Virtual analyses have bee n made on individual implants, but comparison between the three major types is required to determ in the strength and weaknesses of the design concepts. Approach: Three types of implants were modeled in three dimen sions-the straight cylindrical, rectangular taper and anatomical. The siz of the three implants was carefully designed t o fit and fill the canal of a femur reconstructed fro m a computed tomography image dataset. Hip arthroplasty was simulated virtually by inserting t he hip stem into the femoral canal. Finite element method was used in conjunction with a specialized s ub-routine to measure micromotion at the boneimplant interface under loads simulating physiologi cal walking and stair-climbing. Another subroutine was used to assign bone properties based on the grayscale values of the CT image. R sults: All the three types of cementless hip stems were found to be stable under both walking and stair climbing activities. Large micromotion values concentrated a round the proximal and distal part of the stems. Conclusion/Recommendations: The three major types of hip stems were compared in this study and all of them were found to be stable after simulated physiological activities.


INTRODUCTION
Cementless hip stems come in different shapes and sizes. In order to analyze practically the effect of these different geometries on primary stability, they should be grouped into several categories based on their features. There is no consensus at the moment in terms of grouping cementless stems according to their geometry, mostly due to the large variety of cementless stems available in the market. Healy [1] grouped cementless femoral components into 5 basic types with examples of each-the cylindrical distal filling (the AML, the solution), the anatomic, proximal fit and fill (the PCA, the anatomic), combination (the S-ROM, the bridge), dual, tapered wedge (the Omnifit, the Summit) and flat, tapered wedge (the Tri-Lock, the Accolade). Mallory et al. [2] grouped them into three distinct design geometries and philosophies-the extensive porous coating with distal fixation (the AML), the anatomic proximal fixation (the PCA, the Anatomic) and gradual proximal to distal off-loading tapered geometry (the Mallory-Head).
From the search in the literature and websites of implant manufacturers, similar groupings to Mallory et al. [2] was found based on the overall geometry. The first group belongs to hip stems which are not tapered in any plane in the distal half with cylindrical shape. Examples of hip stems within this group are the AML (DePuy) and the Versys (Zimmer). The second grouping is based on stems that have a proximal to distal taper in either or both the sagittal and longitudinal planes such as the Alloclassic (Zimmer) and the Triloc (DePuy). Some tapered designs such as the Mallory-Head also have a posteriorto-anterior taper in the coronal plane [2] . The third group is the Anatomic and was defined as stems designed with an anterior-posterior curve that mimics the natural curve of the human femur. The design therefore has a unique left and right component. Some examples of the anatomical stem are the ABG (Stryker) and the IPS (DePuy). All the three groups were categorized based on the overall geometry of the implant. Micro and macro features on the surface of the implant were ignored in the grouping.
Most cementless femoral stems rely on fixation based on press-fit where there is a contact pressure between the bone and the implant. However, due to the viscoelastic behavior of bone the effectiveness of pressfit is limited through relaxation of contact pressures at the interface [3] . The straight cylindrical stems normally rely on strong cortical support distally for stability, however, there are reports where fixation could be achieved proximally [4] . The tapered stems usually rely on a three-point fixation pattern and a gradual offloading distally [5] . Fixation can also be achieved proximally, such as the CLS stem (Zimmer), or distally, such as the Alloclassic (Zimmer). The curvature in anatomical stems provides maximum contact with the endosteal bones and sometimes with a large proximal segment to match the proximal cavity of the femur. It has been claimed that this design feature optimized resistance to axial, bending and rotational forces [6] . Fixation for the anatomical stems can also be achieved either proximally or distally. The ABG (Stryker) and the IPS (DePuy) has a smaller stem diameter distally suggesting an intended proximal fixation through distal bone overreaming technique. The Profile (DePuy), however, has a standard diameter of stem distally which can be regarded as distally fixed.

