Medical Engineering & Physics 36 (2014) 185–195
Contents lists available at ScienceDirect
Medical Engineering & Physics journal homepage: www.elsevier.com/locate/medengphy
Bone remodeling in the resurfaced femoral head: Effect of cement mantle thickness and interface characteristics ˜ b M.A. Pérez a,∗ , P.-A. Vendittoli c , M. Lavigne c , N. Nuno a
Aragón Institute of Engineering Research (I3A), University of Zaragoza, Zaragoza, Spain Département de génie de la production automatisée, École de technologie supérieure, Laboratoire de recherche en imagerie et orthopédie, Université du Québec, Montreal, Quebec, Canada c Department of Surgery, Université de Montréal, Maisonneuve-Rosemont Hospital, Montreal, Quebec, Canada b
a r t i c l e
i n f o
Article history: Received 21 November 2012 Received in revised form 25 September 2013 Accepted 15 October 2013 Keywords: Resurfacing prosthesis Cement thickness Finite element analysis Bone remodeling Interface characteristics
a b s t r a c t Metal-on-metal hip resurfacing prostheses were re-introduced during the last 10–15 years. These prostheses have the potential to better restore normal function with limited activity restriction, being an option for younger and more active patients. Resurfacing procedures have demonstrated high failure rates in national registers [1,2]. Multiple factors may affect early and long-term HR performance. The influence of femoral cement mantle thickness and different interface characteristics between the prosthesis components on the long-term performance of resurfacing prostheses is still unknown. In the present work, a model was used to predict bone remodeling with different mantle thicknesses and interface characteristics. A very thin cement mantle (0.25 mm) increased bone resorption at the superior femoral head, while greater thickness (1 or 3 mm) had a lesser effect. In all cases, bone apposition was predicted around the stem and at the stem tip. Bone formation and resorption were observed clinically in good agreement with the predictions calculated in simulations. Computed results showed that 1-mm cement mantle thickness combined with a bonded bone–cement interface and a debonded implant–cement interface was an appropriate configuration. Bone remodeling results and computed equivalent strains were correlated. In conclusion, we have been able to demonstrate the importance of choosing an adequate cement mantle thickness. Additionally, computational studies should consider realistic interface characteristics between the components in order to perform simulations closer to reality. © 2013 IPEM. Published by Elsevier Ltd. All rights reserved.
1. Introduction Metal-on-metal hip resurfacing prostheses (HR) are used nowadays as an alternative to total hip arthroplasty (THA), especially for young and active patients [3,4]. HR have some improvements over THA, i.e., minimal femoral bone resection, easier revision, reduction of stress shielding in the proximal femur, more reliable restoration of physiological biomechanics and a lower dislocation rate [5–9]. However, some resurfacing implants presented high failure rates in national registers and published cases series [1,2,10–12]. Klotz et al. [13] obtained that the survival rate after 5–6 years was 96.3%, after 7–8 years 93.8% and after 9–10 years 90%. They detected two major problems: aseptic loosening (34.4%) and fracture of the proximal femur (31.9%). Jameson et al. [14] observed that 96.4% of HR 7 years post-op did not undergo revision surgery. Murray et al. [15]
∗ Corresponding author at: Mechanical Engineering, University of Zaragoza, Ed. Betancourt-Campus Río Ebro, C/María de Luna, 50018 Zaragoza, Spain. Tel.: +34 876 55 52 13; fax: +34 976 76 26 70. E-mail addresses: [email protected], [email protected] (M.A. Pérez).
reflected a ten-year survival of 74% in some particular designs in female patients and in small size joints due to the materials used in the bearing surfaces and the biological reactions they can elicit. These reactions are also responsible for the largest shift away from HR clinically. Multiple factors may affect early and long-term HR performance, for example, surgical technique problems, such as femoral neck notching, improper implant position/seating or poor cementing techniques [16,17], could occur with consequent aseptic loosening or neck fracture [16]. Narrowing of the femoral neck has been observed after HR, although in most cases associated with no adverse clinical or radiological outcome up to a maximum of six years after the initial operation [18]. In the long-term, migration of the femoral component was observed and the femoral components which had not migrated had radiological changes of unknown significance [19]. Previous observations could be directly related with bone remodeling around the HR components. The cementing technique is one of the speculated factors that might contribute to the long-term survival of HR [20,21]. Krause et al. [21] demonstrated that most failures were cemented inappropriately. Among the possible causes were both biological (thermal necrosis or interface biological reactions) or mechanical (inadequate initial fixation).
1350-4533/$ – see front matter © 2013 IPEM. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.medengphy.2013.10.013
186
M.A. Pérez et al. / Medical Engineering & Physics 36 (2014) 185–195
In the latter respect, computational studies could help to predict implant behavior after surgery, mainly when long-term clinical results are still not available. Finite element analysis (FEA) is a well-established method for assessing changes in mechanical stress and strain in complex structures [22]. Kuhl and Balle [23] and Gupta et al. [24] undertook bone remodeling analysis by comparing HR and THA performance. They reported that HR improved bone density distribution in the long-term with respect to THA. Some computational studies have focused on determining post-surgery changes in femoral bone stress and strain that resurfacing prostheses produce [24–32]. Most of previous FE studies have shown that fixing the full length of metallic stems, either by using cement or as a result of osseointegration leads to decreases in stress and strains in the femoral head bone [27,29,33,34]. Gupta et al. [24] concluded that resurfacing caused a reduction of strain in the bone underlying the implant (bone resorption) and elevated strains around the proximal femoral region indicating a potential risk of fracture. Pal et al. [29] indicated that resurfacing led to strain shielding of the bone of the femoral head and periprosthetic bone resorption for all interface stem–bone contact conditions. Taylor [27] observed that increasing the stem diameter and increasing the percentage stem length is in contact with bone both increased the degree of strain shielding. He also concluded that cement mantle thickness had a negligible effect on the load transfer. Similar results were predicted by Pal et al. [30]. They investigated the influence of a short-stem resurfacing component on load transfer and bone remodeling. The short-stem led to a more physiological stress distribution and bone resorption was considerably lower than with a long-stem. They also analyzed the effect of different bone–stem interface conditions. Little et al. [28] found that HR FE models showed bone strains closer to the intact conditions and that bone stresses predicted after resurfacing in both normal and aged femoral neck were not sufficient to be a potential cause of fracture. Several studies have proposed solutions to previous HR limitations. Pal et al. [35] used a ceramic prosthesis instead of metallic. High stress coupled with increased strain shielding in the proximal femoral next region appeared to be a major concern regarding its use as an alternative material. Caouette et al. [36] incorporated a biomimetic stem which did not eliminate the strain shielding effect but reduced it significantly versus a metallic cemented stem. Radcliffe and Taylor [33] analyzed the effect of different cementing techniques on load distribution in the resurfaced proximal femur, reporting that thicker cement mantles increase strain shielding. Others investigated the effect of varus–valgus orientation on load transfer and concluded that valgus alignment is preferential to varus alignment [34,37]. Ong et al. [38] studied the effect of extreme implant orientation and stem canal overreaming on initial bone remodeling stimulus. Rothstock et al. [39] investigated different interior implant geometries, cemented and uncemented solutions predicting that an uncemented press-fit implant could limit bone resorption. A more complex study was performed by Dickinson et al. [40] where they simulated prosthesis-bone interface healing with bone remodeling observing the progressive gap filling at the implant–bone interface. To date, there are few published works on the effect of the cement mantle thickness and interface characteristics on HR [27,30,33]. Cementing technique affects not only cement penetration but also the initial stability of the femoral component [41]. Previous studies did not simulate bone remodeling [27,33], but analyzed the load transfer, i.e., strains for an immediate postsurgery situation. Additionally, little is known about the effect of the bone–cement and cement–implant interface conditions on bone remodeling. Bone–cement interface has always been assumed as completely bonded [27,30,33], and cement–implant interface was mainly assumed as bonded [27,33].
Therefore, the main goal of this study is to investigate the effect of different parameters on bone density increase and/or bone resorption evolution using bone remodeling on a threedimensional (3D) FEA of a resurfaced cemented prosthesis. In particular, different cement mantle thicknesses (0.25, 1 and 3 mm) and interface conditions between the components were varied. A previously developed phenomenological bone remodeling model was tested to predict the bone response post-operatively [42]. The methodology presented in this paper is intended to improve the understanding on previous parameters (cement mantle thickness and interface characteristics between components).
2. Materials and methods The 3D FEA model used in the present study was generated from CT scans of a single right male femur (46 years old). The medical images were segmented using Mimics software (Materialize, Leuven, Belgium) to obtain a personalized geometry of the femur. Finally, the proximal femur was reconstructed using Catia V5 (Dassault Systèmes, Suresnes, France) and the resurfacing prosthesis was implanted. The implant design is based on the clinically used Zimmer Durom implant. While this implant is no longer commercially available, owing to failures related to the acetabular cup, the femoral component geometry of this implant is similar to all the other implants, hence its use in this study. Arthroplasty simulation was oriented at 5◦ valgus with respect to the neutral axis line of the femoral neck (Fig. 1), as recommended by Amstutz et al. [16]. The resurfacing prosthesis was composed of a small and polished stem attached to the spherical component, hereinafter referred to as ‘the implant’. To investigate the effect of cement mantle thickness, three different models were created to represent different gaps between the femoral head and the inner face of the implant: 0.25, 1 and 3 mm keeping the stem geometry constant (diameter of 6 mm). One milimeter is the closest to a current clinically achieved cement mantle thickness [16]. The three different configurations were modeled for a constant femoral component with outer diameter of 50 mm and inner diameter of 42 mm. In reality, the cement is interdigitated with cancellous bone, however for simplicity this was modeled by a layer of cement. The 0.25- and 1-mm configurations ensured no notching of the femoral neck, although notching was necessary for implantation in the 3-mm configuration (Fig. 1). Four-noded tetrahedral solid elements were used in automatic finite element mesh generation with Harpoon v2 (Harpoon Sharc Ltd., Manchester, UK). The different models consisted of approximately 220,000 elements and 40,000 nodes each (Fig. 2). The element size used is inside the asymptotic region of convergence and represents a good trade-off between numerical accuracy and computational cost (results not shown). Bone tissue was considered to be anisotropic and heterogeneous. Mechanical properties of bone were predicted by means of a remodeling computational model, able to evaluate the evolution of bone porosity and anisotropy as a function of the mechanical conditions [42,43]. Following the scheme proposed by Doblaré and García [42] to simulate bone remodeling, three different forces that simulate the gait cycle were considered as loading conditions, specially the forces when the foot touches the floor and the other two alternative situations of abduction and adduction, respectively [42,43]. Hip contact forces were imposed as a uniform load on nodes of the femoral head surface (see Table 1). Only abductor muscle force was considered. Loading configuration, including hip-joint contact force and abductor force, can adequately reproduce in vivo loading [44]. Initially with the femur intact, all elements were considered as bone, presenting an initial homogeneous isotropic density of 0.5 g/cm3 (Fig. 3). After the application of the forces defined above
M.A. Pérez et al. / Medical Engineering & Physics 36 (2014) 185–195
187
Fig. 1. Section of the proximal femur showing (a) implant orientation at 5◦ valgus with respect to the neutral axis line of the femoral head, and (b) 3 different cement mantle thickness configurations.
