MORPHOLOGY AND BIOMECHANICAL CHARACTERISTICS OF THE PROXIMAL FEMUR AFTER IMPACTION ALLOGRAFTING by HANSPETER FREI B.Sc, Bern Institute of Technology, 1994 M.Sc, University of Dundee, 1999 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Department of Mechanical Engineering) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA November 2003 © Hanspeter Frei, 2003 A B S T R A C T For millions of patients world-wide, primary total hip replacement (THR) is an effective way to improve quality of life. Failed THRs are often associated with extensive bone loss which makes the revision difficult. An established technique uses impacted morsellized allograft bone to reconstruct the proximal femur and ensure a rigid accommodation of the cemented revision component. There remains a fundamental lack of understanding of the impaction allografting procedure and its complications, particularly with its morphology and biomechanical characteristics. The objectives of this thesis were to i) describe the morphology after impaction allografting in the femur, ii) incorporate a computer simulation that should help the orthopaedic surgeon to control the morphology during surgery, iii) determine how the morphology affects the immediate strength of the host bone interface, and iv) develop an animal model to investigate the changes in composite morphology and strength with postoperative healing. Around the middle third of the stem, virtually the entire femoral canal was filled with cement, thereby forming a cement-allograft composite, whereas in other locations a pure allograft-host bone interface was found. The computer simulation suggested that cement penetration could be controlled by varying graft impaction and limiting cement volume injection. Cement penetration up to the endosteal surface significantly enhanced the host bone interface strength. In the animal study, the strength of the composite-host bone interface increased significantly at 3 weeks and was higher than the pure allograft construct. In contrast to the composite, the pure allograft construct failed at the cement-allograft interface. At 6 weeks the interface strength of the composite decreased, presumably due to cortical cancellation caused by damaged endosteal circulation. Cement penetration to the endosteal surface appears to be important for immediate postoperative clinical stability. However, the presence of the cement does not allow reconstitution of the host bone stock. Cortical cancellisation and medullary canal widening caused by a damaged endosteal circulation may be responsible for clinically unstable implants postoperatively. These findings suggest that the optimal reconstruction provide clinical stability without the , cement reaching the endosteal surface, thereby enabling revascularisation and subsequent bone remodelling. iii T A B L E O F C O N T E N T S A B S T R A C T ii T A B L E O F C O N T E N T S iv L IST OF T A B L E S vii L IST OF F I G U R E S viii P R E F A C E xii A C K N O W L E D G E M E N T S xiii C H A P T E R I INTRODUCTION 1_ 1.1 T O T A L HIP R E P L A C E M E N T S 2 1 . 2 M A N A G E M E N T OF FAILED F E M O R A L C O M P O N E N T S 4 1 . 3 IMPACTION A L L O G R A F T I N G 1 0 1.3.1 Technique 10 1.3.2 Stem subsidence 14 1.3.3 Incorporation and remodelling of the impacted cancellous allograft 20 1.4 S U M M A R Y 2 4 1 . 5 O B J E C T I V E S 2 6 1 .6 S C O P E 2 6 1 .7 R E F E R E N C E S 2 7 C H A P T E R II A L L O G R A F T IMPACTION A N D C E M E N T P E N E T R A T I O N A F T E R T H E REVIS ION OF F A I L E D T H R WITH IMPACTION A L L O G R A F T I N G : A HLSTOMORPHOMETRIC ANALYSIS IN THE CADAVERIC FEMUR 3 6 2 . 1 INTRODUCTION 3 7 2 . 2 M E T H O D S 3 9 2 . 3 R E S U L T S 4 4 2 . 4 D ISCUSSION 4 9 2 . 5 R E F E R E N C E S 5 3 C H A P T E R III C E M E N T F L O W DURING IMPACTION A L L O G R A F T I N G A FINITE ELEMENT ANALYSIS 5 6 3.1 INTRODUCTION 5 7 3.2 METHODS 5 9 3.3 RESULTS 6 7 3.4 DISCUSSION 72 3.5 REFERENCES 7 6 C H A P T E R IV M E C H A N I C A L C H A R A C T E R I S T I C S OF T H E B O N E - G R A F T - C E M E N T I N T E R F A C E A F T E R IMPACTION A L L O G R A F T I N G 7 8 4.1 INTRODUCTION 7 9 4.2 METHODS 81 4 .3 RESULTS 8 8 4 .4 DISCUSSION 9 5 4 .5 REFERENCES 102 C H A P T E R V B IOLOGICAL A N D M E C H A N I C A L C H A N G E S OF T H E B O N E - G R A F T - C E M E N T INTERFACE A F T E R IMPACTION A L L O G R A F T I N G 105 5.1 INTRODUCTION 106 5.2 METHODS - 109 5.3 RESULTS 115 5.4 DISCUSSION 122 5.5 REFERENCES 127 C H A P T E R VI G E N E R A L DISCUSSION A N D C O N C L U S I O N S 132 6.1 DISCUSSION 133 6.1.1 Morphology 134 6.1.2 Factors that determine morphology 136 6.1.3 Effect of morphology on mechanical characteristics 137 6.1.4 Temporal change of morphology and strength 144 6.1.5 Clinical significance 149 6.2 CONCLUSIONS 151 6.3 CONTRIBUTIONS 152 6.4 FUTURE WORK 153 6.5 REFERENCES 155 v A P P E N D I X A H ISTOLOGICAL P R O C E S S I N G P R O T O C O L S A N D S T A I N S 159 A P P E N D I X B T O R Q U E C E L L 163 A P P E N D I X C DRILL GUIDE F O R B O N E C H A M B E R 166 A P P E N D I X D P U S H - O U T L O A D - D I S P L A C E M E N T C U R V E S 171 A P P E N D I X E B O N E C H A M B E R T O R Q U E - R O T A T I O N C U R V E S 178 A P P E N D I X F S T R E S S C A L C U L A T I O N S 185 A P P E N D I X G A N O V A T A B L E S 187 vi L I S T O F T A B L E S Table 3.1: Cement volumes obtained from the cadaveric femur (CHAPTER II). 60 Table 3.2: Summary of the parameters for the performed simulations. 63 Table 3.3: Summary of the cement volumes for the different simulations. 71 Table 4.1: Regression parameters for the linear regression (y = a + b * %c) where %c is the percentage cement contact with the endosteal surface. 92 Table 4.2: Interface failures types for the composite/ graft-bone interface at different levels. 93 Table 4.3: Interface strength, stiffness, failure energy and ID grouped into the failure Type I, II and III. Mean and (SD). 93 LIST O F FIGURES Figure 1.1: Total hip replacement consists of a metal stem with a metal or ceramic ball and a metal cup with a UHMWPE liner, (adapted from www.jointreplacement.com 2002). 2 Figure 1.2: Endo-Klinik classification for femoral bone stock loss. Grade 1: Radiolucent lines confined to the upper half of the cement mantle. Grade 2: Generalized radiolucent zones and endosteal erosion of the upper femur leading to widening of the medullary cavity. Grade 3: Widening of the medullary cavity by expansion of the upper femur. Grade 4: Gross destruction of the upper third of the femur with involvement of the middle third. 5 Figure 1.3: Schematic illustration of the proximal reconstruction of the femur with a structural allograft, (adapted from Hejna, 1998). 7 Figure 1.4: Impaction allografting according to Gie et al. (Gie et al., 1993a). A: Distal plug with guide wire. B and C: Allograft impaction with distal impactor up to the distal impaction line. D: Allograft impaction with proximal impactor. E: Neo-medullary canal after impaction with proximal, block and 'half moon' impactors. F: Cement pressurisation after retrograde filling. G: Photograph of the medullary canal with a osteolytic cortex after cement and stem were removed. H: Photograph showing the filling of the canal with morsellized allograft. I: Photograph of the neo-medullary canal. J: Impaction allografting construct after the procedure. 11 Figure 1.5: Insufficient graft impaction might result in stem subsidence through graft consolidation and/or shear failure of the graft layer. 16 Figure 1.6: Stem subsidence through host bone and/or cement-allograft interface failure. 17 Figure 1.7: Stem subsidence within the cement mantle due to cement mantle fracture. 18 Figure 1.8: Schematic representation of the four zones observed in autopsies and biopsies. Zone thickness and distribution around the stem vary substantially (Linder, 2000). 21 Figure 2.1 Schematic drawing of the specimens shows the fourteen levels that were matched based on anatomical landmarks. The levels were grouped according to Gruen zones (Gruen et al., 1979). 40 Figure 2.2: Photograph and schematic drawing of five sections in all Gruen zones from a typical specimen. The different patterns in the schematic drawing indicate the materials present in each section. Each section was divided into anterior, posterior, medial and lateral quadrants. 42 Figure 2.3: Particle size distribution of the allograft chips used in this experiment in a semi-logarithmic plot. The ordinate indicates the percentage by weight of particles smaller than the size given by the abscissa. For example, as indicated by the dashed line, 50% of the particles used in this experiment are smaller than 4mm. 44 Figure 2.4: Graft porosity as a function of levels. The error bars indicate the standard deviation. The dashed line represents the porosity of the unimpacted graft. 45 Figure 2.5: Correlation of average impaction force with graft porosity for the proximal and distal impactors. 46 Figure 2.6: Mean distance of the cement mantle, cement penetration and endosteal cortex from the stem surface. The error bars indicate the standard deviation. The standard deviation for the endosteal cortex is not shown for convenience. The guide wire caused the cement mantle in Gruen zone 4 (i.e. Levels 13 and 14). 47 Figure 3.1: Medial-lateral view and sagittal cross-section of the finite element model indicating the site of cement injection and locations where cement leakage was simulated around the seal. 59 Figure 3.2: Porosities and hydraulic conductivities as a function of different levels along the simulated cadaveric femur. The hydraulic conductivities were calculated based on intrinsic permeability data from Beaudoin et. al. 1991. The hydraulic conductivity profile along the femur is determined by the porosity. The hydraulic conductivity decreases with increased cement viscosity. 63 Figure 3.3: Cement intrusion depths as a function of bone porosity into cylindrical bone plugs. The predicted intrusion depths were on average 1mm lower than the experimental data. 64 Figure 3.4: Cross-sections along the femur comparing the measured and predicted cement intrusion. Specific locations are shown in Figure 3.2. Where the cement intrusion increased circumferentially an accurate prediction was possible. 67 Figure 3.5: Evolution in time of the intrusion profile of the mean intrusion depths along the femur for the simulated cadaveric femur. For comparison the measured intrusion profile is shown. 68 Figure 3.6: Pressure gradients along the stem during cement pressurisation at different times of cement pressurisation. After 10s only small changes of the pressure gradient were predicted. The mean pressures measured at three locations along the femur during cement pressurisation during a primary THR's are included for comparison (McCaskie et al., 1997). 69 Figure 3.7: Intrusion profiles of the mean intrusion depths along the femur for the different simulation after stem insertion. 70 Figure 4.1: Schematic drawing of the specimens and the push-out setup. The sections from six specimens were matched to seven levels based on anatomical landmarks and the tip of the stem. The sections were tested in the setup shown on the right. 83 Figure 4.2: Photograph of the three failure modes. Type I failure was a pure interface failure. In Type II failures, remnants of cement remained on the endosteal surface indicating a local al log raft-cement composite failure. Allograft particles remained on the endosteal surface in Type III failures, indicating a local allograft layer failure. 85 Figure 4.3: Typical load-displacement response. The push-out tests were characterized by the initial stiffness calculated between 20% and 80% of the peak load, the peak load, and the failure energy. 88 Figure 4.4: Percentage cement contact %c with the endosteal surface as a function of different levels. The error bars indicate the standard deviation. Highest percentage cement contact was found proximally and then gradually decreased to a pure allograft-bone interface below the tip of the stem. 89 Figure 4.5: Shear T and tension a stresses at failure as a function of levels. The error bars indicate the standard deviation. Lowest shear and tension stresses were found around the tip of the stem. 90 IX F i g u r e 4.6: Scatterplot of shear x and tensile o- stresses at failure as a function of percent cement contact %c. The failure stresses increased with increased percentage cement contact. The lines indicate linear regressions. The regression parameters are summarized in Table 4.1. 91 F i g u r e 4.7: Scatterplot of the apparent interface strength T a p p as a function of interface stiffness k. A moderate correlation was found between strength and stiffness. The line indicates the linear regression. 94 F i g u r e 4.8: Comparison of the apparent interface shear strength T a p p after impaction allografting with interface strength of a primary THR a cemented revision and a cemented re-revision. The interface strengths of after impaction allografting were between a cemented revision and re-revision. The values for the primary THR, the revision and re-revision were adapted from Dohmae et al. (Dohmae et al., 1988). 97 F i g u r e 4.9: Mohr-Coulomb failure envelopes for impacted bone graft, the pure allograft-host bone interface and the composite-host bone interface. The proximal cement-bone interface stresses were determined by finite element analysis of primary total hip replacement for level walking, stair ascent and stair descent activities (Chang et al., 1998). The pure allograft interface failure envelope was estimated from the averaged measured apparent shear strength (0.28MPa) and assuming that the tensile strength does not exceed the maximum measured tensile stress (0.12MPa). Similarly, the composite interface (%c>90%) failure envelope was based on the averaged measured apparent shear strength (1.22MPa) and the maximum measured tensile stress (0.26MPa), which was identical to the tensile strength for the cement-bone interface of primary THRs with no cement interdigitation (Mann et al., 2001). Tick marks on axes indicate 1MPa. 99 F i g u r e 5.1: Schematic diagram and photograph of the rat bone chamber model. The materials and interfaces observed after impaction allografting are re-established in a bone chamber. The endosteal circulation is impaired by a tight fit of the bone chamber to the endosteal surface. 109 F i g u r e 5.2: Typical load-displacement response. The torsion tests were characterized by the initial stiffness calculated between 20% and 80% of the peak load, the peak load, and the failure energy. 113 F i g u r e 5.3: Histology of the pilot study group at 3, 6 and 12 weeks. Four different zones of healing and repair were observed: Z o n e 1: Remodelled viable bone less dense than the allograft bone surrounded by bone marrow. Z o n e 2: Active remodelling front of woven bone. Z o n e 3: Necrotic allograft bone with or without fibrous tissue ingrowth. Z o n e 4: Inflammatory response. 116 F i g u r e 5.4: Histology and histomorphometry for the pure allograft and cement-allograft composite at 0, 3 and 6 weeks. Graphs showing the mean allograft porosity, cortical porosity, percent fibrous tissue and cortical thickness. Allograft porosity in the pure allograft group was significantly lower at 6 weeks (p<0.001). The cortical porosity and cortical thickness increased significantly at 3 and 6 weeks (p<0.002). The percentage fibrous tissue was significantly higher in the pure allograft at 3 and 6 weeks compared with the composite (p<0.005). 118 F i g u r e 5.5: Interface strengths of the allograft and composite at 0, 3 and 6 weeks (mean; SD). The interface strength was significant higher for both the allograft and composite at 3 weeks. The composite strength was significantly higher than the pure allograft (p=0.03). The pure allograft failed in general at the allograft-cement interface and the composite at the host bone interface. 120 x Figure 5.6 Allograft and composite stiffness of the allograft and composite at 0, 3 and 6 weeks (mean; SD). The stiffness did not change with time. The pure allograft stiffness was significantly higher than the composite (p=0.03). 121 Figure 6.1: Proposed modification of the impaction allografting technique. A: Distal plug with guide wire. B to D: Allograft impaction with flattened impactors similar in shape like the distal impactors. E: Drilling of the neo-medullary canal with a tapered reamer. E: Reaming of the canal. G: Femur-allograft construct ready for cement and stem insertion. 135 Figure 6.2: Radial and axial stress distribution of a primary THR with tensile separation (i.e. no tensile loads are transferred) at the cement-cancellous bone interface determined by a two dimensional finite element model. Note: Fully bonded stem and different scales for radial and axial stresses. (Adapted from Weinans et al., 1990). 138 Figure 6.3: Simplified model of the stem-cement-allograft-bone complex around the middle third of the stem to investigate the effect of different cement/ composite layer thicknesses on stem subsidence and allograft and host bone interface failure. It was assumed that the cement/ composite is a thick-walled tube exposed to internal and external stresses (Timoshenko and Goodier, 1987). The internal 20MPa and external stress 3.3MPa were obtained from compressive stress predictions by finite element analysis for the stem-cement and the cement-bone interface of primary THR. (Chang et al., 1998; Norman et al., 2001). The cement/ composite thickness were modelled from 0.5 to 6.75mm with corresponding allograft layer thicknesses of 6.5 to 0.25mm (CHAPTER II). 140 Figure 6.4: Predicted stem subsidence as a function of cement mantle/ composite thickness. With increasing cement mantle thickness, expansion of the inner diameter decreases resulting in less stem subsidence. With the entire section filled, the stem did not subside more than 1.3mm. 141 Figure 6.5: Uniaxial compressive stress in the allograft and the host bone interface as a function of cement mantle/ composite thickness. The patterned areas are required compressive stresses in the allograft or host bone interface to sustain a shear stress of 2.2MPa. The failure criteria are based on Brewster et al. and on the interface strength determined in CHAPTER IV (Brewster et al., 1999). The stress is increasing with decreasing allograft layer thickness. Cement mantle of 2mm which is considered as intact would result in shear failure of the host bone interface and the allograft layer (Masterson etal., 1997). No failure would occur with graft layer equal and smaller than 0.3mm. 142 Figure 6.6: Locking of the cement-allograft in the tapered medullary may prevent stem-composite subsidence. 