MATERIALS AND METHODS
Three implant designs to represent each group were modeled using a commercially available three dimensional modeling software. A generic straight cylindrical stem was developed to represent the first group with a geometry similar to the AML (DePuy). For the tapered group, another hip stem model was developed based on the overall geometry of the Alloclassic (Zimmer). It has a flat straight taper in transverse section and a wedge shape mediolaterally. For the anatomical group, the ABG (Stryker) prosthesis was used as a reference. All the three implants were designed to fit and fill the canal of the femur which has been reconstructed virtually from Computed Tomography (CT) image datasets. All implants have the same length and a homogeneous surface structure and surface finish throughout.
The implants were then positioned inside the femoral canals and the necks of the implants were angulated according to the anteversion angles of the intact femur (Fig. 1). The stems were assigned a linear isotropic material properties resembling titanium alloy (110 GPa), whilst the properties of the bone was assigned according to the grey level values of the CT dataset. An in-house algorithm [7] was used to correlate the grey-level of the CT images using the apparent density through cubic correlation proposed by Carter and Hayes [8] . This relationship was based on the assumption that cancellous and cortical bones are simply at different ends of a continuous spectrum.
Perfect bone-implant contact was assumed initially with the coefficient of friction set to 0.4 and an interference fit of 0.1 mm. In order to predict and simulate unstable femoral components, bone-implant contacts were removed on elements with interface micromotion of more than 50 microns after the first iteration to simulate interfacial bone loss. The models were then loaded in the same physiological loading and the iterations continued until either a stable state micromotion was achieved or loosening was predicted.
Two of the most common physiological activitieswalking and stair cimbing were used in the analysis (Fig. 2). The dataset was obtained from the work of Bitsakos [9] and Duda [10][11][12] in which muscle and joint forces for two physiological activities, walking and stair climbing, were measured using telemetry (Table 1).   Fig. 2: Location of the muscles attachment used by [9] An in-house experimentally validated computer algorithm [11] was used to measure micromotion at the interface and predict instability of the stem. This algorithm calculates the displacement of the stem relative to the endosteal surface of the bone by subtracting displacement values between corresponding nodes at the interface. Non-linear contact analysis was used by assigning target and contact surfaces between the individual parts of the model. The constraint associated with no penetration is implemented by transforming the degrees of freedom of the contact node and applying a boundary condition to the normal displacement. Figure 3 shows the magnitudes and distribution of micromotion for all three implants in both physiological walking and stair climbing. Large micromotions were found in the proximal areas and around the distal stem. The amount of surface areas with micromotion exceeded the threshold value of 50 microns were between 8-10% for all designs. After simulated interfacial bone loss, results were compared again (Fig. 4). In general, all designs were found to be stable with bone loss only increased slightly, up to 13%. The anatomical design was found to be the most stable with a very small increase in surface area above 50 microns. The cylindrical design was the worst in stair climbing with an increase in unfeasible surface area from 9-13%. The tapered design