Fig. 2. (a) Finite element model of a resurfacing prosthesis showing refined mesh density and the four regions used for analyzing the results (ROIL-neck lateral; ROIM-neck medial; ROI7-metaphysis medial; ROI1-metaphysis lateral. (b) Applied boundary conditions and load application points.
during 300 increments of analysis, the actual density distribution of the intact femur was obtained [42,43]. This is the time period needed to get convergent results as is explained in García et al. [45], where the convergence performance is studied in detail. Each increment of analysis represents one day [45].
To validate predictions of the bone remodeling model, predicted bone density distribution was compared with the density values obtained from CT scans and reported in Section 3. The femur FE mesh was imported into Mimics again and different material properties were assigned relating the bone mineral density with the
Table 1 Values of applied forces at the hip joint and abductor muscles for the 3 load cases considered. Case
1 2 3
Cycles/day
6000 2000 2000
Hip-joint
Abductor
X-axis
Y-axis
Z-axis
X-axis
Y-axis
Z-axis
−242.19 −664.20 268.80
−942.40 299.70 −1113.75
−2102.72 −900.00 −1040.85
181.89 59.90 128.97
330.04 343.21 361.01
593.10 −48.87 268.42
188
M.A. Pérez et al. / Medical Engineering & Physics 36 (2014) 185–195
Fig. 3. Steps followed for the finite element simulation.
Hounsfield Units (HU). There exist different relationships between apparent bone density and HU [46,47]. In the present study, the relationship proposed by Peng et al. [47] was used (computed for the femur = 1 + 7.185 × 10-4 YHU g/cm3 ). Then, the proximal part of the femur was removed to receive the prosthesis and cement, and the material properties of the corresponding elements were modified (Fig. 3). This methodology was previously validated. All bone remodeling analyses were performed with Abaqus software v.6.10 (Dassault Systèmes, Suresnes, France), incorporating the bone remodeling model in a UMAT user subroutine. The cement and prosthesis were modeled as linear elastic materials with Poisson’s ratio of 0.3 and Young’s modulus of 2800 MPa and 200 GPa, respectively [33]. The stem was modeled as debonded within bone (no friction: = 0.0). Contact elements were defined to simulate surface-to-surface contact between the stem and the bone. On the other hand, different interface conditions were studied for the bone–cement and cement–implant interfaces. When the comparison of the cement thicknesses was performed,
bone–cement and cement–implant interfaces were considered as bonded. Later, one case with debonded cement–implant interface with friction ( = 0.3) was studied as done by Pal et al. [29], and another case with both cement–implant and bone–cement interfaces debonded with friction ( = 0.3) [48] was analyzed. Contact elements were also defined to simulate surface-to-surface contact between the bone and the cement and the implant. During the simulations, the FE model was rigidly constrained at the distal femur (Fig. 2b). The bone mineral density (BMD) ratio was computed for the whole resurfaced femur and for the different regions shown in Fig. 2 [49]. These regions were identified following a previous work of the literature [49]. The proximal part of the femur was divided in 4 volumes, each one corresponding to the following region: ROILneck lateral, ROIM-neck medial, ROI1-metaphysis lateral and ROI7metaphysis medial) as Fig. 2 indicates. The BMD ratio of the whole resurfaced femur and each region was calculated as:
BMD ratio =
t 0 dV
VOL
(1)
Fig. 4. (a) Initial bone density distribution in the intact femoral head just before the surgery and (b) bone percentage volume under a certain strain level for the three most representative loading cases of walking (Table 1).
M.A. Pérez et al. / Medical Engineering & Physics 36 (2014) 185–195
189
Fig. 5. Equivalent strain distribution in the femoral head at different time increments (incs) without implantation and with the resurfaced prosthesis for the 3 cement mantle thicknesses with bonded bone–cement and cement–implant interfaces (M-medial, L-lateral). Transverse section of the proximal femur.
Fig. 6. Bone density distribution in the femoral head at different time increments (incs) without implantation and with the resurfaced prosthesis for the 3 cement mantle thicknesses with bonded bone–cement and cement–implant interfaces (M-medial, L-lateral). Transverse section of the proximal femur.
190
M.A. Pérez et al. / Medical Engineering & Physics 36 (2014) 185–195
Fig. 7. Mean percentage change in bone mineral density for the whole finite element model and for different four regions for the 3 cement mantle thicknesses with bonded bone–cement and cement–implant interfaces.
where t and 0 were bone density (g/cm3 ) at any iteration and the initial density after implantation, respectively, with VOL (cm3 ) being the total volume of the resurfaced femur or corresponding region. The equivalent strain (Eq. (2)) was also used to compare the load transfer among the different cases analyzed. (2)εeq =
(ε1 −ε2 )2 +(ε2 −ε3 )2 +(ε3 −ε1 )2 2
3. Results 3.1. Bone remodeling model validation (intact proximal femur) Initial bone density distribution before implantation, predicted with the bone remodeling model, is presented in Fig. 4a. In order to validate the bone remodeling model, the performance of the femur with its mechanical properties from the CT scans (HU) has been compared with the ones of the bone remodeling simulation. The femur with the mechanical properties computed from the apparent bone density values obtained from the CT scans (if ≤ 1.2 then E = 2014 2.5 MPA, = 0.2; if > 1.2 then E = 1763 3.2 MPA, = 0.32) [50] and the femur with the properties from the bone remodeling simulation were independently loaded with the three load cases considered for walking (Table 1). Then, the equivalent strain was computed at each integration point of the two approaches and the total volume of bone under certain equivalent strain was evaluated and represented (Fig. 4b) comparing both approaches. The results are very similar; therefore the bone remodeling model used
predicts accurately the bone apparent density obtained from CT scans (HU). 3.2. Influence of cement mantle thickness (with resurfaced prosthesis and bonded interfaces) The equivalent strain distribution for the intact femur and the resurfacing prosthesis, with three cement thicknesses was compared (Fig. 5). An important strain reduction was predicted mainly in the superior femoral head with the resurfacing prosthesis. There were less than 5% between the three cement thicknesses results. A cement mantle thickness of 0.25 mm modified load transfer between the implant and the surrounding bone. Then, higher strains in the inferior neck were predicted with a thinner (0.25 mm) cement mantle (Fig. 5). Higher strains also appeared close to the stem for the three cement thicknesses compared to the intact femur. Bone density evolution was predicted using the bone remodeling theory and has been shown in Fig. 6. As simulation evolved (number of time increment increased), a reduction in bone density in the proximal region (superior lateral and medial femoral head) was qualitatively estimated (Fig. 6). This effect was directly related to strain reduction previously shown in Fig. 5 (strain shielding). An opposite effect was noted along the stem where bone apposition was predicted (Fig. 6). The BMD ratio was computed for the four regions (as defined in Fig. 2) and represented in Fig. 7. Significant differences were clearly observed for the different Gruen regions. At ROIL and ROIM (Fig. 7), a reduction of the BMD ratio (approximately 95%) was predicted for
M.A. Pérez et al. / Medical Engineering & Physics 36 (2014) 185–195
191
Fig. 8. Equivalent strain distribution in the resurfaced femoral head at different time increments (incs) for the different interface conditions and without implantation (Mmedial, L-lateral, B–C bone–cement interface, I–C implant–cement interface). Results computed considering a resurfacing prosthesis with 1-mm cement thickness. Transverse section of the proximal femur.
a cement mantle thickness of 0.25 mm which represented strong bone resorption, and lower resorption was observed for ROIM with 1- and 3-mm thicknesses. At ROI1 and ROI7 a clear increase in the BMD ratio was predicted for the three cement thicknesses, approximately 110% and 112%, respectively. A cement mantle thickness of 1 mm produced the highest increase of the BMD ratio in these regions. When the global BMD ratio was calculated, few changes were found between the 1- and 3-mm thicknesses, where the BMD ratio rose by approximately 108% in both cases. The 0.25-mm configuration showed the lowest BMD ratio.
conditions (Fig. 10). Significant differences were observed among the three models studied and for the different four regions. The highest BMD ratio was predicted when both interfaces were debonded, and the lowest when both interfaces were completely bonded. At ROI1 and ROI7, small differences were observed among the three interface conditions analyzed. At ROIL and ROIM regions, an increase in the BMD ratio was predicted for the completely debonded configuration of approximately 113% and 114%, respectively. When the bone–cement interface was bonded and the cement–implant interface was debonded, the BMD ratio increased 106% and 110% at the ROIL and ROIM regions, respectively (Fig. 10).
3.3. Influence of the interface conditions The different interface conditions were studied for the prosthesis with a cement mantle thickness of 1-mm. Equivalent strain distribution has been represented in Fig. 8. Differences were predicted among the three interface conditions. Equivalent strains higher than the yield strain value (0.7%) were computed at the superior head (where the stem and the implant join) when both interfaces (bone–cement and implant–cement) were debonded with friction. Changing the interface conditions altered how the loads were transferred between the different components through the corresponding interfaces. When the bone–cement was bonded and the cement–implant was debonded, the equivalent strain was similar to the intact femur (Fig. 8). Lower bone apposition was predicted when both interfaces remained bonded (Fig. 9). Higher bone apposition was predicted when the bone–cement and cement–implant interfaces were debonded. This effect was directly related to the high strains at the superior head. The BMD ratio was computed in the four regions in order to compare the different interface
4. Discussion Resurfacing may present some clinical benefits over THA [3,4,7–9]. Some resurfacing designs presented good mid-term clinical results (Birmingham Hip Resurfacing-BHR) versus other with poor results (Durom, ASR, Cormet) [10–12]. Two major problems were detected: aseptic loosening (34.4%) and fracture of the proximal femur (31.9%) [13]. As it was indicated before, HR was indicated to young and active patients. The rate of return to sports after HR appears to be excellent and unequaled by conventional hip prostheses. High-impact sports seem to be compatible with HR, although no long-term studies have analyzed the impact of these activities on wear and/or aseptic loosening [51]. A better understanding of the implant design factors associated with success/failure is needed. There are several design factors (implant position, cement mantle thickness, stem design, etc.) that could be analyzed and, indeed, some of them have been investigated by similar computational methodologies as the one described here
192
M.A. Pérez et al. / Medical Engineering & Physics 36 (2014) 185–195
Fig. 9. Bone density distribution in the resurfaced femoral head at different time increments (incs) for the different interface conditions (M-medial, L-lateral, B–C bone–cement interface, I–C implant–cement interface). Results computed considering a resurfacing prosthesis with 1-mm cement thickness. Transverse section of the proximal femur.