144 Figure 6.7 Schematic drawing illustrating medullary canal widening as observed after primary THR. This phenomenon presumably is caused by a damaged endosteal circulation and may cause stem-cement-allograft construct subsidence after impaction allografting. 148 xi P R E F A C E Sections of this thesis have been submitted as multi-authored papers in refereed journals. Details of the authors' contribution are provided. I agree with the stated contributions of the thesis author, as indicated below. Dr. Thomas R. Oxland (thesis supervisor) Papers in review Frei H, Mitchell P, Masri BA, Duncan CP, Oxland TR. (November 2002) Allograft Impaction and Cement Penetration after the Revision of Failed THR with Impaction Allografting: A Histomorphometric Analysis in the Cadaveric Femur, Journal of Bone and Joint Surgery British Volume. Authors' contribution: Hanspeter Frei was responsible for the original ideas behind the paper, conduction of the experiments, analysis and presentation of the findings, and writing and editing of the original paper. Thomas Oxland was the key editor on this paper. Philip Mitchell, Bassam Masri and Clive Duncan conducted the surgical procedures on the cadaveric femora, stimulated discussion and provided editorial assistance. Frei H, Mitchell P, Masri BA, Duncan CP, Oxland TR. (June 2003) Mechanical Characteristics of the Bone-Graft-Cement Interface after Impaction Allografting, Journal of Orthopaedic Research. Authors' contribution: Hanspeter Frei was responsible for the original ideas behind the paper, conduction of the experiments, analysis and presentation of the findings, and writing and editing of the original paper. Philip Mitchell, Bassam Masri and Clive Duncan conducted the surgical procedures on the cadaveric femora, stimulated discussion and provided editorial assistance. Thomas Oxland was the key editor on this paper. Papers submitted Frei H, Gadala M, Masri BA, Duncan CP, Oxland TR. (October 2003) Cement Flow during Impaction Allografting: A Finite Element Analysis, Journal of Biomechanics. Authors' contribution: Hanspeter Frei was responsible for the original ideas behind the paper, conduction of the experiments, analysis and presentation of the findings, and writing and editing of the original paper. Mohamed Gadala provided technical advice and editorial assistance. Bassam Masri and Clive Duncan stimulated discussion and provided editorial assistance. Thomas Oxland was the key editor on this paper. Frei H, O'Connell J, Masri BA, Duncan CP, Oxland TR. (October 2003) Biological and Mechanical Changes of the Bone-Graft-Cement Interface after Impaction Allografting. Journal of Orthopaedic Research Authors' contribution: Hanspeter Frei was responsible for the original ideas behind the paper, conduction of the experiments, analysis and presentation of the findings, and writing and editing of the original paper. John O'Connell provided advice in the interpretation of the histological sections and editorial assistance. Bassam Masri and Clive Duncan stimulated discussion and provided editorial assistance. Thomas Oxland was the key editor on this paper. xii ACKNOWLEDGEMENTS This thesis is dedicated to my wife Corina for her support, her patience and her love. Without the support and guidance from my supervisor Dr. Tom Oxland, this thesis would not be possible. Even when my ideas and experiments did not turn out the way I expected, he managed to keep me motivated. It was always a great pleasure and privilege to work under his supervision throughout my entire biomechanics career. I wish to thank Dr. Clive Duncan for taking the time to participate in the experiments, for his valuable points of view and for helping me to put my research in perspective of the overall problem. Thanks to Drs. Bas Masri and Phil Mitchell for their participation in the experiments, for organising the surgical tools and implants, and for their clinical knowledge and input. The assistance I received from Dr. John O'Connell with the interpretation of the histological sections, from Dr. Mohamed Gadala with the finite element model was most valuable. Thanks to Dr. Goran Fernlund for his assistance and inspiring discussion. I would like to thank Dr. Jesse Chen for his assistance during surgery and for his skillful preparation of the histological sections. The preparation techniques of xiii bone tissue for histology were learned from Mr. Kevin Gibbon and Mr. George Spurr, which saved me months of experimenting. Thanks to Carolyne, Markus, Anthony, Juay Seng, Simon and the other members of the Division of Orthopaedic Engineering Research for the inspiring discussions. I gratefully acknowledge the George W. Bagby Research Fund, the Canadian Institutes of Health Research, Prof. Ewald Weibel and Prof. Maurice Muller from the Maurice E. Muller Foundation, the Swiss Academy of Engineering Science, and the Robert Mathys foundation for research funding. In addition I thank Mr. Terry Murphy of Stryker Howmedica and Mr. Joe Kudzin of DePuy Canada for providing implants and instruments. xiv CHAPTER I INTRODUCTION INTRODUCTION 1.1 TOTAL HIP REPLACEMENTS Total hip replacement (THR) is a successful treatment for osteoarthritis, rheumatoid arthritis, congenital deformities and traumatic fractures of the hip. This operation gives millions of patients worldwide pain relief and an increased quality of life. In Canada over 18,000 primary THRs were performed in the year 2001 (Canadian Institute for Health Information, 2003). With an average direct hospital cost of US$27,500 per primary THR, the economic burden in the United States exceeded US$5 billion for the 185,000 hip replacements performed in the same year (Agency for Healthcare Cost and Quality, 2003). A THR consists of a metal stem which is cemented (polymethyl-methacrylate (PMMA) cement) or pressfit without cement into the femoral canal. A metal or ceramic head placed on the stem forms a ball and socket joint with an ultra high molecular weight polyethylene (UHMWPE) liner fixed on the pelvic side by a metal cup. Alternatively, the liner might be ceramic or metal and the cup cemented or uncemented (Figure 1.1). Figure 1.1: Total hip replacement consists of a metal stem with a metal or ceramic ball and a metal cup with a UHMWPE liner, (adapted from www.jointreplacement.com 2002). 2 INTRODUCTION Total hip replacements have a finite lifetime. Depending on implant type, survival rates of cemented THRs are around 95% and for uncemented THRs 88% at 10 years (Malchau et al., 2002). Revisions of failed THRs accounted for 11% of all THRs performed in Canada in 2001 and 18% in the U.S. (Agency for Healthcare Cost and Quality, 2003; Canadian Institute for Health Information, 2003). The average direct cost of revision procedures ranged from US$31,000 to US$51,000 depending on the complexity (Barrack et al., 1999). With an aging population and the tendency to perform joint replacements in younger more demanding patients, the number of patients requiring revision of THRs is continuing to increase (Duncan etal., 1998). The main reason for the failure of THRs is aseptic loosening of the components caused by a biological reaction to sub-micron wear debris generated mainly from the UHMWPE cup. The debris migrates into the interface between the bone and the adjacent implant components (i.e. cement, stem) and activates cells called macrophages. The macrophages release cytokines after ingestion of the wear particles which triggers a response leading to bone absorption (osteolysis) and subsequent loosening of the components. It has been shown in animal models that osteolysis can occur also in the absence of wear debris by applying fluctuating pressure to the bone (Van der Vis H.M. et al., 1998; Aspenberg and Van der Vis, 1998). Fluctuating fluid pressures in the femur that are generated during patient activity may cause osteolysis and the resultant fluid flow may distribute the wear particles along the interfaces (Manley et al., 2002). 3 INTRODUCTION Extensive research worldwide is focusing on the development of better wear-resistant bearing materials, the sealing of the bone-cement or bone-implant interfaces, and on the suppression of bone absorption to further extend the lifetime of THRs (Shanbhag et al., 1997; Rahbek et al., 2000; Chiesa et al., 2000). In contrast, this thesis is addressing the treatment of failed THR, and in particular, the revision of failed femoral components since loosening of the stem occurs more frequently than the acetabular cup in primary and revision THRs (Duncan etal., 1998; Malchau etal., 2002). 1.2 MANAGEMENT OF FAILED FEMORAL COMPONENTS The management of failed THRs is multifaceted. Issues such as identifying risk groups, patient selection, surgical techniques, outcome assessment and pre- and postoperative care are important factors in the success of the procedure. It is beyond the scope of this thesis to discuss all these topics and therefore only the surgical options are discussed. The revision of failed THR with associated extensive osteolysis is a challenging and demanding task for the orthopaedic surgeon. The degree of osteolytic bone loss, which needs to be assessed pre-operatively, is an important consideration in the surgical treatment along with other factors such as the patient condition and the surgeon's preference. The femoral bone loss is frequently classified with the Endo-Klinik classification (Engelbrecht and Heinert, 1987) (Figure 1.2). 4 INTRODUCTION Grade 1 Grade 2 Grade 3 Grade 4 Figure 1.2: Endo-Klinik classification for femoral bone stock loss. Grade 1: Radiolucent lines confined to the upper half of the cement mantle. Grade 2: Generalized radiolucent zones and endosteal erosion of the upper femur leading to widening of the medullary cavity. Grade 3: Widening of the medullary cavity by expansion of the upper femur. Grade 4: Gross destruction of the upper third of the femur with involvement of the middle third. Considering the large loads to which THRs are subjected (up to 5 times body weight during walking (Bergmann et al., 1993)), rigid mechanical fixation of the revision components in the remaining bone is difficult to achieve. The quality of the fixation is certainly influenced by the degree of bone deficiency, particularly for Grades 3 and 4 (Leopold etal., 2000). The surgeon is faced with several options to revise failed THRs. These include recementing of a standard or longer stem, the use of a long cementless implant, the replacement of the entire proximal femur with a structural graft from a donor, Girdlestone arthroplasty and impaction allografting (Duncan et al., 1998; 5 INTRODUCTION Leopold and Rosenberg, 2000). The clinical results with these techniques are briefly summarized below. Cemented femoral revisions are inferior to primary THRs with failure rates ranging from 6% to 36% at ten years. A complicating factor is that subsequent re-revision may be more challenging (Mulroy and Harris, 1996; Katz et al., 1997; Hultmark et al., 2000; Malchau et al., 2002; Davis, III et al., 2003). However, in elderly patients with only minor bone loss (Grade 1 and 2), cemented revision with long stems appears to be a durable option (Hultmark ef al., 2000). Excellent clinical results have been reported for cementless femoral revisions with survival rates up to 96% after a follow up between 11 to 16 years (Bohm and Bischel, 2001; Weeden and Paprosky, 2002). Subsiding, clinically unstable implants (up to 45mm) are the major complication associated with uncemented stems. Bohm et al. reported that in 88% of the revised hips with severe proximal bone loss, radiographs showed good or excellent restoration of the proximal part of the femur with Wagner SL stem (Centerpulse Orthopedics Inc., Baar, Switzerland) (Bohm and Bischel, 2001). However, stem subsidence of more than 10mm in 20% of the cases has been reported for this particular design (Grunig et al., 1997; Bohm and Bischel, 2001). With larger proximal defects, rigid fixation of the cementless stems can be achieved only by using longer stems to obtain a pressfit in the distal femoral diaphysis. This may not be possible when the defects extend to the diaphysis and in Grade 4 defects. If the distal fixation is insufficient, the stems may subside and micromotions may prevent bone growth 6 INTRODUCTION onto the stem surface. In the case of failure of the cementless revision, the removal of a partially fixed stem might be difficult and make a re-revision more challenging (Schmidt et al., 2002; Davis, III et al., 2003). No clear advantage has been reported by adding cancellous allograft to cementless femoral revision (Leopold et al., 2000). Severe proximal circumferential bone loss of the femur of more than 5cm in length may be not possible to treat with the above mentioned revision techniques. An alternative is to use a proximal femoral structural allograft cemented to a long stem prosthesis to restore the integrity of the proximal femur (Figure 1.3) (Duncan etal., 1998; Blackley etal., 2001). Although these structural allografts never reconstitute the host bone stock, success rates of 77% at eleven years have been reported for this procedure (Enneking and Mindell, 1991; Blackley et al., 2001). However, this technique is only used in the most complex cases where the patient underwent multiple hip operations and has few other options. F i g u r e 1.3: Schematic illustration of the proximal reconstruction of the femur with a structural allograft, (adapted from Hejna, 1998). 7 INTRODUCTION Girdlestone arthroplasty involves extensive resection or excision of the femoral head and neck. This salvage procedure for failed THR with extensive proximal femoral bone loss provides reasonable pain relief but poor function (Duncan etal., 1998). Impaction allografting is a technique for revision hip reconstruction that was introduced by Slooff et al. in 1984 for the acetabulum and by Gie et al. in 1993 for the femoral side. The method involves the impaction of morsellized allograft bone into a cavity to restore the original bone stock and provide the femur with sufficient strength to rigidly accommodate the replacement component (Slooff et al., 1984; Gie et al., 1993b). Promising survival rates of greater than 88% for short and intermediate term follow-up (minimum 2 to 4 years) have been reported (van Biezen et al., 2000; Pekkarinen et al., 2000; Leopold et al., 2000; Fetzer et al., 2001; de Roeck and Drabu, 2001; Ornstein ef al., 2001; Lind et al., 2002; Nelissen etal., 2002; Piccaluga etal., 2002; Kligman etal., 2002; Ornstein ef al., 2002; Robinson et al., 2002; Ullmark ef al., 2002a). A survivorship of 90.5% at 10 to 11 years was recently reported by the developer of the technique (Halliday ef al., 2003). However, clinical performance and patient outcome are difficult to compare since the surgical technique, inclusion criteria (degree of bone loss), stem geometry, allograft material, cement used and postoperative care vary among different studies and may affect the outcome (Leopold et al., 2000). 8 INTRODUCTION Impaction allografting is associated with a high incidence of major complications that ranges from 8% to 28% including intra- and postoperative fracture, prosthetic dislocation, trochanteric non-union (when trocanteric osteotomies were performed) and stem subsidence (Leopold et al., 2000). The major concerns associated with this technique are clinically unstable, subsiding stems. Stem subsidence up to 31mm has been reported and may cause thigh pain, cement mantle fracture, and prosthetic dislocation (van Biezen et al., 2000; Leopold et al., 2000; Kligman et al., 2002). Although the durability and the indications for this type of reconstruction are controversial, the impaction allografting technique appears to have great potential (Leopold and Rosenberg, 2000; Pekkarinen et al., 2000; Boldt et al., 2001; Fetzer et al., 2001). The reconstruction can be customized to the specific case, the osteolytic bone loss can be restored to facilitate re-revision, and in severe cases impaction allografting may be one of the few options that the patients have. From a mechanical and biological point of view the clinical performance of impaction allografting can be improved by ensuring clinically stable stems and complete restoration of the host bone stock. The next section explains the impaction allografting technique in greater detail and explores other issues with this technique, potential mechanisms leading to clinically unstable implants and the restoration of the host bone stock. 9 INTRODUCTION 1.3 IMPACTION ALLOGRAFTING 1.3.1 Technique The impaction allografting technique used in the published follow-up studies is similar to that described by Gie et al. and is often performed with the X-change revision system (Howmedica Inc, Rutherford, NJ) (Gie et al., 1993a). After the failed stem and the cement have been removed, often with the assistance of a high speed burr, the intramedullary canal is occluded 2cm distal to the most distal lytic defect with an intramedullary plug. A guide wire secured to the intramedullary plug guides the impactors (Figure 1.4A and 1.4G). Two to four fresh frozen femoral heads are morsellized in a bone mill producing bone chips in the range from 2 to 10mm. The intramedullary canal is partially filled with the morsellized allograft bone and then impacted with distal circular impactors (Figure 1.4B and 1.4H). The impaction force is applied with a sliding hammer which is a part of the X-change system. If an adequate impaction is achieved, additional allograft bone chips are inserted and again impacted. This procedure is repeated until the depth of the stem tip on implantation (distal impaction line) is reached (Figure 1.4C). The intramedullary canal is again filled with bone chips and impacted with an impactor that has the shape of an oversized stem (Figure 1.4D). With this procedure a neo-medullary canal is formed which can accommodate a normal sized polished cemented stem (Figure 1.4E and 1.41). The most proximal impaction is achieved with block and 'half moon' impacters to ensure rotational stability of the stem. If adequate impaction has been achieved, 10 INTRODUCTION two packages of low viscosity Simplex cement (Howmedica Inc, Rutherford, NJ) are mixed and injected into the neo-medullary with a cement gun. A proximal femoral seal is then applied and the cement pressurized (Figure 1.4F) until the viscosity of the cement is appropriate for the insertion of the double tapered polished stem. The construct consists of a cortical bone shell, impacted allograft, bone cement and a metal stem after the impaction allograft procedure (Figure 1.4J). Figure 1.4: Impaction allografting according to Gie et al. (Gie ef a/., 1993a). A: Distal plug with guide wire. B and C: Allograft impaction with distal impactor up to the distal impaction line. D: Allograft impaction with proximal impactor. E: Neo-medullary canal after impaction with proximal, block and 'half moon' impactors. F: Cement pressurisation after retrograde filling. G: Photograph of the medullary canal with a osteolytic cortex after cement and stem were removed. H: Photograph showing the filling of the canal with morsellized allograft. I: Photograph of the neo-medullary canal. J: Impaction allografting construct after the procedure. 