DISCUSSION
The results showed that the three categories of hip stems, the cylindrical, the tapered and the anatomical, had similar distribution of micromotion and were all stable when bone loss was simulated. The results are in agreement with published results of actual hip stems belonging to these groups. The AML, the Alloclassic and the ABG are all implants with excellent survival rates in short-term, medium-term and long-term [14][15][16][17][18][19][20][21][22][23][24][25][26] .
The AML is one of the most widely used cementless stems with good clinical outcomes after hip arthroplasty [14] . The results are well-published covering patients with a wide range of ages. Sugiyama et al. [15] reported their 15 years clinical experience with the AML hip prosthesis. Out of 393 AML stems implanted only 6 have been revised, 3 of which were due to loosening. Another study [16] reported a survival rate of 92% at 10.5 years with 88% of the patients had good or excellent clinical results. Kronick et al. [17] reported a follow-up study of 88 AML stems inserted in patients under the age of 50 with a survival rate of 99 and 96% of which has bone ingrown, 3% stable fibrous tissue and only 1% was found unstable.
The FE results were also in agreement with other follow-up studies in terms of predicting bone ingrowth. When solid initial fixation of the AML stem is achieved intraoperatively and radiographically, bone ingrowth reliably occurs whether or not the patient is allowed partial or full weight-bearing post surgery [18,19] . Dense cortical and cancellous bone ingrowth was found with strong attachment of the metal implant to the surrounding bone [20] . No slippage was found at the interface when the retrieved sample was tested under torsional and axial load. Another study reported that an average of 57% of the porous coated area of the stem had bone ingrowth [15] .
For the anatomical design, there are also follow-up reports confirming our FE predictions. The ABG hip stem has been reported to have excellent clinical and radiographic results at short-term [21,22] and middle-term follow-up [23][24][25] . Other anatomical stem designs such as the APR-II and the Profile showed excellent clinical and radiographic results at 10 years follow-up, with no stems revised for aseptic loosening [26,27] . Proximal fixation through bone ingrowth was found in all stems and no patient reporting thigh pain after 3 years.
In vitro experimental comparison studies between a curved anatomical stem and a straight stem [28][29][30] found that at low angles of flexion, the curved and straight stems demonstrated similar patterns of interface micromotion. However, at high torsional moments such as the one observed during stair climbing, the curved stem was found to be more stable than the straight stem. Our FE results, however, did not show a significant difference between the two designs at this physiological loading condition. The reason could be that in the FE models, a perfect fit was assumed with an interference of 0.1 mm throughout the surface of the stem. The FE results showed that the straight stem had 9% surface area in excess of the threshold limit of 50 µm interface micromotion compared to the anatomical stem with 8%. However, when the bone elements were adjusted to simulate bone loss, the area f [1] or the straight stem increased to 12%, but the anatomical stem maintained at 8%. This showed that straight stems were more susceptible to micromotion during stair-climbing when a perfect fit at the interface was not achieved.
The Alloclassic, which has a tapered stem geometry, also showed good clinical outcome with a reported survival rate of 99.3-100% between 5-11 years [31][32][33] . In one of the follow-up reports of the Alloclassic [34] , 98% of the hips were rated good or excellent clinically at a median of 4 years. No stem was classified as definitely loose and no hips required revision. There was also no incidence of femoral osteolysis. Another study [35] reported that only 3 out of 133 stems subsided 2-5 mm and one subsided 5-10 mm within the first year, but no progressive subsidence could be detected beyond this period. Effenberger et al. [36] reported excellent results at 8 years for the Alloclassic with 83% showing no radiolucency and 17% showing radiolucency only proximally. A retrieval study of the Alloclassic [37] found that extensive bone-to-prosthesis apposition occurred at the interface along the stem between 6 weeks to 60 months. Other hip stem designs with tapered geometry have also shown excellent results at 5 years [38] and 10 years [39][40][41] post operatively.
In this study, the models representing the three groups have been modeled in such a way that a proper comparison could be made between them. The actual hip stem design, however, also has other important micro and macro features to enhance primary and secondary stability. The Alloclassic, for example, has a lateral flare feature-a proximal lateral expansion, which is designed to engage the lateral cortex of the femur in the metaphysis. It has been reported that this feature provides extra initial stability in cementless hip stems [42][43] . The ABG has unique semi-circular indentation macrofeatures on both its anterior and posterior sides. This feature can provide additional fixation strength proximally as the distal bone overreaming technique employed for this stem will only allow load transfer in the proximal region. Proximal tapered fins such as the one found in CLS stem and grooves were also used to provide primary and secondary stability.

CONCLUSION
Three types of cementless femoral stems were analyzed for stability using an in-house experimentally validated interface micromotion algorithm. Load simulating physiological activities of walking and stairclimbing was used and micromotion results were compared between them. Interfacial bone loss was simulated to better predict the stability of the stems. The results showed that the three types of femoral stems were stable under both physiological activities.

ACKNOWLEDGEMENT
This project has been supported by the Government research grant no. 01-01-06-SF0034.