[26,27,33,34]. These previous studies all neglected bone remodeling simulations. In the present paper, we analyzed the effect of an important design factor, the cement mantle thickness, on the bone remodeling process. A comparison of our results has been performed with other computational studies of the literature. Equivalent strain distribution was predicted (Fig. 5). Strain shielding was predicted in the superior femoral head, and similar results were demonstrated [25,27,30,33]. Higher strains appeared close to the stem as seen in Taylor [27]. Among the three cement thicknesses, little differences in equivalent strain were computed (Fig. 5); although higher strains in the inferior neck were predicted with a thinner cement mantle (0.25 mm) [33]. Bone resorption in the superior medial and lateral femoral head (Figs. 6 and 7) was predicted. Most previous computational studies showed stress shielding in these regions [25,28,33,34,37,52]. Similar results were predicted in the bone remodeling analysis performed by Gupta et al. [24]. Dickinson et al. [53] predicted neck narrowing to be linked to a shortened horizontal femoral offset in resurfaced hips compared with healthy joints. Additionally, Dickinson et al. [40] predicted a bone mineral density reduction at the superior head-neck junction with an internal-remodeling algorithm, which would be indicative of additional external remodeling of the neck cortex, which is thin in this region. On the other hand, bone formation was predicted along the prosthesis stem and at the stem tip: a global increase of approximately 10% in the BMD ratio was predicted with the present study Figs. 7 and 10). Predicted distal bone formation was also consistent with a densitometry study that discerned changes in bone density around a resurfaced prosthesis in vivo [7]. This effect is also consistent with the report of Pollard et al. [19] where augmented density of femoral neck trabeculation and cortical hypertrophy
were observed. Lian et al. [49] clinically evaluated proximal femoral bone remodeling by dual energy X-ray absorptiometry and concluded that the BMD ratio increased at ROI1 and ROI7, as predicted in the present study (Figs. 7 and 10). A general increment of the BMD ratio (6%) was noted in Lian et al. [49], and tended to rise in a short time period. This value is comparable to that predicted in our analysis (8% for 1-mm cement thickness and bonded interface, Fig. 7; and 10% for 1-mm thickness and bonded bone–cement interface and debonded bone–implant interface, Fig. 10). It is difficult to conclude whether an increase in BMD of the order of up to 10% is relevant or not given the fact that there are a number of errors and parameters that can influence the results. Two parameters were varied in the present study (cement mantle thickness and the contact interface conditions), and the BMD percentage increase predicted is an indication of the tendency for long-term results. The results of the present study show that a very thin cement mantle (0.25 mm) is detrimental to bone, producing significant bone resorption (Figs. 6 and 7). Such a thin mantle would probably cause damage in the bulk cement itself, leading to a high stress level that could affect its long-term performance (i.e. cement fatigue failure). Such a small cement mantle thickness seems more a theoretical possibility than practically feasible. Clinically, it would be very complicated to apply and maintain such a thin layer of cement. Good results were obtained (for both bone remodeling and cement stress) with 1-mm cement mantle thickness, as proposed by Amstutz et al. [16], whereas a 3-mm thick cement mantle showed adequate performance but could have detrimental consequences due to excessive cement thickness (thermal osteonecrosis or improper implant seating) [16,17,54]. A thicker cement mantle will increase the polymerization temperatures at the bone–cement interface, inducing thermal necrosis [54].
M.A. Pérez et al. / Medical Engineering & Physics 36 (2014) 185–195
193
Fig. 10. Mean percentage change in bone mineral density for the whole finite element model and for different four regions for the different interface conditions (B–C bone–cement interface, I–C implant–cement interface) with a cement mantle thickness of 1 mm.
In the present study, the effect of different interface conditions between the prosthesis components was studied. In all cases analyzed, the bone–stem interface was assumed as debonded without friction [29]. Indeed, Pal et al. [29] observed that changes in the bone–stem characteristics did not have any major consequences for the stress/strain pattern in the femoral head. In general, changes in the interface conditions directly affect to the load path, and as a consequence to the bone strain shielding and the bone density evolution. Changes in the implant–cement interface condition (bone–cement interface bonded) from bonded to debonded resulted in noticeable differences in strain distributions and periprosthetic bone adaptation within the resurfaced femoral head (Fig. 8 and 9). For a debonded implant–cement interface, the effect of strain shielding was considerably higher than for a debonded interface. In general, there is a change in the load transfer conditions, debonding the implant from the cement increases the strains and therefore the stimulus is higher increasing the bone formation, as it was observed in our results (Figs. 8 and 9). Assuming both interfaces (bone–cement and implant–cement) as debonded was not a real situation. This interface condition assumes that the cement is completely debonded between the bone and the implant increasing the strains locally. This situation may come from a failure during surgery. High equivalent strain values were locally predicted at the superior femoral head in the last condition (Fig. 8). In most computational studies, bone–cement interface was considered bonded. Only Pal et al. [35] computed the multi-axial Hoffmann failure criterion at the bone–cement interface concluding that this interface was secure against interface debonding. Thus, simulating a bonded bone–cement interface in future studies is reasonable.
4.1. Limitations Although the data are in good agreement with clinical findings [7,16,49] and other computational studies, there are some limitations of our research that deserve comment. The cement mantle was modeled as a uniform layer of pure bone cement. The nonuniformity of real cement mantle (see Fig. 1) has not been simulated [17]. Therefore, load transfer from the prosthesis to the bone has been slightly modified. Sometimes macro cement pocket is located on the implant internal dome surface giving effective implantcement bonding at least locally. Another limitation is related with the the bone–cement interface, there is cement interdigitation within bone which was not considered. Most of computational studies have assumed it as negligible [24,25,27,34,55]. The choice of cementing technique depends on various factors such as heat effects, fixation and curing time [54]. Depending on the cementing technique, the penetration of cement into cancellous bone could be different [56], changing the mechanical properties of the surrounding region. In our study, we considered the bone–cement interface as being completely bonded. This limitation can be solved by incorporating a mixed-mode failure model [57,58] that takes into account the mechanical properties of the bone–cement interface. The increased cement mantle was modeled by keeping the same femoral component diameter and reducing the femoral bone core. Perhaps it should have been modeled by enlarging femoral component size with the same prepared femoral bone. In addition, the curing conditions of bone–cement affect local interface and bone stresses. This will affect load transfer to the surrounding bone. These effects were not taken into account in the current study.
194
M.A. Pérez et al. / Medical Engineering & Physics 36 (2014) 185–195
Another limitation was related to the bone remodeling model used [42] that was unable to simulate external bone remodeling. Narrowing of the femoral neck after hip resurfacing arthroplasty has been clinically described and is mainly circumferential [18]; with the help of an external bone remodeling model, we would have been able to reproduce this effect. External bone remodeling models follow the idea to simulate the growth behavior of biological systems, known as CAO methods [45]. The advantage of using a bone remodeling model to predict the bone material properties with respect to a technique based on CT-scan data is that bone remodeling model takes into account porosity and anisotropy evolution [59]. A further limitation is that a single femur model was used to perform this analysis. In the authors’ opinion, however, this limit does not reduce the importance and generality of the obtained results. Inter-patient variability should be considered in future studies. In the present study, we analyzed walking loads and neglected other activities [36,39,53]. We have also assumed mean values for muscle forces and joint contact forces, ignoring their variability. The muscle insertions and application points are also an approximation to reality because the femur here considered is not exactly the same as the one used by the reference studies in the literature [60,61]. This fact could affect the bone remodeling predictions in the femur. A probabilistic methodology, like the one proposed by Pérez et al. [62], would be more appropriate in order to consider load variability. To improve these results and conclusions, a cement damage model could be incorporated to analyze whether the cement deterioration (cement cracking and cement fatigue properties) could affect the long-term performance of resurfacing prostheses. Additionally, an optimization procedure could have been previously developed to determine the threshold limits of the cement thickness here considered [63]. 5. Conclusion Based our numerical study, we can conclude that HR results in bone resorption at the superior femoral head and in bone apposition around the stem and at the stem tip for the different cement mantles investigated. From the results predicted, 1-mm cement mantle thickness may be an appropriate cement configuration. A bone–cement interface bonded and a cement–implant interface debonded combined with 1-mm cement mantle is numerically closest to the intact femur model analyzed. Therefore, we can summarize that the computational model proposed is a useful tool in order to determine the influence of different factors (cement mantle thickness and interface characteristics between components). It can be additionally adopted to study the influence of other design factors. Declarations Funding: The authors gratefully acknowledge the research support of the Spanish Ministry of Economy and Competitiveness through Research Project 2011-22413. Competing interests None declared. Ethical approval Not required. References [1] Steiger Rd, Miller L, Prosser G, Graves S, Davidson D, Stanford T. Poor outcome of revised resurfacing hip arthroplasty. Acta Orthop 2010;81:72–6.