11 INTRODUCTION This technique is demanding and time-intensive, particularly in severe cases (Ling, 1997). The morsellized allograft needs to be impacted as much as possible to maximise the shear strength of the allograft layer (Brewster et al., 1999). There is a risk that the maximum force during graft impaction may fracture the femur and clearly this depends on the degree of osteolysis and the quality of the remaining cortical shell. The variability in impaction force is likely to create highly variable allograft morphology. To address some of these issues, a radial impaction grafting technique was developed to reduce intraoperative fractures, achieve more reliable and reproducible graft impaction, and allow the use of different implant lengths to facilitate distal fixation (Stulberg, 2002). Only a limited number of patients have been treated with this new technique. Intermediate term clinical results were similar to that reported for the original technique (Hostner et al., 2001; Stulberg, 2002). The major complications were postoperative fractures in two out of fourteen cases (Stulberg, 2002). Stem subsidence was not reported for this series. The possible shortage of allograft bone supply and the risk of disease transmission have led to the exploration of alternatives for allograft bone. Tricalcium phosphate (TCP), hydroxyapatite (HA) and silicate-free bioactive glasses have been investigated in vitro and in animal models for use as substitutes for morsellized allograft (Verdonschot et al., 2001; Griffon et al., 2001a; Griffon et al., 2001b; Bolder et al., 2002; Pratt et al., 2002). Oonishi et. al. used impacted HA particle for the revision of failed acetabular cups with massive 12 INTRODUCTION defects with good clinical results at a minimum follow up of 4 years (Oonishi ef al., 1997). Direct bonding of HA particles with the bone were observed radiologically without morphological changes. The same group is using HA particles smeared to the inner surface of the prepared femoral canal in cemented primary THRs with no femoral component loosening at a minimum follow up of 6 years (Oonishi et al., 2001). It has been shown in rabbits that the bone-cement interface can be enhanced with this technique, but the HA particles are not replaced by bone and therefore contribute only partially to the restoration of the host bone stock (Oonishi ef al., 2000). The synthetic bone graft substitutes available do not combine the strength and osteoconductivity of cancellous allograft bone, and further research is required to explore better materials and evaluate their potential as bone graft substitutes. Different type of stems such as polished tapered, roughened with collars, and PMMA coated stems are used for the revision of failed THRs with the impaction allografting technique (Karrholm et al., 1999; Leopold et al., 2000). The effects of different stem designs are not well understood and there is no specific design with the most favourable clinical results. The original technique described by Gie et al. uses a cemented polished collarless stem to allow initial settling and prevent bonding at the stem-cement interface. Recently the developers of the impaction allografting technique suggested the use of longer stems when the host bone around the tip is compromised or when femoral fracture occurs (Halliday et al., 2003). 13 INTRODUCTION 1.3.2 Stem subsidence The clinical significance of unstable (i.e. subsiding) stems is controversial. Subsidence of the stem does not automatically lead to clinical loosening, since the stem may restabilize within the cement mantle if a polished surface and the tapered stem is used (Ling era/., 1993; Gie etal., 1993b; Nelissen et al., 1995; Elting et al., 1995). However, it is not known if the stem subsides within the cement mantle or the entire construct within the femur (Karrholm et al., 1999). Massive stem subsidence (> 10mm) has been associated with thigh pain, cement mantle fracture, and hip dislocation and is clearly an undesirable result (van Biezen etal., 2000; Leopold etal., 2000; Kligman etal., 2002). Clinical reports of subsiding stems vary substantially among different centres without knowing the causes of clinically unstable implants (van Biezen et al., 2000; Pekkarinen et al., 2000; Leopold et al., 2000; Fetzer et al., 2001; de Roeck and Drabu, 2001; Ornstein et al., 2001; Lind et al., 2002; Nelissen et al., 2002; Piccaluga et al., 2002; Kligman et al., 2002; Ornstein et al., 2002; Robinson et al., 2002; Ullmark et al., 2002b). From a mechanical perspective, stem subsidence after impaction allografting can occur through failure of any material between the stem and the cortex or any of the interfaces between materials. Various reasons for excessive stem subsidence have been reported, including insufficient graft impaction (van Biezen et al., 2000), cement mantle fracture (Masterson et al., 1997), and the incorporation and remodelling process of the graft layer (Franzen et al., 1995). In 14 INTRODUCTION the following paragraphs, the factors which may lead to subsiding and clinically unstable stems are discussed. Subsidence of the stem may result from shear failure or consolidation of the allograft layer. (Figure 1.5) (Berzins et al., 1996). Resistance against shear is a function of friction and interlocking of the particles. Interlocking within the allograft layer is a function of impaction energy, compressive load, the grading and washing of the graft particles (Brewster et al., 1999; Dunlop et al., 2003). The morsellized allograft is confined in a cortical shell and can therefore withstand compressive forces. However, if the graft is loaded in compression, the morsellized allograft consolidates due to the low confined modulus (8MPa to 38.7MPa) and may result in subsidence of the stem-cement complex (Figure 1.5) (Giesen et al., 1999; Voor et al., 2000; Speirs, 2001). Increased graft impaction may increase the confined compressive modulus since higher confined modulus was reported in experiments where the allograft was preconditioned (Giesen et al., 1999; Speirs, 2001). Considering the different shaped impactors used for allograft impaction, graft porosity and cement interdigitation are expected to vary along the femur (Figure 1.4). Although these factors may affect the mechanical properties of the graft layer, graft porosity and cement interdigitation have not been determined in a clinically relevant setup. 15 INTRODUCTION Wmmmm v Stem s Graft consolidation M Cortex Impacted allograft 13 Cement Mantle Shear failure of the graft layer Figure 1.5: Insufficient graft impaction might result in stem subsidence through graft consolidation and/or shear failure of the graft layer. Stem subsidence may also occur due to failure of the cement-allograft and the allograft-host bone interfaces (Figure 1.6). Although important for the success of any THR (Krause etal., 1982; Dohmae etal., 1988; Ling, 1992; MacDonald et al., 1993; Mann et al., 1999; Mann et al., 2001), the interface between the host cortex and the graft material has not been addressed for the impaction allografting technique. The morphology and the mechanical characteristics of the cement-allograft interface after impaction allografting is not known. 1 6 INTRODUCTION Figure 1.6: Stem subsidence through host bone and/or cement-allograft interface failure. Masterson et al. reported a high incidence of cement mantle fractures associated with extensive subsidence (Figure 1.7) (Masterson et al., 1997). An inadequate cement mantle was reported to be the cause of the fracture. Whether the mantle fractured due to the subsidence or the stems subsided due to the fractured mantle could not be determined from this study. In a radiostereometric analysis (RSA) of 18 patients at a follow up of two years, the subsiding polished double tapered stems (average 7.5mm subsidence) showed more cement mantle defects compared with the stable stems (average 1.2mm subsidence) (Nelissen et al., 2002). The cement mantle/composite thicknesses up to 14mm were measured from anteroposterior radiographs and cement mantles less than 2mm were considered defective. No correlation between cement mantle defects and 17 INTRODUCTION stem subsidence was found in another RSA study where 24 patients were revised with roughened stems and small collars (Karrholm et al., 1999). However, interpretation of radiographs after impaction allografting is difficult and the same criteria as in primary THR may not be applicable (Linder, 2000). No detailed histomorphometric analysis of the cement layer after impaction allografting is available. Figure 1.7: Stem subsidence within the cement mantle due to cement mantle fracture. The potential for acrylic bone cement to creep has been studied in a finite element analysis for a primary THR and was associated with stem subsidence (Gie et al., 1993b; Verdonschot and Huiskes, 1997). However, the subsidence of the polished stem due to creep did not exceed 50u.m. Therefore, creep of the cement mantle appears not to be a major reason for stem subsidence. Clinical studies which excluded more severe bone loss have generally lower failure rates (Leopold et al., 2000). A radiostereometic analysis showed 1 8 INTRODUCTION that stem subsidence was significant higher in patients with larger defects (Nelissen etal., 2002). The effect of postoperative weight-bearing (WB) on stem subsidence has been investigated with RSA (Nelissen et al., 2002; Ornstein ef al., 2003). No significant differences in subsidence of the polished tapered stems up to two years postoperatively between unrestricted (50% WB for 6 weeks, after that unrestricted) and restricted (15% WB for 6 weeks, 50% WB until 3 months) WB were found. In structural tests with simulated severe bone deficiency in cadaveric femora, the impaction allografting constructs subsided more than primary hip replacements. However, the mean subsidence of stems did not exceed 2mm (Berzins ef al., 1996; Malkani et al., 1996; Hostner ef al., 2001; Kligman et al., 2003). These results suggest that if the impaction allografting procedure is done properly, the integrity of the proximal femur can be restored without significant subsidence of the stem or the stem-graft-cement construct immediately postoperatively. Biological changes of the allograft layer, the cement-allograft and the allograft-host bone interfaces during incorporation and remodelling may also lead to stem subsidence (Franzen ef al., 1995). Allograft incorporation and remodelling is discussed in more detail in the next section. 19 INTRODUCTION 1.3.3 Incorporation and remodelling of the impacted cancellous allograft Biologically, impacted allograft provides a scaffold for bone ingrowth and remodelling and facilitates cortical healing. The allograft should be completely incorporated and remodelled with host bone to facilitate re-revision and satisfy the theoretical advantages of impaction allografting. In addition, the allograft should maintain sufficient strength during healing to provide clinically stable implants. Histological observations from six human autopsies (from the proximal part or around the tip of the stem) and eight biopsies taken up to 8 years postoperatively showed that the impacted allograft does not remodel completely (Linder, 2000). The allograft appears to be first invaded by fibrovascular tissue presumably from the endosteal surface. The vascular front does not always reach the cement surface, since a layer of avascular allograft bone was observed to be up to 8mm thick. It seems that a bone formation front follows the vascular invasion with intensive bone absorption and formation. Bone remodelling ceases with time leaving the allograft layer only partially incorporated and remodelled. Starting from the endosteal surface, four different zones consisting of different tissue types were identified (Ling etal., 1993; Nelissen etal., 1995; Mikhail etal., 1999; Linder, 2000): i) viable trabecular bone, ii) composite of allograft particles and fibrous tissue iii) dead allograft bone with no vascular invasion and iv) dense fibrous tissue membrane (Figure 1.8). Similar histological observations were made in animal models including goat, sheep and rats simulating impacted graft 20 INTRODUCTION incorporation (Schreurs era/., 1994; Lamerigts et al., 2000; Tagil, 2000; van der DonkS. etal., 2001). • Cortex ^ Viable trabecular bone Composite of allograft particles and fibrous tissue [~3 Dead allograft bone with no vascular invasion ffffl Fibrous tissue membrane I I Cement Mantle Figure 1.8: Schematic representation of the four zones observed in autopsies and biopsies. Zone thickness and distribution around the stem vary substantially (Linder, 2000). Based on basic science and clinical studies, it is not clear the degree to which the allograft layer should be remodelled. Since the histological findings are compatible with good clinical results, complete remodelling may not be necessary (Linder, 2000). In addition, complete remodelling may result in a fibrous tissue membrane formation between the cement-graft interface which caused aseptic loosening of the acetabular cup in a goat model (Schreurs et al., 1994; Schimmel et al., 1998; Aspenberg, 2001). However, an incomplete restoration of the host bone stock may not facilitate re-revision, and therefore limits the indications for the use of impaction allografting. In a rat experiment, the allograft fibrous tissue composite enhanced the unconfined compressive strength compared with the allograft alone (Tagil and 21 INTRODUCTION Aspenberg, 2001). However, the long-term behaviour of this fibrous tissue composite is not known. In the case of a re-revision, the tissue composite and unvascularized bone will be removed, leaving only a few millimetres of viable allograft bone. Growth factors have been successfully used to enhance allograft incorporation and remodelling in a rat model but have been associated with early failures in humans (Wang and Aspenberg, 1994; Tagil et al., 2000; Hostner era/., 2002). The effect of graft impaction on allograft incorporation has been investigated in a rat bone chamber model (Tagil and Aspenberg, 1998). Impaction of morsellized allograft delayed incorporation at 6 weeks, but "catches up" with the unimpacted controls at 12 weeks. Different constant impaction pressures (25MPa or 2500MPa applied for 2 min) used to "impact" the morsellized allograft did not affect graft incorporation. No verification was made whether these pressure values represent a clinically significant range (Tagil and Aspenberg, 1998; Tagil, 2000). Franzen et al. suggested that changes of the allograft layer properties due to revascularisation and remodelling may cause early subsidence of the femoral stem (Franzen et al., 1995). Enneking et al. showed that the strength of structural cortical grafts decreases during revascularisation and remodelling (Enneking et al., 1975). Although these findings are not directly applicable to impacted morsellized grafts, it shows that biological processes have significant effects on graft properties in vivo (Goldberg and Stevenson, 1987). Additional evidence for 22 INTRODUCTION biologically initiated stem subsidence has been reported in a distal femoral defect model in the goat (Lamerigts et al., 2000). Allograft bone impacted into the defect was loaded during incorporation and remodelling. The loading piston subsided 2mm into the impacted allograft in two out of four cases observed at 12 weeks. The exact time of subsidence was not determined. In contrast, no subsidence of the piston occurred even after almost two million cycles in the ex vivo experiment. The effect of load on allograft incorporation is not conclusive (Lamerigts et al., 2000; Wang et al., 2000; Tagil, 2000; van der Donk S. et al., 2002). In the rabbit, significantly more impacted allograft bone was remodelled in loaded proximal tibial joint replacement compared with the unloaded (Wang ef al., 2000). In contrast, in a manually loaded femoral condyle model, no significant differences in remodelled allograft bone were found between the loaded and unloaded group. Similarly no stimulatory effect of load was found in a manually loaded rat bone chamber (Tagil, 2000). The absence of differences in the goat and rat model is most likely due to an insufficient loading stimulus. The loads were applied manually for only short periods whereas in the rabbit model, the implant was continuously loaded even if the animal didn't move. 23 INTRODUCTION 1.4 SUMMARY The expected increasing number of patients requiring revision or re-revision of failed THRs and the associated loss of host bone stock highlights the importance of this research direction. There are treatment options available with excellent clinical results for patients with minimal bone loss and in patients with intact diaphyses which ensure rigid distal fixation. However, in younger active patients with Grade 3 and 4 defects where a re-revision is likely, impaction allografting appears to be a good surgical option. This technique can be customized to the individual patient and the lost bone can be restored which facilitates re-revision. Although the short and midterm clinical results are very promising, significant complications are associated with the impaction allografting technique. Complications like dislocation, thigh pain and clinical loosening may be related to subsidence and clinically unstable implants. In addition, the current technique is difficult and time intensive and the host bone stock does not appear to be sufficiently restored. The clinical performance of the impaction allografting technique is likely to be improved by ensuring clinically stable implants during and after restoration of the host bone stock. Extensive research has focused on characterising and improving the impacted allograft layer and the remodelling process as a function of allograft impaction, allograft preparation and graft types. However, the causes of clinically unstable implants are not well understood. 24 INTRODUCTION Basic data such as the morphology of the cement-allograft-bone construct postoperatively has not been determined. The specific morphology and mechanics of the host bone interface, both immediate postoperatively and after healing, have not been addressed for the impaction allografting technique. These topics are likely important to the issue of stem subsidence and the success of the impaction allografting technique. 25 INTRODUCTION 1.5 OBJECTIVES The focus of this thesis was to determine the morphology and the mechanical characteristics of the proximal femur postoperatively and during the process of healing after impaction allografting. The objectives of this thesis were to: • Determine the morphology of the stem-cement-allograft-bone construct after impaction allografting (CHAPTER II). • Identify surgical parameters that determine the morphology intraoperatively. (CHAPTER III). • Characterize the host bone interface mechanically as a function of morphology (CHAPTER IV). • Develop an animal model to investigate local changes of the host bone and cement-allograft interface as a function of time (CHAPTER V). 1.6 SCOPE The morphology and the host bone interface characteristics after the impaction allografting procedure were investigated in human cadaveric femurs. The surgical parameters that determine the morphology intraoperatively were studied using a finite element model based on a cadaveric femur. A rat tibia bone chamber model was used to investigate local changes of the host bone interface up to 6 weeks postoperatively. 26 INTRODUCTION 1.7 REFERENCES Agency for Healthcare Cost and Quality (2003). HCUPnet, Healthcare Cost and Utilization Project. Agency for Healthcare Cost and Quality, Rockville, MD. http://www.agrq.gov/data/hcup/hcupnet.htm Aspenberg.P. (2001) Impaction grafting. Acta Orthop.Scand. 72, 198-199. Aspenberg.P. and Van der Vis.H.M. (1998) Fluid pressure may cause peri prosthetic osteolysis. Particles are not the only thing. Acta Orthop.Scand. 69, 1 -4. Barrack,R.L, Sawhney.J., Hsu.J., and Cofield.R.H. (1999) Cost analysis of revision total hip arthroplasty. A 5-year followup study. Clin.Orthop. 369 175-178. Bergmann.G., Graichen.F., and Rohlmann.A. (1993) Hip joint loading during walking and running, measured in two patients. J.Biomech. 26, 969-990. Berzins.A., Sumner.D.R., Wasielewski.R.C, and Galante.J.O. (1996) Impacted particulate allograft for femoral revision total hip arthroplasty. In vitro mechanical stability and effects of cement pressurization. J.Arthroplasty 11, 500-506. Blackley.H.R., Davis.A.M., Hutchison,C.R., and Gross.A.E. (2001) Proximal femoral allografts for reconstruction of bone stock in revision arthroplasty of the hip. A nine to fifteen-year follow-up. J.Bone Joint Surg.Am. 83-A, 346-354. Bohm,P. and Bischel,O. (2001) Femoral revision with the Wagner SL revision stem : evaluation of one hundred and twenty-nine revisions followed for a mean of 4.8 years. J.Bone Joint Surg.Am. 83-A, 1023-1031. Bolder.S.B., Verdonschot.N., Schreurs.B.W., and Buma.P. (2002) Acetabular defect reconstruction with impacted morsellized bone grafts or TCP/HA particles. A study on the mechanical stability of cemented cups in.an artificial acetabulum model. Biomaterials 23, 659-666. Boldt.J.G., Dilawari.P., Agarwal,S., and Drabu.K.J. (2001) Revision total hip arthroplasty using impaction bone grafting with cemented nonpolished stems and charnley cups. J.Arthroplasty 16, 943-952. 27 INTRODUCTION Brewster,NT., Gillespie.W.J., Howie,C.R., Madabhushi.S.P., Usmani.A.S., and Fairbairn.D.R. (1999) Mechanical considerations in impaction bone grafting. J.Bone Joint Surg.Br. 81, 118-124. Canadian Institute for Health Information (2003). Canadian Joint Replacement Registry. 2001. Chiesa.R., Tanzi.M.C, Alfonsi.S., Paracchini.L, Moscatelli.M., and Cigada.A. (2000) Enhanced wear performance of highly crosslinked UHMWPE for artificial joints. J.Biomed.Mater.Res 50, 381-387. Davis.C.M., Ill, Berry.D.J., and Harmsen.W.S. (2003) Cemented revision of failed uncemented femoral components of total hip arthroplasty. J.Bone Joint Surg.Am. 85-A, 1264-1269. de Roeck.N'J. and Drabu.K.J. (2001) Impaction bone grafting using freeze-dried allograft in revision hip arthroplasty. J.Arthroplasty 16, 201-206. Dohmae.Y., Bechtold.J.E., Sherman,R.E., Puno.R.M., and Gustilo.R.B. (1988) Reduction in cement-bone interface shear strength between primary and revision arthroplasty. Clin.Orthop. 236, 214-220. Duncan,CP., Masterson.E.L, and Masri.B.A. (1998) Impaction allografting with cement for the management of femoral bone loss. Orthop.Clin.North Am. 29, 297-305. Dunlop.D.G., Brewster.N.T., Madabhushi.S.P., Usmani.A.S., Pankaj.P., and Howie.C.R. (2003) Techniques to improve the shear strength of impacted bone graft: the effect of particle size and washing of the graft. J.Bone Joint Surg.Am. 85-A, 639-646. Elting.J.J., Mikhail.W.E., Zicat.B.A., Hubbell.J.C., Lane.L.E., and House.B. (1995) Preliminary report of impaction grafting for exchange femoral arthroplasty. Clin.Orthop. 319, 159-167. Engelbrecht.E. and Heinert.K. (1987) Klassifikation und Behandlungsrichtlinien von Knochensubstanzverlusten bei Revisionsoperationen am Hueftgelenk- mittelfristige Ergebnisse. Pp. 189-201. Springer Verlag, Hamburg. Enneking,W.F., Burchardt.H., Puhl.J.J., and Piotrowski.G. (1975) Physical and biological aspects of repair in dog cortical-bone transplants. J.Bone Joint Surg.[Am.] 57, 237-252. Enneking,W.F. and Mindell.E.R. (1991) Observations on massive retrieved human allografts. J.Bone Joint Surg.[Am.] 73, 1123-1142. 28 INTRODUCTION Fetzer.G.B., Callaghan.J.J., Templeton.J.E., Goetz.D.D., Sullivan,P.M., and Johnston,R.C. (2001) Impaction allografting with cement for extensive femoral bone loss in revision hip surgery: a 4- to 8-year follow-up study. J.Arthroplasty 16, 195-202. Franzen.H., Toksvig-Larsen.S., Lidgren.L, and Onnerfalt.R. (1995) Early migration of femoral components revised with impacted cancellous allografts and cement. A preliminary report of five patients. J.Bone Joint Surg.Br. 77, 862-864. Gie.G.A., Linder.L, Ling,R.S., Simon,J.P., Slooff.T.J., and Timperley.A.J. (1993a) Contained morselized allograft in revision total hip arthroplasty. Surgical technique. Orthop.Clin.North Am. 24, 717-725. Gie.G.A., Linder.L., Ling.R.S., Simon.J.P., Slooff.T.J., and Timperley.A.J. (1993b) Impacted cancellous allografts and cement for revision total hip arthroplasty. J.Bone Joint Surg.Br. 75, 14-21. Giesen,E.B., Lamerigts,N.M., Verdonschot.N., Buma.P., Schreurs.B.W., and Huiskes.R. (1999) Mechanical characteristics of impacted morsellised bone grafts used in revision of total hip arthroplasty. J.Bone Joint Surg.Br. 81, 1052-1057. Goldberg.V.M. and Stevenson,S. (1987) Natural history of autografts and allografts. Clin.Orthop. 225, 7-16. Griffon,D.J., Dunlop.D.G., Howie.C.R., Gilchrist.T., Salter.D.M., and Healy.D.M. (2001a) Early dissolution of a morsellised impacted silicate-free bioactive glass in metaphyseal defects. J.Biomed.Mater.Res. 58, 638-644. Griffon,D.J., Dunlop.D.G., Howie.C.R., Pratt.J.N., GilchristT.J., and Smith.N. (2001b) An ovine model to evaluate the biologic properties of impacted morselized bone graft substitutes. J.Biomed.Mater.Res. 56, 444-451. Grunig.R., Morscher.E., and Ochsner.P.E. (1997) Three-to 7-year results with the uncemented SL femoral revision prosthesis. Arch.Orthop.Trauma Surg. 116, 187-197. Halliday.B.R., English,H.W., Timperley.A.J., Gie.G.A., and Ling.R.S. (2003) Femoral impaction grafting with cement in revision total hip replacement. Evolution of the technique and results. J.Bone Joint Surg.Br. 85, 809-817. 29 INTRODUCTION Hejna.M.J. (1998) Allograft-Prosthetic Composite Reconstruction. In The Hip (Edited by Sledge.C.B.) Pp. 343-353. Lippincott-Raven, New York. Hostner.J., Hultmark.P., Karrholm.J., Malchau.H., and Tveit.M. (2001) Impaction technique and graft treatment in revisions of the femoral component: laboratory studies and clinical validation. J.Arthroplasty 16, 76-82. Hostner.J., Karrholm.J., and Hultmark.P. (2002) Early failures after femoral revisions using milled allograft bone mixed with O P - 1 , 56th meeting of the Swedish Orthopaedic Association. Hultmark.P., Karrholm.J., Stromberg.C, Herberts.P., Mose.C.H., and Malchau.H. (2000) Cemented first-time revisions of the femoral component: prospective 7 to 13 years' follow-up using second-generation and third-generation technique. J.Arthroplasty 15, 551-561. j Karrholm.J., Hultmark.P., Carlsson.L, and Malchau.H. (1999) Subsidence of a non-polished stem in revisions of the hip using impaction allograft. Evaluation with radiostereometry and dual-energy X-ray absorptiometry. J.Bone Joint Surg.Br. 81, 135-142. Katz.R.P., Callaghan.J.J., Sullivan,P.M., and Johnston,R.C. (1997) Long-term results of revision total hip arthroplasty with improved cementing technique. J.Bone Joint Surg.Br. 79, 322-326. Kligman,M., Con.V., and Roffman.M. (2002) Cortical and cancellous morselized allograft in revision total hip replacement. Clin.Orthop. 401 139-148. Kligman,M., Rotem.A., and Roffman.M. (2003) Cancellous and cortical morselized allograft in revision total hip replacement: A biomechanical study of implant stability. J.Biomech. 36, 797-802. Krause.W.R., Krug.W., and Miller, J. (1982) Strength of the cement-bone interface. Clin.Orthop. 163, 290-299. Lamerigts.N.M., Buma.P., Huiskes.R., Schreurs.W., Gardeniers.J., and Slooff.T.J. (2000) Incorporation of morsellized bone graft under controlled loading conditions. A new animal model in the goat. Biomaterials 21, 741-747. 30 INTRODUCTION Leopold,S.S., Jacobs,J.J., and Rosenberg,A.G. (2000) Cancellous allograft in revision total hip arthroplasty. A clinical review. Clin.Orthop. 371 86-97. Leopold,S.S. and Rosen berg, A.G. (2000) Current status of impaction allografting for revision of a femoral component. Instr.Course.Lect 49, 111-118. Lind.M., Krarup.N., Mikkelsen.S., and Horlyck.E. (2002) Exchange impaction allografting for femoral revision hip arthroplasty: results in 87 cases after 3.6 years' follow-up. J.Arthroplasty 17, 158-164. Linder.L. (2000) Cancellous impaction grafting in the human femur: histological and radiographic observations in 6 autopsy femurs and 8 biopsies. Acta Orthop.Scand 71, 543-552. Ling.R.S. (1992) The use of a collar and precoating on cemented femoral stems is unnecessary and detrimental. Clin.Orthop. 285, 73-83. Ling.R.S. (1997) Femoral component revision using impacted morsellised cancellous graft. J.Bone Joint Surg.Br. 79, 874-875. Ling.R.S., Timperley.A.J., and Linder.L. (1993) Histology of cancellous impaction grafting in the femur. A case report. J.Bone Joint Surg.Br. 75, 693-696. MacDonald,W., Swarts.E., and Beaver.R. (1993) Penetration and shear strength of cement-bone interfaces in vivo. Clin.Orthop. 286, 283-288. Malchau.H., Herberts,P., Sonderman.P., and Oden.A. (2002) Prognosis of total hip replacement. Swedisch National Hip Arthroplasty Registry. Malkani.A.L., Voor.M.J., Fee.K.A., and Bates,CS. (1996) Femoral component revision using impacted morsellised cancellous graft. A biomechanical study of implant stability. J.Bone Joint Surg. Br. 78, 973-978. Manley.M.T., DAntonio.J.A., Capello.W.N., and Edidin.A.A. (2002) Osteolysis: a disease of access to fixation interfaces. Clin.Orthop. 405 129-137. Mann.K.A., Mocarski.R., Damron.L.A., Allen,M.J., and Ayers.D.C. (2001) Mixed-mode failure response of the cement-bone interface. J.Orthop.Res. 19, 1153-1161. 31 INTRODUCTION Mann.KA, Werner,F.W., and Ayers.D.C. (1999) Mechanical strength of the cement-bone interface is greater in shear than in tension. J.Biomech. 32, 1251-1254. Masterson,E.L., Masri,B.A., and Duncan,CP. (1997) The cement mantle in the Exeter impaction allografting technique. A cause for concern. J.Arthroplasty 12, 759-764. Mikhail.W.E., Weidenhielm.L.R., Wretenberg.P., Mikhail.N., and Bauer.T.W. (1999) Femoral bone regeneration subsequent to impaction grafting during hip revision: histologic analysis of a human biopsy specimen. J.Arthroplasty 14, 849-853. Mulroy.W.F. and Harris,W.H. (1996) Revision total hip arthroplasty with use of so-called second-generation cementing techniques for aseptic loosening of the femoral component. A fifteen-year-average follow-up study. J.Bone Joint Surg.Am. 78, 325-330. Nelissen.R.G., Bauer.T.W., Weidenhielm.L.R., LeGolvan.D.P., and Mikhail.W.E. (1995) Revision hip arthroplasty with the use of cement and impaction grafting. Histological analysis of four cases. J.Bone Joint Surg.Am. 77, 412-422. Nelissen.R.G., Valstar.E.R., Poll.R.G., Garling.E.H., and Brand,R. (2002) Factors associated with excessive migration in bone impaction hip revision surgery: a radiostereometric analysis study. J.Arthroplasty 17, 826-833. Oonishi,H., Iwaki.Y., Kin.N., Kushitani.S., Murata.N., Wakitani.S., and Imoto.K. (1997) Hydroxyapatite in revision of total hip replacements with massive acetabular defects: 4-to 10-year clinical results. J.Bone Joint Surg.Br. 79, 87-92. Oonishi,H., Kadoya.Y., Iwaki.H., and Kin.N. (2000) Hydroxyapatite granules interposed at bone-cement interface in total hip replacements: histological study of retrieved specimens. J.Biomed.Mater.Res. 53, 174-180. Oonishi,H., Kadoya.Y., Iwaki.H., and Kin.N. (2001) Total hip arthroplasty with a modified cementing technique using hydroxyapatite granules. J.Arthroplasty 16, 784-789. Ornstein,E., Atroshi.l., Franzen.H., Johnsson.R., Sandquist.P., and Sundberg.M. (2001) Results of hip revision using the Exeter stem, impacted allograft bone, and cement. Clin.Orthop. 389 126-133. 32 INTRODUCTION Ornstein,E., Atroshi,!., Franzen,H., Johnsson.R., Sandquist.P., and Sundberg.M. (2002) Early complications after one hundred and forty-four consecutive hip revisions with impacted morselized allograft bone and cement. J.Bone Joint Surg.Am. 84-A, 1323-1328. Ornstein,E., Franzen,H., Johnsson.R., Stefansdottir.A., Sundberg.M., and Tagil,M. (2003) Hip revision with impacted morselized allografts: unrestricted weight-bearing and restricted weight-bearing have similar effect on migration. A radiostereometry analysis. Arch.Orthop.Trauma Surg. 123, 261-267. Pekkarinen.J., Alho.A., Lepisto.J., Ylikoski.M., Ylinen.P., and Paavilainen.T. (2000) Impaction bone grafting in revision hip surgery. A high incidence of complications. J.Bone Joint Surg. Br. 82, 103-107. Piccaluga.F., Gonzalez,D., V, Encinas Fernandez,J.C., and Pusso,R. (2002) Revision of the femoral prosthesis with impaction allografting and a Charnley stem. A 2- to 12-year follow-up. J.Bone Joint Surg.Br. 84, 544-549. Pratt.J.N., Griffon.D.J., Dunlop.D.G, Smith,N., and Howie.C.R. (2002) Impaction grafting with morsellised allograft and tricalcium phosphate-hydroxyapatite: incorporation within ovine metaphyseal bone defects. Biomaterials 23, 3309-3317. Rahbek.O., Overgaard.S., Jensen.T.B., Bendix.K., and Soballe.K. (2000) Sealing effect of hydroxyapatite coating: a 12-month study in canines. Acta Orthop.Scand. 71, 563-573. Robinson,D.E., Lee.M.B., Smith,E.J., and Learmonth.l.D. (2002) Femoral impaction grafting in revision hip arthroplasty with irradiated bone. J.Arthroplasty 17, 834-840. Schimmel.J.W., Buma.P., Versleyen.D., Huiskes.R., and Slooff.T.J. (1998) Acetabular reconstruction with impacted morselized cancellous allografts in cemented hip arthroplasty: a histological and biomechanical study on the goat. J.Arthroplasty 13, 438-448. Schmidt.J., Porsch.M., Sulk.C, Hillekamp.J., and Schneider.T. (2002) Removal of well-fixed or porous-coated cementless stems in total hip revision arthroplasty. Arch.Orthop.Trauma Surg. 122, 48-50. 33 INTRODUCTION Schreurs.B.W., Buma.P., Huiskes.R., Slagter.J.L, and Slooff.T.J. (1994) Morsellized allografts for fixation of the hip prosthesis femoral component. A mechanical and histological study in the goat. Acta Orthop.Scand. 65, 267-275. Shanbhag.A.S., Hasselman.C.T., and Rubash.H.E. (1997) The John Charnley Award. Inhibition of wear debris mediated osteolysis in a canine total hip arthroplasty model. Clin.Orthop. 344, 33-43. Slooff.T.J., Huiskes.R., van Horn.J., and Lemmens.A.J. (1984) Bone grafting in total hip replacement for acetabular protrusion. Acta Orthop.Scand. 55, 593-596. Speirs (2001) Calcium Phosphate cements composites in revision hip replacement. Master of Applied Science Thesis, University of British Columbia. Stulberg.S.D. (2002) Impaction grafting: doing it right. J.Arthroplasty 17, 147-152. Tagil,M. (2000) The morselized and impacted bone graft. Animal experiments on proteins, impaction and load. Acta Orthop.Scand.Suppl. 290:1-40. Tagil,M. and Aspenberg.P. (1998) Impaction of cancellous bone grafts impairs osteoconduction in titanium chambers. Clin.Orthop. 352, 231-238. Tagil,M. and Aspenberg.P. (2001) Fibrous tissue armoring increases the mechanical strength of an impacted bone graft. Acta Orthop.Scand. 72, 78-82. Tagil,M., Jeppsson.C, and Aspenberg.P. (2000) Bone graft incorporation. Effects of osteogenic protein-1 and impaction. Clin.Orthop. 371, 240-245. Ullmark.G., Hallin.G., and Nilsson.O. (2002a) Impacted corticocancellous allografts and cement for femoral revision of total hip arthroplasty using Lubinus and Charnley prostheses. J.Arthroplasty 17, 325-334. Ullmark.G., Hallin.G., and Nilsson.O. (2002b) Impacted corticocancellous allografts and cement for revision of the femur component in total hip arthroplasty. J.Arthroplasty 17, 140-149. van Biezen,F.C, ten Have.B.L, and Verhaar.J.A. (2000) Impaction bone-grafting of severely defective femora in revision total hip surgery: 21 hips followed for 41-85 months. Acta Orthop.Scand. 71, 135-142. 34 INTRODUCTION van der Donk S., Buma.P., Aspenberg,P., and Schreurs.B.W. (2001) Similarity of bone ingrowth in rats and goats: a bone chamber study. Comp Med. 51, 336-340. van der Donk S., Buma.P., Verdonschot.N., and Schreurs.B.W. (2002) Effect of load on the early incorporation of impacted morsellized allografts. Biomaterials 23, 297-303. Van der Vis H.M., Aspenberg,P., Marti.R.K., Tigchelaar.W., and Van Noorden.C.J. (1998) Fluid pressure causes bone resorption in a rabbit model of prosthetic loosening. Clin.Orthop. 350 201-208. Verdonschot.N. and Huiskes.R. (1997) Acrylic cement creeps but does not allow much subsidence of femoral stems. J.Bone Joint Surg.Br. 79, 665-669. Verdonschot.N., van Hal.C.T., Schreurs.B.W., Buma.P., Huiskes.R., and Slooff.T.J. (2001) Time-dependent mechanical properties of HA/TCP particles in relation to morsellized bone grafts for use in impaction grafting. J.Biomed.Mater.Res. 58, 599-604. Voor.M.J., Nawab.A., Malkani.A.L, and Ullrich,C.R. (2000) Mechanical properties of compacted morselized cancellous bone graft using one-dimensional consolidation testing. J.Biomech 33, 1683-1688. Wang, J.S. and Aspenberg,P. (1994) Basic fibroblast growth factor increases allograft incorporation. Bone chamber study in rats. Acta Orthop.Scand. 