[2] Prosser G, Yates P, Wood D, Graves S, Steiger Rd, Miller L. Outcome of primary resurfacing hip replacement: evaluation of risk factors for early revision. Acta Orthop 2010;81:66–71. [3] Daniel J, Pynsent P, McMinn D. Metal-on-metal resurfacing of the hip in patients under the age of 55 years with osteoarthritis. J Bone Joint Surg Br 2004;86:177–84. [4] Treacy R, McBryde C, Pynsent P. Birmingham hip resurfacing arthroplasty. J Bone Joint Surg Br 2005;87:167–70. [5] McMinn D, Treacy R, Lin K, Pynsent P. Metal on metal surface replacement of the hip. Clin Orthop 1996;329:89–98. [6] Kelley S, Lachiewicz P, Hickman J, Paterno S. Relationship of femoral head and acetabular size to prevalence of dislocation. Clin Orthop 1998;355:163–70. [7] Kishida Y, Sugano N, Nishii T, Miki H, Yamaguchi K, Yoshikawa H. Preservation of the bone mineral density of the femur after surface replacement of the hip. J Bone Joint Surg Br 2004;86:185–9. [8] Silva M, Lee K, Heisel C, Rosa MD, Schmalzried T. The biomechanical results of total hip resurfacing arthroplasty. J Bone Joint Surg Am 2004;86:40–6. [9] Girard J, Lavigne M, Vendittoli P, Roy A. Biomechanical reconstruction of the hip. A randomised study comparing total hip resurfacing and total hip arthroplasty. J Bone Joint Surg Br 2006;88:721–6. [10] Ng V, Arnott L, McShane M. Perspectives in managing an implant recall: revision of 94 Durom Metasul acetabular components. J Bone Joint Surg Am 2011;93:101–5. [11] Naal F, Pilz R, Munzinger U, Hersche O, Leunig M. High revision rate at 5 years after hip resurfacing with the Durom implant. Clin Orthop 2011;469:2598–604. [12] Cohen D. Out of joint: the story of the ASR. BMJ 2011;342:2905. [13] Klotz M, Breusch S, Hassenpflug M, Bitshc R. Results of 5 to 10-year followup after hip resurfacing: a systematic analysis of the literature on long-term results. Orthopade 2012;41:442–51. [14] Jameson S, Baker P, Mason J, Porter M, Deehan D, Reed M. Independent predictors of revision following metal-on-metal hip resurfacing: a retrospective cohort study using National Joint Registry data. J Bone Joint Surg Br 2012;94:746–54. [15] Murray DW, Grammatopoulos G, Pandit H, Gundle R, Gill HS, McLardy-Smith P. The ten-year survival of the Birmingham hip resurfacing: an independent series. J Bone Joint Surg Br 2012;94:1180–6. [16] Amstutz H, Campbell P, Duff ML. Fracture of the neck of the femur after surface arthroplasty of the hip. J Bone Joint Surg Am 2004;86:1874–7. [17] Morlock M, Bishop N, Ruther W, Delling G, Hahn M. Biomechanical, morphological and histological analysis of early failures in hip resurfacing arthroplasty. Proc Inst Mech Eng [H] 2006:220. [18] Hing C, Young D, Dalziel R, Bailey M, Back D, Shimmin A. Narrowing of the neck in resurfacing arthroplasty of the hip. J Bone Joint Surg Br Vol 2007;89:1019–24. [19] Pollard T, Baker R, Eastaugh-Waring S, Bannister G. Treatment of the young active patient with osteoarthritis of the hip. J Bone Joint Surg Br 2006;88:592–600. [20] Beckmann J, Goldapp C, Ringleff K, Schaumburger J, Grifka J, Perlick L. Cementing technique in femoral resurfacing. Arch Orthop Trauma Surg 2009;129:1317–25. [21] Krause M, Breer S, Hahn M, Ruther W, Morlock MM, Amling M, et al. Cementation and interface analysis of early failure cases after hip-resurfacing arthroplasty. Int Orthop 2012;36:1333–40. [22] Prendergast P. Finite element models in tissue mechanics and orthopaedic implant design. Clin Biomech 1997;12:343–66. [23] Kuhl E, Balle F. Computational modelling of hip replacement surgery: total hip replacement vs. hip resurfacing. Technische Mechanik 2005;25:1–9. [24] Gupta S, New A, Taylor M. Bone remodelling inside a cemented resurfaced femoral head. Clin Biomech 2006;21:594–602. [25] Wanatabe Y, Shiba N, Matsuo S, Higuchi F, Tagawa Y, Inoue A. Biomechanical study of the resurfacing hip arthroplasty. J Arthroplasty 2000;15:505–11. [26] Ong K, Kurtz S, Manley M, Rushton N, Mohammed N, Field R. Biomechanics of the Birmingham hip resurfacing arthroplasty. J Bone Joint Surg Br 2006;88:1110–5. [27] Taylor M. Finite element analysis of the resurfaced femoral head. Proc Inst Mech Eng [H] 2006;220:289–97. [28] Little J, Taddei F, Viceconti M, Murray D, Gill H. Changes in femur stress after hip resurfacing arthroplasty: response to physiological loads. Clin Biomech 2007:22. [29] Pal B, Gupta S, New A. A numerical study of failure mechanisms in the cemented resurfaced femur: effects of interface characteristics and bone remodelling. Proc Inst Mech Eng [H] 2009;223:471–84. [30] Pal B, Gupta S, New A. Influence of the change in stem length on the load transfer and bone remodelling for a cemented resurfaced femur. J Biomech 2010;43:2908–14. [31] Dickinson AS, Taylor AC, Jeffers JR, Browne M. Performance of the resurfaced hip. Part 2: the influence of prosthesis stem design on remodelling and fracture of the femoral neck. Proc Inst Mech Eng H 2010;224:841–51. [32] Long JP, Santner TJ, Bartel DL. Hip resurfacing increases bone strains associated with short-term femoral neck fracture. J Orthop Res 2009;27: 1319–25. [33] Radcliffe I, Taylor M. Investigation into the effect of cementing techniques on load transfer in the resurfaced femoral head: a multi-femur finite element analysis. Clin Biomech 2007;22:422–30. [34] Radcliffe I, Taylor M. Investigation into the effect of varus–valgus orientation on load transfer in the resurfaced femoral head: a multi-femur finite element analysis. Clin Biomech 2007;22:780–6.
M.A. Pérez et al. / Medical Engineering & Physics 36 (2014) 185–195 [35] Pal B, Gupta S, New A. Design considerations for ceramic resurfaced femoral head: effect of interface characteristics on failure mechanisms. Comput Methods Biomech Biomed Eng 2010;13:143–55. ˜ N. Anisotropic bone remodel[36] Caouette C, Bureau M, Vendittoli PMML, Nuno ing of a biomimetic metal-on-metal hip resurfacing implant. Med Eng Phys 2012;34:559–65. [37] Long J, Bartel D. Surgical variables affect the mechanics of a hip resurfacing system. Clin Orthop 2006;453:115–22. [38] Ong KL, Day JS, Kurtz SM, Field RE, Manley MT. Role of surgical position on interface stress and initial bone remodeling stimulus around hip resurfacing arthroplasty. J Arthroplasty 2009;24:1137–42. [39] Rothstock S, Uhlenbrock A, Bishop N, Laird L, Nassutt R, Morlock M. Influence of interface condition and implant design on bone remodelling and failure risk for the resurfaced head. J Biomech 2011:44. [40] Dickinson A, Taylor A, Browne M. Implant–bone interface healing and adaptation in resurfacing hip replacement. Comput Methods Biomech Biomed Eng 2012;15:935–47. [41] Bitsch R, Jäger S, Lürssen M, Loidolt T, Schmalzried T, Weiss S. The influence of cementing technique in hip resurfaceing arthroplasty on the initial stability of the femoral component. Int Orthop 2011;35:1759–65. [42] Doblaré M, García J. Application of an anisotropic bone-remodeling model based on a damage-repair theory to the analysis of the proximal femur before and after total hip replacement. J Biomech 2001;34:1157–70. ˜ [43] García J, Doblaré M, Cegonino J. Bone remodelling simulation: a tool for implant design. Comput Mater Sci 2002;25:100–14. [44] Stolk J, Verndonschot N, Huiskes R. Hip-joint and abductor-muscle forces adequately represent in vivo loading of a cemented total hip reconstruction. J Biomech 2001;34:917–26. [45] García J, Martínez M, Doblaré M. An anisotropic internal–external bone adaptation model based on a combination of CAO and continuum damage mechanics technologies. Comput Methods Biomech Biomed Eng 2001;4:355–78. [46] Rho J, Hobatho M, Ashman R. Relations of mechanical properties to density and CT numbers in human bone. Med Eng Phys 1995;17:347–55. [47] Peng L, Bai J, Zeng X, Zhou Y. Comparison of isotropic and orthotropic material property assignments on femoral finite element models under two loading conditions. Med Eng Phys 2007;28:227–33. [48] Janssen D, Mann KA, Verdonschot N. Micro-mechanical modeling of the cement–bone interface: the effect of friction, morphology and material properties on the micromechanical response. J Biomech 2008;41:3158–63. [49] Lian Y, Pei F, Yoo M, Chenq J, Fatou C. Changes of the bone mineral density in proximal femur following total hip resurfacing arthroplasty in osteonecrosis of femoral head. J Orthop Res 2008;26:453–9.
195
[50] Beaupre GS, Orr TE, Carter DR. An approach for time-dependent bone modeling and remodeling—theoretical development. J Orthop Res 1990;8:651–61. [51] Girard J, Miletic B, Deny A, Migaud H, Fouilleron N. Can patients return to highimpact physical activities after hip resurfacing? A prospective study. Int Orthop 2013;37:1019–24. [52] Huiskes R, Strens P, Heck Jv, Slooff T. Interface stresses in the resurfaced hip: finite element analyis of load transmissions in the femoral head. Acta Orthop Scand 1985;56:474–8. [53] Dickinson A, Taylor A, Browne M. Performance of the resurfaced hip. Part 1: the influence of the prosthesis size and positioning on the remodelling and fracture of the femoral neck. Proc Inst Mech Eng [H] 2010;224:841–51. ˜ N, Madrala A, García-Aznar J, Doblaré M. Computacional mod[54] Pérez M, Nuno elling of bone cement polymerization: temperature and residual stresses. Comput Biol Med 2009;39:751–9. [55] Taddei F, Martelli S, Gill H, Cristofolini L, Viceconti M. Finite element modelling of resurfacing hip prosthesis: estimation of accuracy through experimental validation. J Biomech Eng 2010;132:1–11. [56] Beaulé P, Matar W, Poitras P, Smit K, May O. 2008 Otto Aufranc Award: component design and technique affect cement penetration in hip resurfacing. Clin Orthop 2009;467:84–93. [57] Pérez M, Palacios J. Comparative finite element analysis of the debonding process in different concepts of cemented hip implants. Ann Biomed Eng 2010;38:2093–106. [58] Moreo P, Pérez M, García-Aznar J, Doblaré M. Modelling the mixed-mode failure of cement–bone interfaces. Eng Fract Mech 2006;73:1379–95. [59] Pérez M, Fornells P, Doblaré M, García-Aznar J. Comparative analysis of bone remodeling models with respect to computerised tomography-based finite element models of bone. Comput Methods Biomech Biomed Eng 2010;12: 101–8. [60] Bergmann G, Deuretzbacher G, Heller M, Graichen F, Rohlmann A, Strauss J. Hip contact forces and gait patterns from routine activities. J Biomech 2001;34:859–71. [61] Heller M, Bergmann G, Deuretzbacher G, Dürselen L, Pohl M, Claes L, et al. Musculo-skeletal loading conditions at the hip during walking and stair climbing. J Biomech 2001;34:883–93. [62] Pérez M, Grasa J, García-Aznar J, Bea J, Doblaré M. Probabilistic analysis of the influence of the bonding degree of the stem–cement interface in the performance of cemented hip prostheses. J Biomech 2006;39: 1859–72. [63] Vena P, Verdonschot N, Contro R, Huiskes R. Sensitivity analysis and optimal shape design for bone–prosthesis interfaces in a femoral head surface replacement. Comput Methods Biomech Biomed Eng 2000;3:245–56.