65, 27-31. Wang.J.S., Tagil,M., and Aspenberg,P. (2000) Load-bearing increases new bone formation in impacted and morselized allografts. Clin.Orthop. 378, 274-281. Weeden.S.H. and Paprosky.W.G. (2002) Minimal 11-year follow-up of extensively porous-coated stems in femoral revision total hip arthroplasty. J.Arthroplasty 17, 134-137. 35 CHAPTER II ALLOGRAFT IMPACTION AND CEMENT PENETRATION AFTER THE REVISION OF FAILED THR WITH IMPACTION ALLOGRAFTING: A HISTOMORPHOMETRIC ANALYSIS IN THE CADAVERIC FEMUR Abstract: We studied the amount of graft impaction, cement penetration and impaction forces in a clinically relevant impaction allografting set-up. The cancellous bone was removed proximally and local diaphyseal lytic defects were i simulated in six human cadaveric femurs. After the impaction grafting procedure, the specimens were sectioned and prepared for histomorphometric analysis. The graft porosity was lowest in Gruen zone 4 (52%) and highest in Gruen zone 1 (76%). At the level of Gruen zones 6&2, virtually the entire cross-section was filled with bone cement. The cement sometimes reached the endosteal surface also in other Gruen zones. The mean peak impaction forces exerted with the impactors were negatively correlated with the graft porosity. The presented data will serve as a baseline for future investigations of the impaction allografting procedure and may help in the development of bone graft substitutes for revision hip replacement. 36 ALLOGRAFT IMPACTION AND CEMENT PENETRATION 2.1 INTRODUCTION With the potential to restore host bone stock, impaction allografting is an attractive technique for the treatment of failed total hip replacements (THR). However, a high prevalence of stem subsidence (>10mm) and intra- and postoperative fracture have been associated with the procedure (Elting et al., 1995; Leopold and Rosenberg, 2000). Other clinical studies reported more promising results (Franzen et al., 1995; Boldt et al., 2001; de Roeck and Drabu, 2001; Fetzeret al., 2001; Knight and Helming, 2000; Ullmark et al., 2002). The mechanical and biological environments to which the allograft bone is subjected in a revision hip arthroplasty scenario are not well described. For example, most research has focused on morsellized allograft bone alone, omitting the infiltration of the cement into the graft (Brewster et al., 1999; Brodt et al., 1998; Lamerigts et al., 2000; Tagil, 2000; Ullmark and Nilsson, 1999). The structural properties and thickness of this graft-cement composite layer may be an important factor that determines the strength of the cement-graft-cortical bone complex. The graft porosity after impaction, which has been shown to influence both the shear strength of the allograft layer and the incorporation of the graft, has not been determined in a clinically relevant set-up (Brodt et al., 1998; Tagil and Aspenberg, 1998). Analysis of retrieved specimens has focused mainly on a qualitative description of allograft bone incorporation and did not describe cement penetration or graft porosity (Linder, 2000; Ling et al., 1993; Nelissen et al., 1995; Ullmark and Obrant, 2002). In addition, the impaction 37 ALLOGRAFT IMPACTION AND CEMENT PENETRATION forces that are required to achieve firm packing of the bone chips remain unknown, hence the variability seen in clinical practice and the variable, but often high, prevalence of fractures. An incomplete cement mantle, which may lead to cement mantle fracture and stem subsidence, has been described earlier but no detailed histomorphometric analysis is available (Masterson et al., 1997). Therefore, the goal of this study was to comprehensively describe the morphology of the graft-cement-cortical bone complex as it appears after an impaction allografting procedure to provide a basis for future studies. The specific objectives in a clinically relevant cadaveric model were to determine the: i Impaction forces during graft impaction, ii Graft porosity as a function of location within the femur, iii Cement mantle thickness, iv Cement penetration into the allograft bone, and v Residual cement-free graft layer thickness. 38 ALLOGRAFT IMPACTION AND CEMENT PENETRATION 2.2 METHODS The femoral necks from six fresh frozen human cadaveric femurs were osteotomized. Using a high-speed burr the entire cancellous bone was removed from the proximal femoral metaphysis and local diaphyseal lytic defects were created to simulate cavitary bone loss at the time of revision total hip arthroplasty. Three experienced surgeons performed the impaction allografting procedure, each on two cadaveric femurs using the X-Change revision system (Howmedica Inc, Rutherford, NJ, USA). Prior to impaction, the intramedullary canal was occluded 2cm distal to the most distal lytic defect with a plastic plug and two Kirschner wires to avoid plug subsidence. Three to four fresh frozen femoral heads for each specimen were thawed in water, cut into 10x10x25mm blocks and morsellized in a Lere bone mill (DePuy, Warsaw, IN, USA). None of the articular cartilage, fat or bone marrow was removed from morsellized allograft chips. The bone graft was not defatted. Similar to Brewster et. al., the bone chip particle size distribution was determined by sieve analysis (Brewster et al., 1999). The bone chips were passed through sieves with successively smaller pores, using mesh sizes 0.6, 1, 2, 2.36, 3.35, 4 and 8mm (0.6 to 4mm Fisher Scientific, Nepean, ON, 8mm custom made). A load cell (MC3-6-1000, ATM I Inc., Watertown, MA, USA) was placed below the femur to measure the impaction forces during the procedure. The forces measured by the load cell were acquired at a frequency of 2.5kHz and stored on a personal computer. 39 ALLOGRAFT IMPACTION AND CEMENT PENETRATION In general, the procedure was conducted according to the guidelines of the X-Change revision system with one exception. The surgeon should start using the proximal impactors when the distal impaction line (i.e. depth that the tip of the stem will reach on implantation) with the distal impactors is reached (Figure 2 . 1 ) . In four out of six specimens the distal impactors were used past the distal impaction line. The appropriate neomedullary canal position was achieved by impacting the proximal impactors into the pre-impacted bone graft. In one specimen high impaction forces caused a longitudinal fracture which was fixed with cerclage wires. Gruen zone 1 Gruen zone 2 Gruen zone 3 Level 6 Gruen zone 7 Level 7 Level 8 Level 9 Level 10 Level 11 Level 12 Gruen zone 6 Gruen zone 5 Levels Gruen zone 1 Levels Gruen zone 7&1 Levels Gruen zone 6&2 Levels Gruen zone 5&3 Gruen zone 4 Level 13 Level 14 Gruen zone 4 Distal impaction line Levels Gruen zone 4 Figure 2.1: Schematic drawing of the specimens shows the fourteen levels that were matched based on anatomical landmarks. The levels were grouped according to Gruen zones (Gruen et al., 1979). 40 ALLOGRAFT IMPACTION AND CEMENT PENETRATION After impaction, two packages of low viscosity Simplex cement (Howmedica Inc, Rutherford, NJ, USA) were mixed and injected into the neo-medullary canal. After the canal had been filled in a retrograde manner, a proximal femoral seal was applied and the cement pressurised with a Prism II cement gun (DePuy, Warsaw, IN). The pressure was maintained until the viscosity of the cement was appropriate for the insertion of the double tapered polished Exeter stem (Howmedica Inc, Rutherford, NJ, USA). After the cement had polymerized completely, anterioposterior and mediolateral radiographs were obtained to verify the stem position before it was removed. Coloured bone cement prepared by mixing Simplex cement and food colouring (Scott-Bathgate LTD, Winnipeg, MB, Canada) was used to fill the cavity produced by the stem to distinguish the bone cement from the stem in the histological sections. The femurs were cut in 6mm thick transverse sections with a diamond saw (Exact Technologies Inc., Oklahoma City, OK, USA). Twenty-seven to thirty-six sections were obtained from each femur. Anatomical landmarks and the tip of the stem were used to match the sections to fourteen levels (Figure 2.1). These levels were grouped into five levels of Gruen zones and processed for non-decalcified histomorphometric analysis (Figure 2.1) (Gruen et al., 1979). The sections were fixed in 10% buffered Formalin and then dehydrated in 70%, 95% and 100% ethanol. The dehydrated sections were infiltrated with a light-curing resin (Technovit 7200, Kulzer Ltd, Wehrheim, Germany) in an incubator. Subsequently, the resin was polymerized under ultraviolet light. The bone 41 ALLOGRAFT IMPACTION AND CEMENT PENETRATION cement was not dissolved during this procedure. From each 6mm section a 400um section was cut, ground to lOOum and stained with Alizarin Red S. The histological sections were photographed with a high resolution digital camera (1600 x 1200 pixels, CoolPix 950 Nikon, USA) and stored for image analysis. The impacted allograft porosity, the cement mantle thickness, the cement penetration and the remaining allograft layer thickness were measured in anterior, posterior, lateral and medial quadrants for each section using IMAGE-PRO 4.5 (Mediacybernetics, Silver Spring, MD, USA) (Figure 2.2). Posterior quadrant Gruen zone 1 Gruen zone 7&1 Stem ] Cement mantle 10mm Gruen zone 6&2 Gruen zone 5&3 Gruen zone 4 I Cortex Graft-cement composite I I Graft layer Figure 2.2: Photograph and schematic drawing of five sections in all Gruen zones from a typical specimen. The different patterns in the schematic drawing indicate the materials present in each section. Each section was divided in anterior, posterior, medial and lateral quadrants. 42 ALLOGRAFT IMPACTION AND CEMENT PENETRATION The porosity was defined by the void area (including the cement area) divided by the total area. Note that this is a two-dimensional measure of the porosity and not a volumetric porosity. The porosity of the unimpacted morsellized allograft was determined by loosely packing four moulds (23x33x15mm, Exact Technologies Inc., Oklahoma City, OK, USA) with graft chips. These specimens were processed and analysed as with the femoral sections. Allograft porosity, mean cement mantle thickness, mean cement penetration and mean allograft layer thickness were compared with a two-way repeated measures analysis of variance (ANOVA) with the level and the quadrants as factors. The mean peak impaction force was calculated from the measured force and grouped into a mean peak distal and proximal impaction force according to the type of impactor used. Student-Newman Keuls analysis was used for post hoc comparisons and the significance level was 95%. 43 ALLOGRAFT IMPACTION AND CEMENT PENETRATION 2 .3 RESULTS The particle size distribution of the bone chips used in this experiment is shown in Figure 2.3. Particles greater than 10mm were in general not used for the impaction allografting procedure. 0.1 1 H 10 Particle size (mm) Figure 2.3: Particle size distribution of the allograft chips used in this experiment in a semi-logarithmic plot. The ordinate indicates the percentage by weight of particles smaller than the size given by the abscissa. For example, as indicated by the dashed line, 50% of the particles used in this experiment are smaller than 4mm. The porosity of the impacted graft was highest proximally in Gruen zone 1 (75%) and lowest in Gruen zone 4 (52%) and this difference was statistically significant (p=0.003; APPENDIX G, Table 1) (Figure 2.4). No significant difference was observed among the four quadrants (i.e. anterior, posterior, 44 ALLOGRAFT IMPACTION AND CEMENT PENETRA TION lateral and medial). The porosity of the unimpacted graft was 81% (SD 3%). The graft porosity distally (Gruen zone 4) was negatively correlated with the mean peak impaction force exerted with the distal impactors (r=0.95) and proximally with mean peak force exerted with the proximal impactors (r=0.81) (Figure 2.5). The maximum impaction force measured was 1630N with the distal and 2150N with the proximal impactors. 100 Gruen zone 7/1 Gruen zone 6/2 Gruen zone 5/3 Gruen zone 4 Level! Level2 Level3 Level4 Level5 Level6 Level7 Level8 Level9 LevellO Level! 1 Levell2 Level!3 Level14 Figure 2.4: Graft porosity as a function of levels. The error bars indicate the standard deviation. The dashed line represents the porosity of the unimpacted graft. 45 ALLOGRAFT IMPACTION AND CEMENT PENETRATION 1600 1400 1200 | 1000 c o ro u u u Q. E ro 600 ro OJ < 400 200 30 o Proximal impactor A Distal impactor 40 50 60 Graft porosity [%] 70 80 90 Figure 2.5: Correlation of average peak impaction force with graft porosity for the proximal and distal impactors. The average cement mantle thickness in Gruen zone 1 to 5&3 was 1.7mm (SD 0.4mm). No significant differences were found between the four quadrants (p=0.18) or along the length of the stem, except level 14 had a significantly thinner cement mantle compared with Level 1 and 2 (p=0.03; APPENDIX G, Table 2) (Figure 2.6). In general, the cement mantle was not uniform in the quadrants as shown in Figure 2.2. It exceeded 2mm in some places and was absent in others. 46 ALLOGRAFT IMPACTION AND CEMENT PENETRATION Levsll Level2 Level3 Level4 Level5 Level6 Level7 Level8 Level9 LevellO Level!! Level!2 Level!3 Level!4 |—a—Cement mantle — 0 — Cement penetration line Endosteal Cortex | Figure 2.6: Mean distance of the cement mantle, cement penetration and endosteal cortex from the stem surface. The error bars indicate the standard deviation. The standard deviation for the endosteal cortex is not shown for convenience. The guide wire caused the cement mantle in Gruen zone 4 (i.e. Levels 13 and 14). Below the tip of the stem (Gruen zone 4), the mean cement penetration was significantly lower (p< 0.001) compared with the proximal part of the femur (Figure 2.6). In the posterior quadrant of Level 5 to Level 9 (Figure 2.1) the mean cement penetration was significantly higher compared with the other quadrants (p=0.002; APPENDIX G, Table 3). There was only a weak correlation (r=0.55; p<0.001) between cement penetration and graft porosity. Similar to the cement penetration, the mean distance from the stem to the cortex in the posterior quadrant of Level 5 to 8 was significantly larger than in the other quadrants (p< 0.001) 47 ALLOGRAFT IMPACTION AND CEMENT PENETRATION The averaged residual cement-free impacted graft layer at the level of Gruen zones 6&2 was significantly thinner (0.5mm; SD 0.4mm; p< 0.01; APPENDIX G, Table 4) compared with Gruen zones 1, 7&1, and 4 (Figure 2.6). There were no significant differences among the quadrants. 48 ALLOGRAFT IMPACTION AND CEMENT PENETRATION 2 . 4 DISCUSSION The current study describes comprehensively the morphology of the stem-cement-graft-complex after the impaction allografting procedure in a cadaver model. Clinical studies, retrieval analysis, and basic biomechanical properties of morsellized allograft bone are available, but the morphology of the stem-cement-graft complex has not been described until now. This baseline information can be used to change or improve the impaction allografting technique. As with any cadaveric model, this study has some limitations. The impaction allografting procedure was performed under ideal conditions. The femur was rigidly fixed, which may lead to tighter impaction of the graft chips. Since there was no intramedullary bleeding in the model, one may presume that this would influence the cement penetration.. However, experimental measurements for primary THR's have shown that cement penetration was not affected substantially by the intramedullary bleeding pressure (Breusch ef al., 2002; Majkowski etal., 1994). The particle size distribution of bone particles used in this study was similar to that reported by Brewster et al.(Brewster et al., 1999). The graft porosity was highest proximally at the level of Gruen zone 1 and only 5% lower than the unimpacted graft. In contrast, at the level of Gruen zone 4, the graft porosity was 29% lower compared with the unimpacted graft. This shows a higher effectiveness of the distal impactors since the averaged peak 49 ALLOGRAFT IMPACTION AND CEMENT PENETRATION impaction forces were generally lower at this level compared with the proximal impactors (Figure 2.4). No significant differences in graft porosity were found among the quadrants, which indicates that the different anterior-posterior and medial-lateral tap angles of the proximal impactors did not affect graft porosity. The relatively large variations of the graft porosity in Gruen zones 5&3 are most likely due to the slight variation in the impacting technique. Two of the three surgeons used the distal impactors past the recommended distal impaction line (X-Change System Operative Technique, Howmedica Inc, Rutherford, NJ, USA), which resulted in a tighter packing of the bone graft around the distal third of the stem. Higher impaction force on the proximal impactors was required to get the neomedullary canal in place. However, this tighter graft impaction around the tip may be beneficial for the stability of the construct. The correlation of the average peak impaction force with the graft porosity indicates that a higher average peak impaction force results in denser packing of the graft chips. The mean pure cement mantle thickness (excluding the cement that had penetrated the allograft layer) was somewhat lower than expected. The cement mantle exceeded the required 2mm in some areas and can be considered as inadequate or absent in others (Figure 2.2). The significance of the thin average and partially incomplete cement mantle is not fully understood for the impaction allografting construct. The extensive cement penetration which produces a substantial cement-graft composite implies that a thicker pure cement mantle may not be necessary. However, bone particles within the cement mantle may 50 ALLOGRAFT IMPACTION AND CEMENT PENETRATION give rise to cement mantle or bone-cement composite fractures which were reported earlier to be associated with stem subsidence (Masterson et al., 1997). In this study, the inadequate cement mantles were related to undersized proximal impactors, which have subsequently been redesigned. Additional study is required to further elucidate the biological and mechanical implications of these observations. The mean cement penetration was higher than expected. In Gruen zones 6&2, almost the entire cross-section was filled with bone cement. The simulated lytic defects in this zone were generally filled with the graft-cement composite. It remains a question to what extent these defects are remodelled. In other Gruen zones, the cement sometimes contacted the endosteal cortex. This could be beneficial if no cortical remodelling is required. On the other hand, Dai et al. showed that inorganic bone particles embedded in bone cement could be replaced by new bone and form a viable cement-bone interface (Dai et al., 1991). The shear strength of this viable interface was significantly higher compared with cement only. However, it is not known if these findings are applicable to the impaction allografting case. The weak correlation between mean cement penetration and graft porosity suggests that other factors such as the pressure gradient of the pressurised cement and the presence of fat and bone marrow may determine cement penetration. 51 ALLOGRAFT IMPACTION AND CEMENT PENETRATION The residual graft layer was smallest in Gruen zones 6&3 and largest proximally. The optimal graft layer thickness probably depends on the location. However, biopsies and autopsies showed that only a few millimetres of the allograft bone is replaced by viable bone. Therefore a thin graft layer with a viable bone-cement interface may be sufficient for a successful impaction allografting procedure. This investigation will serve as a baseline for future studies of the mechanical and biological processes that make the impaction allografting a successful procedure and will help to determine the morphological properties of new materials able to replace allograft bone. 52 ALLOGRAFT IMPACTION AND CEMENT PENETRATION 2 . 5 REFERENCES Boldt.J.G., Dilawari.P., Agarwal,S., and Drabu.K.J. (2001) Revision total hip arthroplasty using impaction bone grafting with cemented nonpolished stems and charnley cups. J.Arthroplasty 16, 943-952. Breusch.S., Heisel.C, Muller.J., Borchers.T., and Mau.H. (2002) Influence of cement viscosity on cement interdigitation and venous fat content under in vivo conditions: a bilateral study of 13 sheep. Acta Orthop.Scand. 73, 409-415. Brewster,NT., Gillespie.W.J., Howie.C.R., Madabhushi.S.P., Usmani.A.S., and Fairbairn.D.R. (1999) Mechanical considerations in impaction bone grafting. J.Bone Joint Surg.Br. 81, 118-124. Brodt.M.D., Swan,CC, and Brown,T.D. (1998) Mechanical behavior of human morselized cancellous bone in triaxial compression testing. J.Orthop.Res. 16, 43-49. Dai.K.R., Liu.Y.K., Park.J.B., Clark.C.R., Nishiyama.K., and Zheng.Z.K. (1991) Bone-particle-impregnated bone cement: an in vivo weight-bearing study. J.Biomed.Mater.Res. 25, 141-156. de Roeck.N.J. and Drabu.K.J. (2001) Impaction bone grafting using freeze-dried allograft in revision hip arthroplasty. J.Arthroplasty 16, 201-206. Elting.J.J., Mikhail.W.E., Zicat.B.A., Hubbell.J.C, Lane.L.E., and House.B. (1995) Preliminary report of impaction grafting for exchange femoral arthroplasty. Clin.Orthop. 319, 159-167. Fetzer.G.B., Callaghan.J.J., Templeton.J.E., Goetz.D.D., Sullivan,P.M., and Johnston,R.C. (2001) Impaction allografting with cement for extensive femoral bone loss in revision hip surgery: a 4- to 8-year follow-up study. J.Arthroplasty 16, 195-202. Franzen,H., Toksvig-Larsen.S., Lidgren.L, and Onnerfalt.R. (1995) Early migration of femoral components revised with impacted cancellous allografts and cement. A preliminary report of five patients. J.Bone Joint Surg.Br. 77, 862-864. Gruen,T.A., McNeice.G.M., and Amstutz.H.C. (1979) "Modes of failure" of cemented stem-type femoral components: a radiographic analysis of loosening. Clin.Orthop. 141, 17-27. 53 ALLOGRAFT IMPACTION AND CEMENT PENETRATION Knight,J.L. and Helming,C. (2000) Collarless polished tapered impaction grafting of the femur during revision total hip arthroplasty: pitfalls of the surgical technique and follow-up in 31 cases. J.Arthroplasty 15, 159-165. Lamerigts.N.M., Buma.P., Huiskes.R., Schreurs.W., Gardeniers.J., and Slooff.T.J. (2000) Incorporation of morsellized bone graft under controlled loading conditions. A new animal model in the goat. Biomaterials 21, 741-747. Leopold,S.S. and Rosenberg,A.G. (2000) Current status of impaction allografting for revision of a femoral component. Instr.Course.Lect 49, 111-118. Linder.L. (2000) Cancellous impaction grafting in the human femur: histological and radiographic observations in 6 autopsy femurs and 8 biopsies. Acta Orthop.Scand 71, 543-552. Ling.R.S., Timperley.A.J., and Linder.L. (1993) Histology of cancellous impaction grafting in the femur. A case report. J.Bone Joint Surg.Br. 75, 693-696. Majkowski.R.S., Bannister.G.C, and Miles.A.W. (1994) The effect of bleeding on the cement-bone interface. An experimental study. Clin.Orthop. 299, 293-297. Masterson,E.L., Masri,B.A., and Duncan,CP. (1997) The cement mantle in the Exeter impaction allografting technique. A cause for concern. J.Arthroplasty 12, 759-764. Nelissen.R.G., Bauer.T.W., Weidenhielm.L.R., LeGolvan.D.P., and Mikhail.W.E. (1995) Revision hip arthroplasty with the use of cement and impaction grafting. Histological analysis of four cases. J.Bone Joint Surg.Am. 77, 412-422. Tagil,M. (2000) The morselized and impacted bone graft. Animal experiments on proteins, impaction and load. Acta Orthop.Scand.Suppl. 290, 1-40. Tagil,M. and Aspenberg.P. (1998) Impaction of cancellous bone grafts impairs osteoconduction in titanium chambers. Clin.Orthop. 352, 231-238. Tagil,M. and Aspenberg.P. (2001) Fibrous tissue armoring increases the mechanical strength of an impacted bone graft. Acta Orthop.Scand. 72, 78-82. Ullmark.G., Hallin.G., and Nilsson.O. (2002) Impacted corticocancellous allografts and cement for femoral revision of total hip arthroplasty using Lubinus and Charnley prostheses. J.Arthroplasty 17, 325-334. 54 ALLOGRAFT IMPACTION AND CEMENT PENETRATION Ullmark.G. and Nilsson.O. (1999) Impacted corticocancellous allografts: recoil and strength. J.Arthroplasty 14, 1019-1023. Ullmark.G. and Obrant.K.J. (2002) Histology of impacted bone-graft incorporation. J.Arthroplasty 17, 150-157. 55 CHAPTER III CEMENT FLOW DURING IMPACTION ALLOGRAFTING A FINITE ELEMENT ANALYSIS Abstract: The optimum amount of cement interdigitation in the revision of failed hip replacements with impaction allografting is not known. Excessive cement intrusion may prevent the remodeling of the bone graft or the cortex if the cement reaches the endosteal surface. On the other hand, insufficient cement interdigitation may lead to stem subsidence. With a three-dimensional finite element model the effect of cement viscosity, the magnitude and duration of pressurization and the distribution of the porosity along the femur were investigated. The finite element model predicted the cement intrusion into impacted bone graft accurately when compared with observations in a cadaver. Cement viscosity, cement pressure and duration of pressurisation changed the depth of cement intrusion but maintained a similar profile. The distribution of the permeability along the femur determines the intrusion profile. Cement viscosity, the applied pressure or the duration of the pressurisation, can be adjusted to limit the cement volume injected into the medullary canal and therefore prevent the cement reaching the endosteal surface. 56 CEMENT FLOW DURING IMPACTION ALLOGRAFTING 3.1 INTRODUCTION The revision of failed total hip replacement (THR) associated with extensive osteolytic bone loss is a challenge for the orthopaedic surgeon. With the potential to restore the host bone stock, impaction allografting is an attractive revision procedure which has been increasingly used since it was introduced (Slooff et al., 1984; Gie etal., 1993). In the impaction allografting procedure, the intramedullary canal of the femur is filled with impacted morsellized allograft bone after removal of the failed femoral component and the cement. With successive impaction of the bone graft into the intramedullary canal, the osteolytic defects are filled and a neo-medullary canal is formed which can accommodate a normal sized polished stem. Similar to a primary THR, the neo-medullary canal is filled in a retrograde manner with low viscosity acrylic bone cement. A proximal femoral seal is applied and the cement is pressurized with a cement gun to achieve cement interdigitation with the impacted graft. The pressure is maintained until the viscosity of the cement is appropriate for the insertion of a double tapered polished stem. The optimal amount of cement interdigitation in the revision of failed hip replacements with impaction allografting is not known. Excessive cement intrusion may prevent the remodeling of the bone graft or the cortex if the cement reaches the endosteal surface (CHAPTER II). On the other hand, insufficient cement interdigitation may lead to stem subsidence which has been associated with the impaction allografting technique (Elting et al., 1995; Leopold and 57 CEMENT FLOW DURING IMPACTION ALLOGRAFTING Rosenberg, 2000). For primary total hip replacements, adequate bone cement intrusion into the cancellous bone bed has been shown to be important for success (Rey, Jr. etal., 1987; Mann etal., 1997; Churchill etal., 2001). The parameters that determine the cement intrusion into the cancellous bone bed are not well understood. The magnitude of the applied pressure during pressurisation, the quality of proximal seal, the viscosity of the cement, intramedullary bleeding, the permeability of the cancellous bone bed and the pressure generated during stem insertion have been reported to affect the amount of cement intrusion (Markolf and Amstutz, 1976; Song et al., 1994; Majkowski etal., 1994; McCaskie etal., 1997; Churchill etal., 2001). An analysis of these parameters, particularly for impaction allografting, has not been conducted and should be useful in optimizing the technique. ( Therefore the purpose of this study was to investigate how cement viscosity, cement pressure magnitude and duration of the pressurisation and impacted allograft porosity affect the cement flow into the allograft bone. A three-dimensional finite element model of the surgical technique was validated and used to assess these parameters. 58 CEMENT FLOW DURING IMPACTION ALLOGRAFTING 3.2 METHODS The geometry for the three dimensional finite element model was obtained from one cadeveric femur on which the impaction allografting procedure was performed (CHAPTER II). The endosteal surface defined the boundary of the model (Figure 3.1). The cement volumes in the mantle and the impacted graft as well as the porosities at different sections measured for this femur are summarized in Table 3.1 and Figure 3.2 respectively. Medial-lateral view Sagittal cross-section Figure 3.1: Medial-lateral view and sagittal cross-section of the finite element model indicating the site of cement injection and locations where cement leakage was simulated around the seal. 59 CEMENT FLOW DURING IMPACTION ALLOGRAFTING Table 3.1: Cement volumes obtained from the cadaveric femur (CHAPTER II). Volume [cm3] A: Total cement used (two packages) Measured 99.7 B: Stem volume Measured 11.3 C: Cement mantle volume Measured 10.4 D: Cement injected during pressurization (A-(B+C)) 78.0 E: Cement volume within impacted bone Measured 23.5 F: Cement volume lost through seal (D-E) 54.5 G: Cement volume in femur after stem insertion (E+C) 33.9 The cement flow through the impacted bone graft and the cement flow in the neo-medullary canal were modeled with the partial saturated flow capabilities from ABAQUS (Hibbitt, Karlsson & Sorensen Inc., Pawtucket, R.I.). In ABAQUS, a porous medium is modeled by attaching finite element mesh to the solid phase through which a fluid can flow. When the medium is partially saturated, two fluids, a wetting and a non-wetting, exist at the same point. It is assumed that the non-wetting fluid can diffuse through the medium and across the boundary so that its pressure is small enough that it can be neglected (i.e. air flow in a porous medium). This implies that the medium is fully saturated at a wetting fluid pressure greater than zero. The negative fluid pressure in a partially saturated medium like soils represents capillary effects. For the purpose of this simulation the atmospheric pressure was assumed to be of a negative magnitude depending on the pressure applied during pressurisation. The flow of the wetting fluid is described by Darcy's law: -k dd vw= ~£ (1) snp dX 60 CEMENT FLOW DURING IMPACTION ALLOGRAFTING where s is the saturation [dimensionless; 0.05-1], np is the porosity of the impacted bone [%], vw is the volumetric flow rate per unit area [ms"1], k is the hydraulic conductivity [ms"1] and .M. is the gradient of the hydrostatic head 5X [dimensionless]. The hydraulic conductivity k is a function of bone porosity and cement viscosity and therefore a function of time. The hydraulic conductivity is related to the fluid independent intrinsic permeability K p [m2] by: where yw is the specific weight of the fluid [Nm"3 ] and r| the dynamic viscosity [Pas] of the fluid. The intrinsic permeability has been determined experimentally for different bone porosities (Beaudoin et al., 1991; Grimm and Williams, 1997; Nauman etal., 1999; Baroud etal., 2003). Experimental investigations of bone cement have characterised the bone cement as a pseudoplastic fluid with a dynamic viscosity r| given by (Krause et al., 1982). 77 = Kf-' (2) where K is the consistency index, n the power index describing the flow behaviour and y[sA] is the shear rate. Both K and n are functions of the elapsed time after mixing. For Simplex P cement (Howmedica) K= 2 6 9 e 0 0 0 6 5 t and n=-61 CEMENT FLOW DURING IMPACTION ALLOGRAFTING 0.0003t+0.67 where t is the time after mixing in minutes (Krause et al., 1982). The shear rate yean be calculated by (Krause etal., 1982): '3n + lY AQ^ r An ym-3 j (3) where 3 n + 1 is a correction factor for non-Newtonian fluids, Q [ m V 1 ] is An the volumetric flow rate and r [m] is the radius of the outlet. The volumetric flow rate was estimated based on the cement volume injected during cement pressurisation divided by the time (Q=1.42cm3s"1) (Tables 3.1 and 3.2). The radius of the nozzle was 5.5mm. The shear rate was assumed to be constant in the domain of the model. The calculated hydraulic conductivities as a function of level and time are summarized in Figure 3.2. The saturation, s, in Equation 1 is a function of the pore pressure. This relationship changes with different media/ fluid combinations and with fluid absorption and exsorption (Nguyen and Durso, 1983). This relationship has to be determined experimentally and is not available for trabecular bone and cement. The gradient of the pressure-saturation relationship determines the smallest time increment for a given element size (ABAQUS/Standard User manual, Hibbitt, Karlsson & Sorensen Inc., Pawtucket, R.I.). An average element length of 2mm and a saturation increase to the third power with the pressure allowed time increments of 0.25 seconds. 62 CEMENT FLOW DURING IMPACTION ALLOGRAFTING Table 3.2: Summary of the parameters for the performed simulations. Simulation Starting times after mixing [s] Duration of Pressure Pressurization Stem insertion Pressurization [s] [kPa] Cadaveric femur (Frei et. al. 2003) 125 195 55 88 Increased viscosity (late stem insertion) 287 357 55 88 Increased pressure (25%) 125 195 55 110 Increased pressure (50%) 125 195 55 132 Increased duration of pressurization 125 210 70 88 Uniform averaged graft porosity ( 72%) 125 195 55 88 Uniform averaged graft porosity tight seal 125 33.5 18.5 88 Figure 3.2: Porosities and hydraulic conductivities as a function of different levels along the simulated cadaveric femur. The hydraulic conductivities were calculated based on intrinsic permeability data from Beaudoin et. al. 1991. The hydraulic conductivity profile along the femur is determined by the porosity. The hydraulic conductivity decreases with increased cement viscosity. 63 CEMENT FLOW DURING IMPACTION ALLOGRAFTING Further, it was assumed that the hydraulic conductivity does not change with the saturation. To verify this approach, cylindrical bone plugs with different porosities where modeled according to a cement intrusion experiment (Rey, Jr. et al., 1987). Different pressures where simulated and the intrusion depths at a saturation of s > 0.5 were compared favourably with experimental results (Figure 3.3). 12.0 -, - 0 " Br-.--•Or' -Rey et.al. 140kPa -e- Model 140kPa 40 45 50 55 60 65 70 Bone Porosity [%] 75 80 85 90 Figure 3.3: Cement intrusion depths as a function of bone porosity into cylindrical bone plugs. The predicted intrusion depths were on average 1 mm lower than the experimental data. 64 CEMENT FLOW DURING IMPACTION ALLOGRAFTING To model the cement flow in the neo-medullary canal (i.e. a non-porous media), ABACUS requires an "equivalent" intrinsic permeability Kp' to be estimated. This was done by combining Darcy's law for flow in porous media with the governing equation of cement flow in tubes. Darcy's Law can be written as: K'AP Q = ^Lr-A (4) TjL where AP is the applied pressure [Pa], L the length [m], A is the averaged cross-sectional area [m2] of the neo-medullary canal, Q is the volumetric flow rate [ m V 1 ] and r) is the dynamic viscosity [Pas]. The cement flow in tubes is governed by (Bohl, 1991): <*> where r is an average radius [m] of the canal. By substituting Equation (4) in (5) the "equivalent" intrinsic permeability is: Kr-r~ (6) The cement intrusion depths did not change when the cement flow simulated by the "equivalent" intrinsic permeability was included into the models of the cylindrical bone plugs (Figure 3.3) (Rey, Jr. etal., 1987). It was assumed that no cement flowed into the bone graft during retrograde filling. Cement pressurization was simulated by introducing cement at the top of the stem and allowing it to flow into the domain of the model (Figure 3.1). The pressurisation pressure magnitude was determined by matching the 65 CEMENT FLOW DURING IMPACTION ALLOGRAFTING cement volume that flowed into the model during pressurization with cement volume in the cement mantle and the bone graft from the cadaveric femur (Table 3.1). At four nodes the cement was allowed to drain freely to simulate cement loss at the proximal seal (Figure 3.1). Stem insertion was simulated by applying a time averaged pressure gradient along the stem. The pressure gradient was based on experimental results and was kept constant for 3s (McCaskie et al., 1997). It was assumed that between the end of pressurisation and the beginning of stem insertion no pressure drop in the model occurred and that no cement was lost proximally during stem insertion. No displacement of the solid phase was allowed during pressurisation and stem insertion. The model was validated by comparing the averaged cement intrusion depth predicted by the model with the averaged measured cement intrusion at different levels from the tested cadaveric femur. Simulations were performed to determine the effect of increased cement viscosity, increased pressure (25% and 50%), increased duration of the pressurisation (55 vs. 70 seconds) and uniform porosity (72%) (Table 3.2). To investigate the effect of the proximal seal on cement intrusion, the procedure was simulated with averaged uniform porosity and no loss at the seal. 66 Cement Flow During Impaction Allografting 3.3 RESULTS The predicted and measured cement intrusions at different sections for a pressure of 88kPa are compared in Figures 3.4 and 3.5. The overall averaged mean intrusion depth difference between the prediction and the cadaveric measurements for all sections was 1.1mm. The cement intrusion increased circumferentially from the neo-medullary canal to the periphery. Where the cement reached the endosteal surface, the flow direction changed and the cement followed the boundary of the model. The evolution of the cement intrusion profiles during pressurisation along the femur as a function of time is shown in Figure 3.5. Section 5 Section 6 Section 12 Section 13 Section 15 Section 16 Section 17 Section 18 — — — Measured FEM prediction Stem Endosteal surface Figure 3.4: Cross-sections along the femur comparing the measured and predicted cement intrusion. Specific locations are shown in Figure 3.2. Where the cement intrusion increased circumferentially an accurate prediction was possible. 