Contents lists available at ScienceDirect
Medical Engineering & Physics journal homepage: www.elsevier.com/locate/medengphy
Bone remodeling in the resurfaced femoral head: Effect of cement mantle thickness and interface characteristics ˜ b M.A. Pérez a,∗ , P.-A. Vendittoli c , M. Lavigne c , N. Nuno a
Aragón Institute of Engineering Research (I3A), University of Zaragoza, Zaragoza, Spain Département de génie de la production automatisée, École de technologie supérieure, Laboratoire de recherche en imagerie et orthopédie, Université du Québec, Montreal, Quebec, Canada c Department of Surgery, Université de Montréal, Maisonneuve-Rosemont Hospital, Montreal, Quebec, Canada b
a r t i c l e
i n f o
Article history: Received 21 November 2012 Received in revised form 25 September 2013 Accepted 15 October 2013 Keywords: Resurfacing prosthesis Cement thickness Finite element analysis Bone remodeling Interface characteristics
a b s t r a c t Metal-on-metal hip resurfacing prostheses were re-introduced during the last 10–15 years. These prostheses have the potential to better restore normal function with limited activity restriction, being an option for younger and more active patients. Resurfacing procedures have demonstrated high failure rates in national registers [1,2]. Multiple factors may affect early and long-term HR performance. The influence of femoral cement mantle thickness and different interface characteristics between the prosthesis components on the long-term performance of resurfacing prostheses is still unknown. In the present work, a model was used to predict bone remodeling with different mantle thicknesses and interface characteristics. A very thin cement mantle (0.25 mm) increased bone resorption at the superior femoral head, while greater thickness (1 or 3 mm) had a lesser effect. In all cases, bone apposition was predicted around the stem and at the stem tip. Bone formation and resorption were observed clinically in good agreement with the predictions calculated in simulations. Computed results showed that 1-mm cement mantle thickness combined with a bonded bone–cement interface and a debonded implant–cement interface was an appropriate configuration. Bone remodeling results and computed equivalent strains were correlated. In conclusion, we have been able to demonstrate the importance of choosing an adequate cement mantle thickness. Additionally, computational studies should consider realistic interface characteristics between the components in order to perform simulations closer to reality. © 2013 IPEM. Published by Elsevier Ltd. All rights reserved.
1. Introduction Metal-on-metal hip resurfacing prostheses (HR) are used nowadays as an alternative to total hip arthroplasty (THA), especially for young and active patients [3,4]. HR have some improvements over THA, i.e., minimal femoral bone resection, easier revision, reduction of stress shielding in the proximal femur, more reliable restoration of physiological biomechanics and a lower dislocation rate [5–9]. However, some resurfacing implants presented high failure rates in national registers and published cases series [1,2,10–12]. Klotz et al. [13] obtained that the survival rate after 5–6 years was 96.3%, after 7–8 years 93.8% and after 9–10 years 90%. They detected two major problems: aseptic loosening (34.4%) and fracture of the proximal femur (31.9%). Jameson et al. [14] observed that 96.4% of HR 7 years post-op did not undergo revision surgery. Murray et al. [15]
∗ Corresponding author at: Mechanical Engineering, University of Zaragoza, Ed. Betancourt-Campus Río Ebro, C/María de Luna, 50018 Zaragoza, Spain. Tel.: +34 876 55 52 13; fax: +34 976 76 26 70. E-mail addresses: [email protected], [email protected] (M.A. Pérez).
reflected a ten-year survival of 74% in some particular designs in female patients and in small size joints due to the materials used in the bearing surfaces and the biological reactions they can elicit. These reactions are also responsible for the largest shift away from HR clinically. Multiple factors may affect early and long-term HR performance, for example, surgical technique problems, such as femoral neck notching, improper implant position/seating or poor cementing techniques [16,17], could occur with consequent aseptic loosening or neck fracture [16]. Narrowing of the femoral neck has been observed after HR, although in most cases associated with no adverse clinical or radiological outcome up to a maximum of six years after the initial operation [18]. In the long-term, migration of the femoral component was observed and the femoral components which had not migrated had radiological changes of unknown significance [19]. Previous observations could be directly related with bone remodeling around the HR components. The cementing technique is one of the speculated factors that might contribute to the long-term survival of HR [20,21]. Krause et al. [21] demonstrated that most failures were cemented inappropriately. Among the possible causes were both biological (thermal necrosis or interface biological reactions) or mechanical (inadequate initial fixation).
1350-4533/$ – see front matter © 2013 IPEM. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.medengphy.2013.10.013
186
M.A. Pérez et al. / Medical Engineering & Physics 36 (2014) 185–195
In the latter respect, computational studies could help to predict implant behavior after surgery, mainly when long-term clinical results are still not available. Finite element analysis (FEA) is a well-established method for assessing changes in mechanical stress and strain in complex structures [22]. Kuhl and Balle [23] and Gupta et al. [24] undertook bone remodeling analysis by comparing HR and THA performance. They reported that HR improved bone density distribution in the long-term with respect to THA. Some computational studies have focused on determining post-surgery changes in femoral bone stress and strain that resurfacing prostheses produce [24–32]. Most of previous FE studies have shown that fixing the full length of metallic stems, either by using cement or as a result of osseointegration leads to decreases in stress and strains in the femoral head bone [27,29,33,34]. Gupta et al. [24] concluded that resurfacing caused a reduction of strain in the bone underlying the implant (bone resorption) and elevated strains around the proximal femoral region indicating a potential risk of fracture. Pal et al. [29] indicated that resurfacing led to strain shielding of the bone of the femoral head and periprosthetic bone resorption for all interface stem–bone contact conditions. Taylor [27] observed that increasing the stem diameter and increasing the percentage stem length is in contact with bone both increased the degree of strain shielding. He also concluded that cement mantle thickness had a negligible effect on the load transfer. Similar results were predicted by Pal et al. [30]. They investigated the influence of a short-stem resurfacing component on load transfer and bone remodeling. The short-stem led to a more physiological stress distribution and bone resorption was considerably lower than with a long-stem. They also analyzed the effect of different bone–stem interface conditions. Little et al. [28] found that HR FE models showed bone strains closer to the intact conditions and that bone stresses predicted after resurfacing in both normal and aged femoral neck were not sufficient to be a potential cause of fracture. Several studies have proposed solutions to previous HR limitations. Pal et al. [35] used a ceramic prosthesis instead of metallic. High stress coupled with increased strain shielding in the proximal femoral next region appeared to be a major concern regarding its use as an alternative material. Caouette et al. [36] incorporated a biomimetic stem which did not eliminate the strain shielding effect but reduced it significantly versus a metallic cemented stem. Radcliffe and Taylor [33] analyzed the effect of different cementing techniques on load distribution in the resurfaced proximal femur, reporting that thicker cement mantles increase strain shielding. Others investigated the effect of varus–valgus orientation on load transfer and concluded that valgus alignment is preferential to varus alignment [34,37]. Ong et al. [38] studied the effect of extreme implant orientation and stem canal overreaming on initial bone remodeling stimulus. Rothstock et al. [39] investigated different interior implant geometries, cemented and uncemented solutions predicting that an uncemented press-fit implant could limit bone resorption. A more complex study was performed by Dickinson et al. [40] where they simulated prosthesis-bone interface healing with bone remodeling observing the progressive gap filling at the implant–bone interface. To date, there are few published works on the effect of the cement mantle thickness and interface characteristics on HR [27,30,33]. Cementing technique affects not only cement penetration but also the initial stability of the femoral component [41]. Previous studies did not simulate bone remodeling [27,33], but analyzed the load transfer, i.e., strains for an immediate postsurgery situation. Additionally, little is known about the effect of the bone–cement and cement–implant interface conditions on bone remodeling. Bone–cement interface has always been assumed as completely bonded [27,30,33], and cement–implant interface was mainly assumed as bonded [27,33].
Therefore, the main goal of this study is to investigate the effect of different parameters on bone density increase and/or bone resorption evolution using bone remodeling on a threedimensional (3D) FEA of a resurfaced cemented prosthesis. In particular, different cement mantle thicknesses (0.25, 1 and 3 mm) and interface conditions between the components were varied. A previously developed phenomenological bone remodeling model was tested to predict the bone response post-operatively [42]. The methodology presented in this paper is intended to improve the understanding on previous parameters (cement mantle thickness and interface characteristics between components).
2. Materials and methods The 3D FEA model used in the present study was generated from CT scans of a single right male femur (46 years old). The medical images were segmented using Mimics software (Materialize, Leuven, Belgium) to obtain a personalized geometry of the femur. Finally, the proximal femur was reconstructed using Catia V5 (Dassault Systèmes, Suresnes, France) and the resurfacing prosthesis was implanted. The implant design is based on the clinically used Zimmer Durom implant. While this implant is no longer commercially available, owing to failures related to the acetabular cup, the femoral component geometry of this implant is similar to all the other implants, hence its use in this study. Arthroplasty simulation was oriented at 5◦ valgus with respect to the neutral axis line of the femoral neck (Fig. 1), as recommended by Amstutz et al. [16]. The resurfacing prosthesis was composed of a small and polished stem attached to the spherical component, hereinafter referred to as ‘the implant’. To investigate the effect of cement mantle thickness, three different models were created to represent different gaps between the femoral head and the inner face of the implant: 0.25, 1 and 3 mm keeping the stem geometry constant (diameter of 6 mm). One milimeter is the closest to a current clinically achieved cement mantle thickness [16]. The three different configurations were modeled for a constant femoral component with outer diameter of 50 mm and inner diameter of 42 mm. In reality, the cement is interdigitated with cancellous bone, however for simplicity this was modeled by a layer of cement. The 0.25- and 1-mm configurations ensured no notching of the femoral neck, although notching was necessary for implantation in the 3-mm configuration (Fig. 1). Four-noded tetrahedral solid elements were used in automatic finite element mesh generation with Harpoon v2 (Harpoon Sharc Ltd., Manchester, UK). The different models consisted of approximately 220,000 elements and 40,000 nodes each (Fig. 2). The element size used is inside the asymptotic region of convergence and represents a good trade-off between numerical accuracy and computational cost (results not shown). Bone tissue was considered to be anisotropic and heterogeneous. Mechanical properties of bone were predicted by means of a remodeling computational model, able to evaluate the evolution of bone porosity and anisotropy as a function of the mechanical conditions [42,43]. Following the scheme proposed by Doblaré and García [42] to simulate bone remodeling, three different forces that simulate the gait cycle were considered as loading conditions, specially the forces when the foot touches the floor and the other two alternative situations of abduction and adduction, respectively [42,43]. Hip contact forces were imposed as a uniform load on nodes of the femoral head surface (see Table 1). Only abductor muscle force was considered. Loading configuration, including hip-joint contact force and abductor force, can adequately reproduce in vivo loading [44]. Initially with the femur intact, all elements were considered as bone, presenting an initial homogeneous isotropic density of 0.5 g/cm3 (Fig. 3). After the application of the forces defined above
M.A. Pérez et al. / Medical Engineering & Physics 36 (2014) 185–195
187
Fig. 1. Section of the proximal femur showing (a) implant orientation at 5◦ valgus with respect to the neutral axis line of the femoral head, and (b) 3 different cement mantle thickness configurations.