6 7 CEMENT FLOW DURING IMPACTION ALLOGRAFTING 16 Figure 3.5: Evolution in time of the intrusion profile of the mean intrusion depths along the femur for the simulated cadaveric femur. For comparison the measured intrusion profile is shown. Ten seconds after the start of pressurisation the pressure gradient along the stem was developed with only minor changes for the remaining time (Figure 3.6). Proximally the pressure drop was 62% due to the seal. The pressure gradient increased with increased viscosity compared with the other simulations (Figure 3.6). The pressure gradient was uniform along the stem with a tight proximal seal. 68 CEMENT FLOW DURING IMPACTION ALLOGRAFTING 40000 35000 30000 _ 25000 CL § 20000 in V) 2> 15000 10000 5000 0 ! | ! • ' I t ! 1 i • ! 1 : | j M : 1 j x : | 1""4?lMS9? - - _ T i m a Cf* Time 10s 1 HUD JO -< "T: r • Time 16s ! '•• Time 29s End of pressurisation • McCaskie e ta l . 1997 — • — 5 5 s increased viscosity i ! : ; i . . . . . . . . j . . . . . . . . t ' i \ | 1 f \ """j — i — 1 j - • -\ 1 -I - j - i ! F i g u r e 3.6: Pressure gradients along the stem during cement pressurisation at different times of cement pressurisation. After 10s only small changes of the pressure gradient were predicted. The mean pressures measured at three locations along the femur during cement pressurisation during a primary THR's are included for comparison (McCaskie et al., 1997). By increasing the viscosity of the cement, the cement intrusion and the cement volume in the bone graft decreased compared with the cadaveric femur (Figure 3.7, Table 3.3). Increased cement pressure during pressurisation increased cement intrusion mainly proximally and the cement volume in the bone graft. More than two packages of cement were required to establish continuous cement flow during pressurization when the pressure was increased by 50%. Increased pressures resulted in a higher loss of cement volume at the seal. The 69 CEMENT FLOW DURING IMPACTION ALLOGRAFTING intrusion profile was not affected when the pressure was increased. Increased duration of the pressurization by 27% resulted only in a marginal increase of cement intrusion and cement volume in the graft (Figure 3.7 and Table 3.3). In contrast to the other simulations, the uniform porosity resulted in a change in the intrusion profile particularly proximally, in sections 1 to 7 and distally, sections 15 to 18 with (Figure 3.7). A tight proximal seal changed the proximal intrusion profile only marginally. FEM prediction femur Increased duration Increased viscosity Endosteal surface • - Increased pressure (50%) - - Increased pressure (25%) — Uniform porosity Figure 3.7: Intrusion profiles of the mean intrusion depths along the femur for the different simulation after stem insertion. 70 CEMENT FLOW DURING IMPACTION ALLOGRAFTING T a b l e 3.3: Summary of the cement volumes for the different simulations. Cement Volume [cm3] Injected Lost in graft in graft during through before after pressurization proximal stem stem seal insertion insertion Cadaveric femur simulated 78.6 53.3 25.3 37.2 Increased viscosity (late stem insertion) 42.5 27.7 14.8 22.6 Increased pressure (25%) 92.0 63.1 28.8 38.8 Increased pressure (50%) 110.2 77.9 32.3 42.3 Increased duration of pressurization 84.5 57.0 27.5 38.7 Uniform averaged graft porosity (72%) 102.6 84.3 18.3 29.0 Uniform averaged graft porosity tight seal 24.8 0.0 24.8 34.7 71 CEMENT FLOW DURING IMPACTION ALLOGRAFTING 3.4 DISCUSSION The importance of cement penetration into the impacted bone graft in this revision hip replacement technique is unknown. A degree of cement intrusion into the impacted allograft bone is likely important for the success of the procedure, whereas cement intrusion to the endosteal surface may prevent cortical remodelling (CHAPTER II). This finite element analysis investigated how cement intrusion can be controlled intraoperatively by varying cement viscosity, magnitude and duration of pressurization and graft porosity. To inject the same amount of cement found in the cadaveric femur into the model of an empty impacted allograft network in 55s, a pressure of 88kPa was required. This cement pressurisation pressure may be increased by the pressure required to displace remnants of fat and bone marrow entrapped in the impacted allograft bone, which could not be modeled with the finite element formulation available in ABAQUS. Most cement gun systems can generate pressure magnitudes of 340kPa and more (Churchill ef al., 2001). Intramedullary bleeding pressure does not affect cement intrusion substantially (Breusch et al., 2002). The predicted pressure profile along the stem was similar in magnitude compared with the mean pressures measured during primary THRs where the femoral canal was power brushed and lavaged (Majkowski et al., 1994; McCaskie ef al., 1997). The permeability data were obtained from unimpacted cancellous bone, which may be different for impacted allograft bone (Beaudoin ef al., 1991). If the 72 CEMENT FLOW DURING IMPACTION ALLOGRAFTING permeability-porosity relationship does not change with impacted allograft bone, a decreased permeability, for example, would simply result in an increase of the calculated pressure during pressurization since the cement flow was modeled with the linear Darcy's law. The assumed saturation-pressure relationship has been verified by simulating experimental data (Rey, Jr. et al., 1987), which were slightly underestimated (Figure 3.3). However, this may reflect the nature of the impacted bone graft where the flow resistance is increased by an arbitrary orientation of the trabeculae (Nauman etal., 1999). It was assumed that no cement was lost during stem insertion which implies that all the cement in the neo-medullary canal was forced into the impacted allograft. The high pressures reported during stem insertion support this assumption (Song etal., 1994; McCaskie etal., 1997; Churchill etal., 2001). For this simulation 32% of the total cement volume in the allograft bone and cement mantle were forced into the allograft during stem insertion. The finite element model predicted the cement intrusion into impacted bone graft accurately where cement intrusion increased circumferentially in a rather uniform manner (Figure 3.4). The accuracy in the other sections may be increased if the variation of the porosity within each section were modeled. Increased cement pressure and duration of pressurisation increased the depth of cement intrusion but maintained a similar profile (Figures 3.5 and 3.7). Late cement pressurisation and stem insertion decreased cement intrusion. The change in pressure gradient along the intramedullary canal was not sufficient to 73 CEMENT FLOW DURING IMPACTION ALLOGRAFTING affect the intrusion profile (Figures 3.6 and 3.7). Similar intrusion depths can be achieved by increasing the pressurisation pressure with higher viscosity cement. An altered intrusion profile was predicted with a uniform porosity along the femur. It appears that the cement intrusion of the distal two thirds of the stem is mainly determined by the graft permeability which is consistent with experimental observation where significant correlation between cement porosity and cement intrusion was found (CHAPTER II). Cement intrusion at the proximal third of the femur was mainly determined by the geometry (Figure 3.7). The proximal seal affected cement intrusion only marginally. The porosity appears to be the most effective way to alter the cement flow in the impacted allograft bone. The porosity can be controlled by the impaction force during graft impaction (CHAPTER II). The impaction force also determines the shear strength of the graft layer (Brewster et al., 1999). Therefore the effect of cement intrusion and graft porosity on graft layer properties and structural properties of the impaction allografting construct needs to be determined. With the cement viscosity, the pressure magnitude and the duration of pressurisation, the intrusion depth can be controlled intra-operatively. These variables depend on the magnitude of the pressure required to displace residual fluid in the trabecular network and the quality of the proximal seal, which are difficult to determine and may change with each patient. Excessive cement penetration can be avoided by limiting the cement volume injected during pressurisation. For this purpose the cement volume 74 CEMENT FLOW DURING IMPACTION ALLOGRAFTING injected into the femur and the losses at the seal need to be measured during pressurisation. It is not known how much cement intrusion into the impacted morsellized allograft bone is necessary for the success of the revision of THR's with impaction allografting. The cement intrusion up to the endosteal surface may enhance the strength of the construct and might be desirable at some levels, where cortical bone remodelling is not crucial. Based on this finite element analysis the intrusion profile along the stem is most effectively controlled by the graft porosity whereas the intrusion depths can be controlled by the injected cement volumes. However, experiments are required to investigate how different intrusion profiles and intrusion depths affect the strength and the biology of the impaction allografting construct. 75 CEMENT FLOW DURING IMPACTION ALLOGRAFTING 3.5 REFERENCES Baroud.G., Wu.J.Z., Bohner.M., Sponagel.S., and Steffen.T. (2003) How to determine the permeability for cement infiltration of osteoporotic cancellous bone. Med.Eng. Phys. 25, 283-288. Beaudoin.A.J., Mihalko.W.M., and Krause.W.R. (1991) Finite element modelling of polymethylmethacrylate flow through cancellous bone. J.Biomech. 24, 127-136. Bohl.W. (1991) Stoffstrome in Geschlossenen Rohrleitungen. In Technische Stromungslehre Pp. 119-156. Vogel, Wurzburg. Breusch.S., Heisel.C, Muller.J., Borchers.T., and Mau.H. (2002) Influence of cement viscosity on cement interdigitation and venous fat content under in vivo conditions: a bilateral study of 13 sheep. Acta Orthop.Scand. 73, 409-415. Brewster,NT., Gillespie.W.J., Howie.C.R., Madabhushi.S.P., Usmani.A.S., and Fairbairn.D.R. (1999) Mechanical considerations in impaction bone grafting. J.Bone Joint Surg.Br. 81, 118-124. Churchill,D.L., Incavo.S.J., Uroskie.J.A., and Beynnon.B.D. (2001) Femoral stem insertion generates high bone cement pressurization. Clin.Orthop. 393, 335-344. Elting.J.J., Mikhail.W.E., Zicat.BA, Hubbell.J.C., Lane.L.E., and House.B. (1995) Preliminary report of impaction grafting for exchange femoral arthroplasty. Clin.Orthop. 319, 159-167. Gie.G.A., Linder.L., Ling.R.S., Simon.J.P., Slooff.T.J., and Timperley.A.J. (1993) Impacted cancellous allografts and cement for revision total hip arthroplasty. J.Bone Joint Surg.Br. 75, 14-21. Grimm,M.J. and Williams,J.L. (1997) Measurements of permeability in human calcaneal trabecular bone. J.Biomech. 30, 743-745. Krause.W.R., Miller,J., and Ng.P. (1982) The viscosity of acrylic bone cements. J.Biomed.Mater.Res. 16, 219-243. Leopold,S.S. and Rosenberg,A.G. (2000) Current status of impaction allografting for revision of a femoral component. Instr.Course.Lect. 49, 111-118. 76 CEMENT FLOW DURING IMPACTION ALLOGRAFTING Majkowski.R.S., Bannister.G.C., and Miles,A.W. (1994) The effect of bleeding on the cement-bone interface. An experimental study. Clin.Orthop. 299, 293-297. Mann.K.A., Ayers.D.C, Werner.F.W., Nicoletta.R.J., and Fortino.M.D. (1997) Tensile strength of the cement-bone interface depends on the amount of bone interdigitated with PMMA cement. J.Biomech. 30, 339-346. Markolf.K.L. and Amstutz.H.C. (1976) Penetration and flow of acrylic bone cement. Clin.Orthop. 121, 99-102. McCaskie,A.W., Barnes,M.R., Lin.E., Harper.W.M., and Gregg,P.J. (1997) Cement pressurisation during hip replacement. J.Bone Joint Surg.Br. 79, 379-384. Nauman.E.A., Fong.K.E., and Keaveny.T.M. (1999) Dependence of intertrabecular permeability on flow direction and anatomic site. Ann.Biomed.Eng. 27, 517-524. Nguyen, H.V. and Durso.D.F. (1983) Absorption of water by fiber webs: an illustration of diffusion transport. Tappi Journal 66. 76-79. Rey.R.M., Jr., Paiement.G.D., McGann.W.M., Jasty.M., Harrigan.T.P., Burke.D.W., and Harris,W.H. (1987) A study of intrusion characteristics of low viscosity cement Simplex-P and Palacos cements in a bovine cancellous bone model. Clin.Orthop. 215, 272-278. Slooff.T.J., Huiskes.R., van Horn.J., and Lemmens.A.J. (1984) Bone grafting in total hip replacement for acetabular protrusion. Acta Orthop.Scand. 55, 593-596. Song.Y., Goodman,S.B., and Jaffe,R.A. (1994) An in vitro study of femoral intramedullary pressures during hip replacement using modern cement technique. Clin.Orthop. 302, 297-304. 77 CHAPTER IV MECHANICAL CHARACTERISTICS OF THE BONE-GRAFT-CEMENT INTERFACE AFTER IMPACTION ALLOGRAFTING Abstract: The host bone-graft-cement interface after impaction allografting has not been characterised, although it is a potential site of subsidence. After the impaction grafting procedure was performed in six cadaveric femurs, the specimens were sectioned in 6mm transverse sections and push-out tests were performed. The interface properties varied significantly along the femur largely due to different interface morphologies. Around the lesser trochanter, the apparent interface shear strength was highest and lowest around the tip of the stem. There was a significant positive correlation between the percentage cement contact and the apparent shear strength (1^=0.52). The sections failed in 69% of the cases through a pure interface failure, 19% failed with local cement layer and 12% with a local graft layer failure. The apparent interface strength was on average 89% lower than values reported for primary total hip replacements and were similar to cemented revisions proximally and lower distally. 78 MECHANICAL CHARACTERISTICS OF THE INTERFACE 4.1 INTRODUCTION From a mechanical perspective, stem subsidence after impaction allografting can occur through failure of any material between the stem and the cortex or any of the interfaces between materials. Various reasons for excessive stem subsidence have been reported, including cement mantle fracture (Masterson et al., 1997), insufficient graft impaction (van Biezen era/., 2000) and the incorporation and remodelling process of the graft layer (Franzen et al., 1995). Problems with cement mantle fractures have been addressed (Haddad and Duncan, 1999) and extensive research has focussed on graft layer properties (Berzins et al., 1996; Brodt et al., 1998; Giesen et al., 1999; Ullmark and Nilsson, 1999; Voor et al., 2000) and graft incorporation (Tagil and Aspenberg, 1998; Tagil et al., 1999; Lamerigts etal., 2000). Although important for the success of any THR (Krause et al., 1982; Dohmae et al., 1988; Ling, 1992; MacDonald et al., 1993; Mann et al., 1997; Mann et al., 2001), the interface between the host cortex and the graft material has not been addressed for the impaction allografting technique. For the primary THR, the shear strength of the cement-bone interface was found to be positively correlated with the amount of cement interdigitation into trabecular bone (Mann et al., 1997). Removal of trabecular bone, to simulate a revision scenario in a cadaveric model, decreased the shear strength of the cement-bone interface by up to 93% compared with primary THR (Dohmae et al., 1988). 79 MECHANICAL CHARACTERISTICS OF THE INTERFACE The interface morphology at the endosteal surface after impaction allografting was determined recently by histological analysis of cadaveric femurs (CHAPTER II). In addition to the impaction allograft-host bone interface, an impaction allograft-cement composite-host bone interface was identified. Around the middle third of the stem virtually the entire intramedullary canal was filled with a composite of allograft bone and cement, whereas below the tip of the stem the cement did not reach the endosteal surface. Given these variations of interface morphology along the femur, one would expect differences in mechanical interface properties. Therefore, we hypothesized that the mechanical characteristics of the impaction allografting-host bone interface would exhibit substantial differences along the length of the stem and that the degree of cement penetration of the endosteal cortex would explain a significant percentage of the variation. For the investigation, an in vitro human cadaveric model was used. 80 MECHANICAL CHARACTERISTICS OF THE INTERFACE 4 . 