Fig. 2. (a) Finite element model of a resurfacing prosthesis showing refined mesh density and the four regions used for analyzing the results (ROIL-neck lateral; ROIM-neck medial; ROI7-metaphysis medial; ROI1-metaphysis lateral. (b) Applied boundary conditions and load application points.
during 300 increments of analysis, the actual density distribution of the intact femur was obtained [42,43]. This is the time period needed to get convergent results as is explained in García et al. [45], where the convergence performance is studied in detail. Each increment of analysis represents one day [45].
To validate predictions of the bone remodeling model, predicted bone density distribution was compared with the density values obtained from CT scans and reported in Section 3. The femur FE mesh was imported into Mimics again and different material properties were assigned relating the bone mineral density with the
Table 1 Values of applied forces at the hip joint and abductor muscles for the 3 load cases considered. Case
1 2 3
Cycles/day
6000 2000 2000
Hip-joint
Abductor
X-axis
Y-axis
Z-axis
X-axis
Y-axis
Z-axis
−242.19 −664.20 268.80
−942.40 299.70 −1113.75
−2102.72 −900.00 −1040.85
181.89 59.90 128.97
330.04 343.21 361.01
593.10 −48.87 268.42
188
M.A. Pérez et al. / Medical Engineering & Physics 36 (2014) 185–195
Fig. 3. Steps followed for the finite element simulation.
Hounsfield Units (HU). There exist different relationships between apparent bone density and HU [46,47]. In the present study, the relationship proposed by Peng et al. [47] was used (computed for the femur = 1 + 7.185 × 10-4 YHU g/cm3 ). Then, the proximal part of the femur was removed to receive the prosthesis and cement, and the material properties of the corresponding elements were modified (Fig. 3). This methodology was previously validated. All bone remodeling analyses were performed with Abaqus software v.6.10 (Dassault Systèmes, Suresnes, France), incorporating the bone remodeling model in a UMAT user subroutine. The cement and prosthesis were modeled as linear elastic materials with Poisson’s ratio of 0.3 and Young’s modulus of 2800 MPa and 200 GPa, respectively [33]. The stem was modeled as debonded within bone (no friction: = 0.0). Contact elements were defined to simulate surface-to-surface contact between the stem and the bone. On the other hand, different interface conditions were studied for the bone–cement and cement–implant interfaces. When the comparison of the cement thicknesses was performed,
bone–cement and cement–implant interfaces were considered as bonded. Later, one case with debonded cement–implant interface with friction ( = 0.3) was studied as done by Pal et al. [29], and another case with both cement–implant and bone–cement interfaces debonded with friction ( = 0.3) [48] was analyzed. Contact elements were also defined to simulate surface-to-surface contact between the bone and the cement and the implant. During the simulations, the FE model was rigidly constrained at the distal femur (Fig. 2b). The bone mineral density (BMD) ratio was computed for the whole resurfaced femur and for the different regions shown in Fig. 2 [49]. These regions were identified following a previous work of the literature [49]. The proximal part of the femur was divided in 4 volumes, each one corresponding to the following region: ROILneck lateral, ROIM-neck medial, ROI1-metaphysis lateral and ROI7metaphysis medial) as Fig. 2 indicates. The BMD ratio of the whole resurfaced femur and each region was calculated as:
BMD ratio =
t 0 dV
VOL
(1)
Fig. 4. (a) Initial bone density distribution in the intact femoral head just before the surgery and (b) bone percentage volume under a certain strain level for the three most representative loading cases of walking (Table 1).
M.A. Pérez et al. / Medical Engineering & Physics 36 (2014) 185–195
189
Fig. 5. Equivalent strain distribution in the femoral head at different time increments (incs) without implantation and with the resurfaced prosthesis for the 3 cement mantle thicknesses with bonded bone–cement and cement–implant interfaces (M-medial, L-lateral). Transverse section of the proximal femur.
Fig. 6. Bone density distribution in the femoral head at different time increments (incs) without implantation and with the resurfaced prosthesis for the 3 cement mantle thicknesses with bonded bone–cement and cement–implant interfaces (M-medial, L-lateral). Transverse section of the proximal femur.
190
M.A. Pérez et al. / Medical Engineering & Physics 36 (2014) 185–195
Fig. 7. Mean percentage change in bone mineral density for the whole finite element model and for different four regions for the 3 cement mantle thicknesses with bonded bone–cement and cement–implant interfaces.
where t and 0 were bone density (g/cm3 ) at any iteration and the initial density after implantation, respectively, with VOL (cm3 ) being the total volume of the resurfaced femur or corresponding region. The equivalent strain (Eq. (2)) was also used to compare the load transfer among the different cases analyzed. (2)εeq =
(ε1 −ε2 )2 +(ε2 −ε3 )2 +(ε3 −ε1 )2 2
3. Results 3.1. Bone remodeling model validation (intact proximal femur) Initial bone density distribution before implantation, predicted with the bone remodeling model, is presented in Fig. 4a. In order to validate the bone remodeling model, the performance of the femur with its mechanical properties from the CT scans (HU) has been compared with the ones of the bone remodeling simulation. The femur with the mechanical properties computed from the apparent bone density values obtained from the CT scans (if ≤ 1.2 then E = 2014 2.5 MPA, = 0.2; if > 1.2 then E = 1763 3.2 MPA, = 0.32) [50] and the femur with the properties from the bone remodeling simulation were independently loaded with the three load cases considered for walking (Table 1). Then, the equivalent strain was computed at each integration point of the two approaches and the total volume of bone under certain equivalent strain was evaluated and represented (Fig. 4b) comparing both approaches. The results are very similar; therefore the bone remodeling model used
predicts accurately the bone apparent density obtained from CT scans (HU). 3.2. Influence of cement mantle thickness (with resurfaced prosthesis and bonded interfaces) The equivalent strain distribution for the intact femur and the resurfacing prosthesis, with three cement thicknesses was compared (Fig. 5). An important strain reduction was predicted mainly in the superior femoral head with the resurfacing prosthesis. There were less than 5% between the three cement thicknesses results. A cement mantle thickness of 0.25 mm modified load transfer between the implant and the surrounding bone. Then, higher strains in the inferior neck were predicted with a thinner (0.25 mm) cement mantle (Fig. 5). Higher strains also appeared close to the stem for the three cement thicknesses compared to the intact femur. Bone density evolution was predicted using the bone remodeling theory and has been shown in Fig. 6. As simulation evolved (number of time increment increased), a reduction in bone density in the proximal region (superior lateral and medial femoral head) was qualitatively estimated (Fig. 6). This effect was directly related to strain reduction previously shown in Fig. 5 (strain shielding). An opposite effect was noted along the stem where bone apposition was predicted (Fig. 6). The BMD ratio was computed for the four regions (as defined in Fig. 2) and represented in Fig. 7. Significant differences were clearly observed for the different Gruen regions. At ROIL and ROIM (Fig. 7), a reduction of the BMD ratio (approximately 95%) was predicted for
M.A. Pérez et al. / Medical Engineering & Physics 36 (2014) 185–195
191
Fig. 8. Equivalent strain distribution in the resurfaced femoral head at different time increments (incs) for the different interface conditions and without implantation (Mmedial, L-lateral, B–C bone–cement interface, I–C implant–cement interface). Results computed considering a resurfacing prosthesis with 1-mm cement thickness. Transverse section of the proximal femur.
a cement mantle thickness of 0.25 mm which represented strong bone resorption, and lower resorption was observed for ROIM with 1- and 3-mm thicknesses. At ROI1 and ROI7 a clear increase in the BMD ratio was predicted for the three cement thicknesses, approximately 110% and 112%, respectively. A cement mantle thickness of 1 mm produced the highest increase of the BMD ratio in these regions. When the global BMD ratio was calculated, few changes were found between the 1- and 3-mm thicknesses, where the BMD ratio rose by approximately 108% in both cases. The 0.25-mm configuration showed the lowest BMD ratio.
conditions (Fig. 10). Significant differences were observed among the three models studied and for the different four regions. The highest BMD ratio was predicted when both interfaces were debonded, and the lowest when both interfaces were completely bonded. At ROI1 and ROI7, small differences were observed among the three interface conditions analyzed. At ROIL and ROIM regions, an increase in the BMD ratio was predicted for the completely debonded configuration of approximately 113% and 114%, respectively. When the bone–cement interface was bonded and the cement–implant interface was debonded, the BMD ratio increased 106% and 110% at the ROIL and ROIM regions, respectively (Fig. 10).
3.3. Influence of the interface conditions The different interface conditions were studied for the prosthesis with a cement mantle thickness of 1-mm. Equivalent strain distribution has been represented in Fig. 8. Differences were predicted among the three interface conditions. Equivalent strains higher than the yield strain value (0.7%) were computed at the superior head (where the stem and the implant join) when both interfaces (bone–cement and implant–cement) were debonded with friction. Changing the interface conditions altered how the loads were transferred between the different components through the corresponding interfaces. When the bone–cement was bonded and the cement–implant was debonded, the equivalent strain was similar to the intact femur (Fig. 8). Lower bone apposition was predicted when both interfaces remained bonded (Fig. 9). Higher bone apposition was predicted when the bone–cement and cement–implant interfaces were debonded. This effect was directly related to the high strains at the superior head. The BMD ratio was computed in the four regions in order to compare the different interface
4. Discussion Resurfacing may present some clinical benefits over THA [3,4,7–9]. Some resurfacing designs presented good mid-term clinical results (Birmingham Hip Resurfacing-BHR) versus other with poor results (Durom, ASR, Cormet) [10–12]. Two major problems were detected: aseptic loosening (34.4%) and fracture of the proximal femur (31.9%) [13]. As it was indicated before, HR was indicated to young and active patients. The rate of return to sports after HR appears to be excellent and unequaled by conventional hip prostheses. High-impact sports seem to be compatible with HR, although no long-term studies have analyzed the impact of these activities on wear and/or aseptic loosening [51]. A better understanding of the implant design factors associated with success/failure is needed. There are several design factors (implant position, cement mantle thickness, stem design, etc.) that could be analyzed and, indeed, some of them have been investigated by similar computational methodologies as the one described here
192
M.A. Pérez et al. / Medical Engineering & Physics 36 (2014) 185–195
Fig. 9. Bone density distribution in the resurfaced femoral head at different time increments (incs) for the different interface conditions (M-medial, L-lateral, B–C bone–cement interface, I–C implant–cement interface). Results computed considering a resurfacing prosthesis with 1-mm cement thickness. Transverse section of the proximal femur.
[26,27,33,34]. These previous studies all neglected bone remodeling simulations. In the present paper, we analyzed the effect of an important design factor, the cement mantle thickness, on the bone remodeling process. A comparison of our results has been performed with other computational studies of the literature. Equivalent strain distribution was predicted (Fig. 5). Strain shielding was predicted in the superior femoral head, and similar results were demonstrated [25,27,30,33]. Higher strains appeared close to the stem as seen in Taylor [27]. Among the three cement thicknesses, little differences in equivalent strain were computed (Fig. 5); although higher strains in the inferior neck were predicted with a thinner cement mantle (0.25 mm) [33]. Bone resorption in the superior medial and lateral femoral head (Figs. 6 and 7) was predicted. Most previous computational studies showed stress shielding in these regions [25,28,33,34,37,52]. Similar results were predicted in the bone remodeling analysis performed by Gupta et al. [24]. Dickinson et al. [53] predicted neck narrowing to be linked to a shortened horizontal femoral offset in resurfaced hips compared with healthy joints. Additionally, Dickinson et al. [40] predicted a bone mineral density reduction at the superior head-neck junction with an internal-remodeling algorithm, which would be indicative of additional external remodeling of the neck cortex, which is thin in this region. On the other hand, bone formation was predicted along the prosthesis stem and at the stem tip: a global increase of approximately 10% in the BMD ratio was predicted with the present study Figs. 7 and 10). Predicted distal bone formation was also consistent with a densitometry study that discerned changes in bone density around a resurfaced prosthesis in vivo [7]. This effect is also consistent with the report of Pollard et al. [19] where augmented density of femoral neck trabeculation and cortical hypertrophy
were observed. Lian et al. [49] clinically evaluated proximal femoral bone remodeling by dual energy X-ray absorptiometry and concluded that the BMD ratio increased at ROI1 and ROI7, as predicted in the present study (Figs. 7 and 10). A general increment of the BMD ratio (6%) was noted in Lian et al. [49], and tended to rise in a short time period. This value is comparable to that predicted in our analysis (8% for 1-mm cement thickness and bonded interface, Fig. 7; and 10% for 1-mm thickness and bonded bone–cement interface and debonded bone–implant interface, Fig. 10). It is difficult to conclude whether an increase in BMD of the order of up to 10% is relevant or not given the fact that there are a number of errors and parameters that can influence the results. Two parameters were varied in the present study (cement mantle thickness and the contact interface conditions), and the BMD percentage increase predicted is an indication of the tendency for long-term results. The results of the present study show that a very thin cement mantle (0.25 mm) is detrimental to bone, producing significant bone resorption (Figs. 6 and 7). Such a thin mantle would probably cause damage in the bulk cement itself, leading to a high stress level that could affect its long-term performance (i.e. cement fatigue failure). Such a small cement mantle thickness seems more a theoretical possibility than practically feasible. Clinically, it would be very complicated to apply and maintain such a thin layer of cement. Good results were obtained (for both bone remodeling and cement stress) with 1-mm cement mantle thickness, as proposed by Amstutz et al. [16], whereas a 3-mm thick cement mantle showed adequate performance but could have detrimental consequences due to excessive cement thickness (thermal osteonecrosis or improper implant seating) [16,17,54]. A thicker cement mantle will increase the polymerization temperatures at the bone–cement interface, inducing thermal necrosis [54].
M.A. Pérez et al. / Medical Engineering & Physics 36 (2014) 185–195
193
Fig. 10. Mean percentage change in bone mineral density for the whole finite element model and for different four regions for the different interface conditions (B–C bone–cement interface, I–C implant–cement interface) with a cement mantle thickness of 1 mm.
In the present study, the effect of different interface conditions between the prosthesis components was studied. In all cases analyzed, the bone–stem interface was assumed as debonded without friction [29]. Indeed, Pal et al. [29] observed that changes in the bone–stem characteristics did not have any major consequences for the stress/strain pattern in the femoral head. In general, changes in the interface conditions directly affect to the load path, and as a consequence to the bone strain shielding and the bone density evolution. Changes in the implant–cement interface condition (bone–cement interface bonded) from bonded to debonded resulted in noticeable differences in strain distributions and periprosthetic bone adaptation within the resurfaced femoral head (Fig. 8 and 9). For a debonded implant–cement interface, the effect of strain shielding was considerably higher than for a debonded interface. In general, there is a change in the load transfer conditions, debonding the implant from the cement increases the strains and therefore the stimulus is higher increasing the bone formation, as it was observed in our results (Figs. 8 and 9). Assuming both interfaces (bone–cement and implant–cement) as debonded was not a real situation. This interface condition assumes that the cement is completely debonded between the bone and the implant increasing the strains locally. This situation may come from a failure during surgery. High equivalent strain values were locally predicted at the superior femoral head in the last condition (Fig. 8). In most computational studies, bone–cement interface was considered bonded. Only Pal et al. [35] computed the multi-axial Hoffmann failure criterion at the bone–cement interface concluding that this interface was secure against interface debonding. Thus, simulating a bonded bone–cement interface in future studies is reasonable.
4.1. Limitations Although the data are in good agreement with clinical findings [7,16,49] and other computational studies, there are some limitations of our research that deserve comment. The cement mantle was modeled as a uniform layer of pure bone cement. The nonuniformity of real cement mantle (see Fig. 1) has not been simulated [17]. Therefore, load transfer from the prosthesis to the bone has been slightly modified. Sometimes macro cement pocket is located on the implant internal dome surface giving effective implantcement bonding at least locally. Another limitation is related with the the bone–cement interface, there is cement interdigitation within bone which was not considered. Most of computational studies have assumed it as negligible [24,25,27,34,55]. The choice of cementing technique depends on various factors such as heat effects, fixation and curing time [54]. Depending on the cementing technique, the penetration of cement into cancellous bone could be different [56], changing the mechanical properties of the surrounding region. In our study, we considered the bone–cement interface as being completely bonded. This limitation can be solved by incorporating a mixed-mode failure model [57,58] that takes into account the mechanical properties of the bone–cement interface. The increased cement mantle was modeled by keeping the same femoral component diameter and reducing the femoral bone core. Perhaps it should have been modeled by enlarging femoral component size with the same prepared femoral bone. In addition, the curing conditions of bone–cement affect local interface and bone stresses. This will affect load transfer to the surrounding bone. These effects were not taken into account in the current study.
194
M.A. Pérez et al. / Medical Engineering & Physics 36 (2014) 185–195
Another limitation was related to the bone remodeling model used [42] that was unable to simulate external bone remodeling. Narrowing of the femoral neck after hip resurfacing arthroplasty has been clinically described and is mainly circumferential [18]; with the help of an external bone remodeling model, we would have been able to reproduce this effect. External bone remodeling models follow the idea to simulate the growth behavior of biological systems, known as CAO methods [45]. The advantage of using a bone remodeling model to predict the bone material properties with respect to a technique based on CT-scan data is that bone remodeling model takes into account porosity and anisotropy evolution [59]. A further limitation is that a single femur model was used to perform this analysis. In the authors’ opinion, however, this limit does not reduce the importance and generality of the obtained results. Inter-patient variability should be considered in future studies. In the present study, we analyzed walking loads and neglected other activities [36,39,53]. We have also assumed mean values for muscle forces and joint contact forces, ignoring their variability. The muscle insertions and application points are also an approximation to reality because the femur here considered is not exactly the same as the one used by the reference studies in the literature [60,61]. This fact could affect the bone remodeling predictions in the femur. A probabilistic methodology, like the one proposed by Pérez et al. [62], would be more appropriate in order to consider load variability. To improve these results and conclusions, a cement damage model could be incorporated to analyze whether the cement deterioration (cement cracking and cement fatigue properties) could affect the long-term performance of resurfacing prostheses. Additionally, an optimization procedure could have been previously developed to determine the threshold limits of the cement thickness here considered [63]. 5. Conclusion Based our numerical study, we can conclude that HR results in bone resorption at the superior femoral head and in bone apposition around the stem and at the stem tip for the different cement mantles investigated. From the results predicted, 1-mm cement mantle thickness may be an appropriate cement configuration. A bone–cement interface bonded and a cement–implant interface debonded combined with 1-mm cement mantle is numerically closest to the intact femur model analyzed. Therefore, we can summarize that the computational model proposed is a useful tool in order to determine the influence of different factors (cement mantle thickness and interface characteristics between components). It can be additionally adopted to study the influence of other design factors. Declarations Funding: The authors gratefully acknowledge the research support of the Spanish Ministry of Economy and Competitiveness through Research Project 2011-22413. Competing interests None declared. Ethical approval Not required. References [1] Steiger Rd, Miller L, Prosser G, Graves S, Davidson D, Stanford T. Poor outcome of revised resurfacing hip arthroplasty. Acta Orthop 2010;81:72–6.
[2] Prosser G, Yates P, Wood D, Graves S, Steiger Rd, Miller L. Outcome of primary resurfacing hip replacement: evaluation of risk factors for early revision. Acta Orthop 2010;81:66–71. [3] Daniel J, Pynsent P, McMinn D. Metal-on-metal resurfacing of the hip in patients under the age of 55 years with osteoarthritis. J Bone Joint Surg Br 2004;86:177–84. [4] Treacy R, McBryde C, Pynsent P. Birmingham hip resurfacing arthroplasty. J Bone Joint Surg Br 2005;87:167–70. [5] McMinn D, Treacy R, Lin K, Pynsent P. Metal on metal surface replacement of the hip. Clin Orthop 1996;329:89–98. [6] Kelley S, Lachiewicz P, Hickman J, Paterno S. Relationship of femoral head and acetabular size to prevalence of dislocation. Clin Orthop 1998;355:163–70. [7] Kishida Y, Sugano N, Nishii T, Miki H, Yamaguchi K, Yoshikawa H. Preservation of the bone mineral density of the femur after surface replacement of the hip. J Bone Joint Surg Br 2004;86:185–9. [8] Silva M, Lee K, Heisel C, Rosa MD, Schmalzried T. The biomechanical results of total hip resurfacing arthroplasty. J Bone Joint Surg Am 2004;86:40–6. [9] Girard J, Lavigne M, Vendittoli P, Roy A. Biomechanical reconstruction of the hip. A randomised study comparing total hip resurfacing and total hip arthroplasty. J Bone Joint Surg Br 2006;88:721–6. [10] Ng V, Arnott L, McShane M. Perspectives in managing an implant recall: revision of 94 Durom Metasul acetabular components. J Bone Joint Surg Am 2011;93:101–5. [11] Naal F, Pilz R, Munzinger U, Hersche O, Leunig M. High revision rate at 5 years after hip resurfacing with the Durom implant. Clin Orthop 2011;469:2598–604. [12] Cohen D. Out of joint: the story of the ASR. BMJ 2011;342:2905. [13] Klotz M, Breusch S, Hassenpflug M, Bitshc R. Results of 5 to 10-year followup after hip resurfacing: a systematic analysis of the literature on long-term results. Orthopade 2012;41:442–51. [14] Jameson S, Baker P, Mason J, Porter M, Deehan D, Reed M. Independent predictors of revision following metal-on-metal hip resurfacing: a retrospective cohort study using National Joint Registry data. J Bone Joint Surg Br 2012;94:746–54. [15] Murray DW, Grammatopoulos G, Pandit H, Gundle R, Gill HS, McLardy-Smith P. The ten-year survival of the Birmingham hip resurfacing: an independent series. J Bone Joint Surg Br 2012;94:1180–6. [16] Amstutz H, Campbell P, Duff ML. Fracture of the neck of the femur after surface arthroplasty of the hip. J Bone Joint Surg Am 2004;86:1874–7. [17] Morlock M, Bishop N, Ruther W, Delling G, Hahn M. Biomechanical, morphological and histological analysis of early failures in hip resurfacing arthroplasty. Proc Inst Mech Eng [H] 2006:220. [18] Hing C, Young D, Dalziel R, Bailey M, Back D, Shimmin A. Narrowing of the neck in resurfacing arthroplasty of the hip. J Bone Joint Surg Br Vol 2007;89:1019–24. [19] Pollard T, Baker R, Eastaugh-Waring S, Bannister G. Treatment of the young active patient with osteoarthritis of the hip. J Bone Joint Surg Br 2006;88:592–600. [20] Beckmann J, Goldapp C, Ringleff K, Schaumburger J, Grifka J, Perlick L. Cementing technique in femoral resurfacing. Arch Orthop Trauma Surg 2009;129:1317–25. [21] Krause M, Breer S, Hahn M, Ruther W, Morlock MM, Amling M, et al. Cementation and interface analysis of early failure cases after hip-resurfacing arthroplasty. Int Orthop 2012;36:1333–40. [22] Prendergast P. Finite element models in tissue mechanics and orthopaedic implant design. Clin Biomech 1997;12:343–66. [23] Kuhl E, Balle F. Computational modelling of hip replacement surgery: total hip replacement vs. hip resurfacing. Technische Mechanik 2005;25:1–9. [24] Gupta S, New A, Taylor M. Bone remodelling inside a cemented resurfaced femoral head. Clin Biomech 2006;21:594–602. [25] Wanatabe Y, Shiba N, Matsuo S, Higuchi F, Tagawa Y, Inoue A. Biomechanical study of the resurfacing hip arthroplasty. J Arthroplasty 2000;15:505–11. [26] Ong K, Kurtz S, Manley M, Rushton N, Mohammed N, Field R. Biomechanics of the Birmingham hip resurfacing arthroplasty. J Bone Joint Surg Br 2006;88:1110–5. [27] Taylor M. Finite element analysis of the resurfaced femoral head. Proc Inst Mech Eng [H] 2006;220:289–97. [28] Little J, Taddei F, Viceconti M, Murray D, Gill H. Changes in femur stress after hip resurfacing arthroplasty: response to physiological loads. Clin Biomech 2007:22. [29] Pal B, Gupta S, New A. A numerical study of failure mechanisms in the cemented resurfaced femur: effects of interface characteristics and bone remodelling. Proc Inst Mech Eng [H] 2009;223:471–84. [30] Pal B, Gupta S, New A. Influence of the change in stem length on the load transfer and bone remodelling for a cemented resurfaced femur. J Biomech 2010;43:2908–14. [31] Dickinson AS, Taylor AC, Jeffers JR, Browne M. Performance of the resurfaced hip. Part 2: the influence of prosthesis stem design on remodelling and fracture of the femoral neck. Proc Inst Mech Eng H 2010;224:841–51. [32] Long JP, Santner TJ, Bartel DL. Hip resurfacing increases bone strains associated with short-term femoral neck fracture. J Orthop Res 2009;27: 1319–25. [33] Radcliffe I, Taylor M. Investigation into the effect of cementing techniques on load transfer in the resurfaced femoral head: a multi-femur finite element analysis. Clin Biomech 2007;22:422–30. [34] Radcliffe I, Taylor M. Investigation into the effect of varus–valgus orientation on load transfer in the resurfaced femoral head: a multi-femur finite element analysis. Clin Biomech 2007;22:780–6.
M.A. Pérez et al. / Medical Engineering & Physics 36 (2014) 185–195 [35] Pal B, Gupta S, New A. Design considerations for ceramic resurfaced femoral head: effect of interface characteristics on failure mechanisms. Comput Methods Biomech Biomed Eng 2010;13:143–55. ˜ N. Anisotropic bone remodel[36] Caouette C, Bureau M, Vendittoli PMML, Nuno ing of a biomimetic metal-on-metal hip resurfacing implant. Med Eng Phys 2012;34:559–65. [37] Long J, Bartel D. Surgical variables affect the mechanics of a hip resurfacing system. Clin Orthop 2006;453:115–22. [38] Ong KL, Day JS, Kurtz SM, Field RE, Manley MT. Role of surgical position on interface stress and initial bone remodeling stimulus around hip resurfacing arthroplasty. J Arthroplasty 2009;24:1137–42. [39] Rothstock S, Uhlenbrock A, Bishop N, Laird L, Nassutt R, Morlock M. Influence of interface condition and implant design on bone remodelling and failure risk for the resurfaced head. J Biomech 2011:44. [40] Dickinson A, Taylor A, Browne M. Implant–bone interface healing and adaptation in resurfacing hip replacement. Comput Methods Biomech Biomed Eng 2012;15:935–47. [41] Bitsch R, Jäger S, Lürssen M, Loidolt T, Schmalzried T, Weiss S. The influence of cementing technique in hip resurfaceing arthroplasty on the initial stability of the femoral component. Int Orthop 2011;35:1759–65. [42] Doblaré M, García J. Application of an anisotropic bone-remodeling model based on a damage-repair theory to the analysis of the proximal femur before and after total hip replacement. J Biomech 2001;34:1157–70. ˜ [43] García J, Doblaré M, Cegonino J. Bone remodelling simulation: a tool for implant design. Comput Mater Sci 2002;25:100–14. [44] Stolk J, Verndonschot N, Huiskes R. Hip-joint and abductor-muscle forces adequately represent in vivo loading of a cemented total hip reconstruction. J Biomech 2001;34:917–26. [45] García J, Martínez M, Doblaré M. An anisotropic internal–external bone adaptation model based on a combination of CAO and continuum damage mechanics technologies. Comput Methods Biomech Biomed Eng 2001;4:355–78. [46] Rho J, Hobatho M, Ashman R. Relations of mechanical properties to density and CT numbers in human bone. Med Eng Phys 1995;17:347–55. [47] Peng L, Bai J, Zeng X, Zhou Y. Comparison of isotropic and orthotropic material property assignments on femoral finite element models under two loading conditions. Med Eng Phys 2007;28:227–33. [48] Janssen D, Mann KA, Verdonschot N. Micro-mechanical modeling of the cement–bone interface: the effect of friction, morphology and material properties on the micromechanical response. J Biomech 2008;41:3158–63. [49] Lian Y, Pei F, Yoo M, Chenq J, Fatou C. Changes of the bone mineral density in proximal femur following total hip resurfacing arthroplasty in osteonecrosis of femoral head. J Orthop Res 2008;26:453–9.
195
[50] Beaupre GS, Orr TE, Carter DR. An approach for time-dependent bone modeling and remodeling—theoretical development. J Orthop Res 1990;8:651–61. [51] Girard J, Miletic B, Deny A, Migaud H, Fouilleron N. Can patients return to highimpact physical activities after hip resurfacing? A prospective study. Int Orthop 2013;37:1019–24. [52] Huiskes R, Strens P, Heck Jv, Slooff T. Interface stresses in the resurfaced hip: finite element analyis of load transmissions in the femoral head. Acta Orthop Scand 1985;56:474–8. [53] Dickinson A, Taylor A, Browne M. Performance of the resurfaced hip. Part 1: the influence of the prosthesis size and positioning on the remodelling and fracture of the femoral neck. Proc Inst Mech Eng [H] 2010;224:841–51. ˜ N, Madrala A, García-Aznar J, Doblaré M. Computacional mod[54] Pérez M, Nuno elling of bone cement polymerization: temperature and residual stresses. Comput Biol Med 2009;39:751–9. [55] Taddei F, Martelli S, Gill H, Cristofolini L, Viceconti M. Finite element modelling of resurfacing hip prosthesis: estimation of accuracy through experimental validation. J Biomech Eng 2010;132:1–11. [56] Beaulé P, Matar W, Poitras P, Smit K, May O. 2008 Otto Aufranc Award: component design and technique affect cement penetration in hip resurfacing. Clin Orthop 2009;467:84–93. [57] Pérez M, Palacios J. Comparative finite element analysis of the debonding process in different concepts of cemented hip implants. Ann Biomed Eng 2010;38:2093–106. [58] Moreo P, Pérez M, García-Aznar J, Doblaré M. Modelling the mixed-mode failure of cement–bone interfaces. Eng Fract Mech 2006;73:1379–95. [59] Pérez M, Fornells P, Doblaré M, García-Aznar J. Comparative analysis of bone remodeling models with respect to computerised tomography-based finite element models of bone. Comput Methods Biomech Biomed Eng 2010;12: 101–8. [60] Bergmann G, Deuretzbacher G, Heller M, Graichen F, Rohlmann A, Strauss J. Hip contact forces and gait patterns from routine activities. J Biomech 2001;34:859–71. [61] Heller M, Bergmann G, Deuretzbacher G, Dürselen L, Pohl M, Claes L, et al. Musculo-skeletal loading conditions at the hip during walking and stair climbing. J Biomech 2001;34:883–93. [62] Pérez M, Grasa J, García-Aznar J, Bea J, Doblaré M. Probabilistic analysis of the influence of the bonding degree of the stem–cement interface in the performance of cemented hip prostheses. J Biomech 2006;39: 1859–72. [63] Vena P, Verdonschot N, Contro R, Huiskes R. Sensitivity analysis and optimal shape design for bone–prosthesis interfaces in a femoral head surface replacement. Comput Methods Biomech Biomed Eng 2000;3:245–56.