2 METHODS The femoral necks from six fresh frozen human cadaveric femora were osteotomized. Using a high-speed burr the cancellous bone was removed from the proximal femoral metaphysis and local diaphyseal lytic defects were created to simulate cavitary bone loss as seen in revision total hip arthroplasty. The endosteal surface roughness achieved was expected to be similar to that of a femur in a revision total hip arthroplasty as judged by two experienced revision hip arthroplasty surgeons. The high speed burr was used because of its almost routine use in revision total hip arthroplasty at our institution (Paprosky et al., 2001). Three surgeons performed the impaction allografting procedure, each on two cadaveric femora using the X-Change revision system (Howmedica Inc, Rutherford, NJ, USA). Prior to impaction, the intramedullary canal was occluded 2cm distal to the most distal lytic defect with a plastic plug and two Kirschner wires to avoid plug subsidence. Three to four fresh frozen femoral heads for each specimen were thawed in water, cut into 10x10x25mm blocks and morsellized in a Lere bone mill (DePuy, Warsaw, IN, USA). None of the articular cartilage, fat or bone marrow was removed from the morsellized allograft chips. After impaction two packages of low viscosity Simplex cement (Howmedica Inc, Rutherford, NJ, USA) were mixed and injected into the neo-medullary canal. After the canal had been filled in a retrograde manner, a proximal femoral seal was applied and the cement pressurised with a Prism II cement gun (DePuy, Warsaw, IN). The pressure was maintained until the viscosity of the cement was appropriate for the 8 1 MECHANICAL CHARACTERISTICS OF THE INTERFACE insertion of the double tapered polished collarless Exeter stem (Howmedica Inc, Rutherford, NJ, USA). After the cement had polymerized completely, anteroposterior and mediolateral radiographs were obtained to document the stem position. The femoral stem was then removed and the resultant void was filled with Simplex cement. The femurs were cut in 6mm thick transverse sections with a diamond saw (Exakt Technologies Inc., Oklahoma City, OK, USA). The sections were frozen at -20°C. Twenty-seven to thirty-six sections were obtained from each femur. From below the lesser trochanter every other section was selected for mechanical testing. Anatomical landmarks and the tip of the stem were used to match the sections to seven levels (Figure 4.1). The other sections were processed for histomorphometrical analysis, the results of which are reported in CHAPTER II. 82 MECHANICAL CHARACTERISTICS OF THE INTERFACE t i \ Level 1 t I | Level 2 [ Level 3 Level 4 IDE Polymeric Support Polymeric Plunger Actuatorl with LVDT Distal Femoral Section Level 5 i Level 6 Adjustable Support Proximal Load Cell / r ;.r Level 7 Figure 4.1: Schematic drawing of the specimens and the push-out setup. The sections from six specimens were matched to seven levels based on anatomical landmarks and the tip of the stem. The sections were tested in the setup shown on the right. Push-out tests were performed on these transverse sections to mechanically characterise the impaction allograft/ composite-host bone interface (Figure 4.1). To account for the irregular shape of the sections, plungers and supports were customised to fit each individual section. For this purpose, the perimeter of the endosteal cortex was copied to a 5mm thick acrylic-polymer plate. From that plate, the plunger and the support were machined by following the perimeter of the endosteal surface with a 3mm-mill bit. The clearance between plunger and distal endosteal surface was kept as small as possible but 8 3 MECHANICAL CHARACTERISTICS OF THE INTERFACE was at least 0.5mm. The proximal cortical ring was glued (LOCTITE, Div Henkel Canada Ltd, Etobicoke ON, Canada) to a support such that the plug was pushed in the direction from smaller canal diameter (i.e. distal) to larger canal diameter (i.e. proximal) (Figure 4.1). The push-out test was performed in a servo-hydraulic testing machine (DynaMight, Instron Corp, Canton, MA, USA) in displacement control at a rate of 0.5mm/s. Axial force was measured during the test by a load cell (Sensotec 250LBS, Columbus, Ohio, USA). The axial deflection of the polymeric supports was determined over the entire loading range after the test and was subtracted from the measured interface displacement (ID). The failure modes were determined by macroscopic inspection. Three different failure types were identified (Figure 4.2): Type I: Pure failure of the allograft/ composite-host bone interface. Type II: Failure of the allograft/ composite-host bone interface with local allograft/ cement composite failure. Type III: Failure of the allograft/ composite-host bone interface with local allograft failure. 84 MECHANICAL CHARACTERISTICS OF THE INTERFACE Type I Type II Type III Figure 4.2: Photograph of the three failure modes. Type I failure was a pure interface failure. In Type II failures, remnants of cement remained on the endosteal surface indicating a local allograft/ cement composite failure. Allograft particles remained on the endosteal surface in Type III failures, indicating a local allograft layer failure. Transverse photographs of the proximal and distal cortical rings were taken with a high-resolution digital camera (1600 x 1200 pixels, CoolPix 950 Nikon, Melville, NY, USA). From these photographs, perimeter lengths of the endosteum and the tapers were determined with IMAGE-PRO 4.5 (Mediacybernetics, Silver Spring, MD, USA) and custom software. The height of each section was measured with a dial gauge. The apparent area was calculated by multiplying the height by the average of the proximal and distal perimeter lengths. The average taper angle was obtained by averaging the distance between the smaller and larger perimeter divided by the height of the section. 85 MECHANICAL CHARACTERISTICS OF THE INTERFACE From the recorded load-displacement curves (APPENDIX D), the failure load was defined as the point where a zero slope was reached the first time which was, in general, the peak load. The stiffness was determined by a linear regression between 20% and 80% of the failure load and the failure energy by numerical integration of the load-displacement curve up to the peak load (Figure 4.3). The failure load, the stiffness, and the failure energy were divided by the apparent area to obtain the apparent interface shear strength (x a pp), apparent interface stiffness (k), and the apparent interface failure energy (G). Interface shear (x) and tensile stresses (a) were calculated using conventional stress analysis (Gere and Timoshenko, 1990) (APPENDIX F). From the two adjacent levels which were processed for histomorphometrical analysis, the average porosity and the average percentage cement contact (%c) were calculated (i.e. %c=100%; the cement reached the entire endosteal surface circumferentially; %c=0%; no cement reached the endosteal surface) (CHAPTER II). Apparent interface shear strength ( T a p p ) , interface shear stress (x), tensile stress (a), interface displacement to failure (ID), stiffness (k) and failure energy (G) were compared using a one-way repeated measures analysis of variance (ANOVA) with the levels as the factor. Parametric factorial one-way ANOVA was used to compare these parameters (x a p p , x, a, ID, k, G) with the failure types as factors. Stepwise linear multiple regression models were used to determine the mechanical response parameters as a function of average percentage cement 86 MECHANICAL CHARACTERISTICS OF THE INTERFACE contact (%c) and taper angle. Student-Newman Keuls analysis was used for post hoc comparisons and the significance level was 95%. 87 MECHANICAL CHARACTERISTICS OF THE INTERFACE 4.3 RESULTS In general the specimens showed an initial linear load-displacement response up to the peak load followed by a large strain softening region until complete debonding of the interface (Figure 4.3). There was substantial variability in the measured parameters among the different specimens. The average percentage cement contact (%c) was not significantly different between levels 1 to 4. However, these levels had a significantly higher average percentage cement contact (%c) compared with levels 6 to 7 (p< 0.01) (Figure 4.4). There was a pure a I log raft-bone interface (%c=0) at level 7 below the tip of the stem. Figure 4.3: Typical load-displacement response (see APPENDIX D for all load-displacement curves). The push-out tests were characterized by the initial stiffness calculated between 20% and 80% of the peak load, the peak load, and the failure energy. 88 MECHANICAL CHARACTERISTICS OF THE INTERFACE The apparent strength x a p p and the shear stress x were significantly-different among the levels with the highest shear strength at level 3 and lowest at levels 6 and 7 (p<0.02) (Figure 4.5). The tensile stresses a at level 1 were significantly higher than the other levels (p<0.02) and were significantly lower than the shear stresses T at all levels (p<0.0001). Similar to the shear stress x the lowest initial stiffnesses were at levels 6 and 7 (p<0.07). The highest energy to failure was observed at level 3, and the lowest was at levels 6 and 7 (p<0.07). The average displacement to failure was 0.9mm (SD 0.41) and was not significantly different between the levels (p= 0.41). 100% n 1 : : 1 : : ; , Figure 4.4: Percentage cement contact %c with the endosteal surface as a function of different levels. The error bars indicate the standard deviation. Highest percentage cement contact was found proximally and then gradually decreased to a pure allograft-bone interface below the tip of the stem. 89 MECHANICAL CHARACTERISTICS OF THE INTERFACE The apparent interface shear strength x a p p and shear stress x were significantly correlated with the percentage cement contact %c, with the relationships explaining 52% of the variance for each parameter (Figure 4.6 for x ; Table 4.1 for both). There was a poor correlation between tensile stress a and percentage cement contact %c (r*= 0.17). In addition, the percentage cement contact %c explained 36% of the variability of the stiffness and 35% of the 90 MECHANICAL CHARACTERISTICS OF THE INTERFACE variability of the failure energy. When the taper angle was included into a stepwise linear multiple regression of the measured parameters as a separate variable, it was not significant and increased the quality of the regression only marginally. The displacement to failure was not correlated with %c. The linear regression parameters are summarised in Table 4.1. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Cement contact area %c [%] Figure 4.6: Scatterplot of shear T and tensile CT stresses at failure as a function of percent cement contact %c. The failure stresses increased with increased percentage cement contact. The lines indicate linear regressions. The regression parameters are summarized in Table 4.1. 91 MECHANICAL CHARACTERISTICS OF THE INTERFACE Table 4.1 Regression parameters for the linear regression (y = a + b * %c) where %c is the percentage cement contact with the endosteal surface. Parameter a b r2 Apparent strength, T a p p [MPa] 0.248 0.842 0.527 Shear, x [MPa] 0.240 0.817 0.519 Tension, a [MPa] 0.047* 0.151 0.168 Stiffness, k [Mpa mm"1] 0.510* 1.982 0.355 Failure energy, G [Nmm] 0.155* 0.694 0.352 * Indicates p < 0.18 for the other values p < 0.05 69% of the sections exhibited a pure interface failure (Type I) without composite or graft layer failure. Local cement failure (Type II) was found in 19% of the sections and 12% of the sections showed local graft failure (Type III) (Table 4.2). Sections with Type II failures were significantly stronger and tougher compared with Type I and Type III failures (p< 0.03) (Table 4.3). Stiffness (p=0.16), interface displacement (p=0.55) and percentage contact %c (p=0.06) were not statistically different among the groups. None of the measured parameters were significantly different between Type I and Type III failure. 92 MECHANICAL CHARACTERISTICS OF THE INTERFACE Table 4.2 Interface failures types for the composite/ graft-bone interface at different levels. Type 1 Type II Type III Level 1 1 (2%) 3 (7%) 2 (5%) Level 2 3 (7%) 2 (5%) 1 (2%) Level 3 5 (10%) 1 (2%) 0 Level 4 5 (12%) 1 (2%) 0 Level 5 4 (14%) 1 (2%) 1 (2%) Level 6 6 (10%) 0 0 Level 7 5 (12%) 0 1 (2%) Total 29 (69%) 8 (19%) 5 (12%) Table 4.3 Interface strength, stiffness, failure energy and ID grouped into the failure Type I, II and III. Mean and (SD) Type I Type II Type III Apparent strength, x a p p [MPa] 0.57 (0.42) 1.19 (0.24)* 0.53 (0.42) Shear, x [MPa] 0.56 (0.42) 1.12 (0.26)* 0.51 (0.42) Tension, a [MPa] 0.07 (0.06) 0.33 (0.21)* 0.11 (0.11) Stiffness, k [Mpa mm"1] 1.30 (1.16) 2.39 (1.3) 1.53 (1.82) Failure energy, G [N mm"1] 0.40 (0.37) 1.02 (0.5)* 0.34 (0.34) Displacement ID [mm] 0.94 (0.43) 0.82 (0.31) 0.76 (0.37) Percent contact c% [%] 0.43 (0.41) 0.80 (0.21)* 0.57 (0.32) * Indicates significant higher values (p < 0.03) Thirteen of the forty-two sections below level 4 had a pure allograft-host bone interface (0% contact). The averaged apparent interface strength of these sections was x a p p = 0.28MPa (SD 0.15), the shear stress x= 0.27MPa (SD 0.15), the tensile stress a= 0.04MPa (SD 0.03), the failure energy G=0.19Nmm"1 (SD 93 MECHANICAL CHARACTERISTICS OF THE INTERFACE 0.12), the stiffness k= 0.51 MPa mm"1(SD 0.69) and the interface displacement ID=1.1mm (SD 0.43). There was no correlation between impacted allograft porosity with the mechanical response parameters (T a p p , x, a, ID, k, G) for these thirteen sections (CHAPTER II). For all tests, there was a moderate correlation between apparent shear strength T a p p and the stiffness (r2= 0.56) (Figure 4.7). CD CL 1.8 1.6 1.4 S 1.2 Q . Q. OJ JZ 1.0 -.—1 O ) a CD 55 0.8 c 5 0.6 < 0.4 0.2 0.0 y = 0.258x + 0.299 0 R2 = 0.5607 o o o o r o o - ° O o o o 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 Stiffness [MPa/mm] 4.5 5.0 Figure 4.7: Scatterplot of the apparent interface strength T A P P as a function of interface stiffness k. A moderate correlation was found between strength and stiffness. The line indicates the linear regression. 94 MECHANICAL CHARACTERISTICS OF THE INTERFACE 4.4 DISCUSSION The impaction allograft-host bone interface and/or the impaction allograft-cement composite-host bone interface is a potential site of subsidence for this revision hip reconstruction. The cement-bone interface has been extensively studied for primary THR, but has been neglected for the impaction allografting procedure. This study may help to enhance our understanding of the technique, and should be important for the development and evaluation of bone graft substitutes. As with other biomechanical investigations, this study has some limitations. The different sections were pushed from the smaller to the larger diameter, which is in the opposite direction from which the femur is loaded in vivo. This technique results in a mixed mode (combined shear and tensile) failure of the interfaces and has been used for primary THR (Dohmae et al., 1988). Mixed mode failures are clearly clinically relevant since in addition to shear stresses significant tensile stresses were predicted by finite element analyses at the host bone interface of primary total hip replacements (Verdonschot and Huiskes, 1996; Chang et al., 1998). The tensile and shear stress components were calculated approximately based on the apparent shear stress x app and an averaged taper angle (APPENDIX F). Pure shear and plane stress was assumed and no movement effects were included in the calculations. The majority of stem subsidence occurs in the first postoperative year which indicates that biological processes may alter allograft and interface properties (Nelissen et al., 2002; 95 MECHANICAL CHARACTERISTICS OF THE INTERFACE Ornstein et al., 2003). These processes were not included in the model and therefore the reported properties are only applicable to an immediately postoperative scenario. The amount of cement penetration into the graft varied considerably along the length of the femur. At levels 2 and 3 the percentage cement contact was on average 86% and then gradually decreased to a pure al log raft-bone interface below the tip of the stem. The relatively large variations of percentage cement contact at levels 4 and 5 were most likely due to the variation in the impacting technique. In four out of six specimens the distal impactors were used past the recommended distal impaction line, which resulted in tighter packing of the bone graft around the distal third of the stem (X-Change System Operative Technique, Howmedica Inc, Rutherford, NJ, USA). Although only a moderate correlation was found between cement penetration and allograft porosity (r=0.5), this may have decreased the cement penetration into the impacted allograft bone (CHAPTER II) Similar to the present study, Dohmae et al. measured cement-bone interface shear strengths of primary THR by push-out tests (Dohmae ef al., 1988). They reported shear strengths between 4MPa and 8MPa which were in agreement with others (MacDonald ef al., 1993; Mann et al., 1997; Mann ef al., 1999; Mann et al., 2001). After testing, they removed the cement plug and enlarged the endosteal diameter with a burr and the sections were re-cemented to simulate a cemented revision, and repeated the same procedure to simulate a re-revision. The first revision reduced the interface shear strength by 80% and 96 MECHANICAL CHARACTERISTICS OF THE INTERFACE the re-revision by 93% due to the loss of cement- trabecular bone interdigitation. After impaction allografting the average apparent interface shear strength x a p P measured in the present study was on average 89% lower than in a primary THR and appears to suggest that the values are similar to a cemented revision proximally and are lower distally (Dohmae et al., 1988) (Figure 4.8). x y ^. sa — B -- ^ ^ ^ — P r e s e n t study - B - Primary THR i -A -Cemented Revision -©-Cemented Re-Revison \ ; ^ < X 1 — ~* —e-—._____ G - © - — : — • Figure 4.8: Comparison of the apparent interface shear strength xapP after impaction allografting with interface strength of a primary THR a cemented revision and a cemented re-revision. The interface strengths of after impaction allografting were between a cemented revision and re-revision. The values for the primary THR, the revision and re-revision were adapted from Dohmae etal. (Dohmae ef al., 1988). 97 MECHANICAL CHARACTERISTICS OF THE INTERFACE The correlations between the measured parameters and percentage cement contact %c showed that cement penetration up to the endosteal surface increases the strength, stiffness and toughness of the interface (Table 4.1). This situation may be viable biologically as it was shown previously that inorganic bone or hydroxyapatite particles embedded in bone cement could be replaced by new bone to form a viable cement-bone interface (Dai et al., 1991; Kwon et al., 1997). In these studies, the shear strength of the viable interface was significantly higher compared with cement only. However, it is not known if these findings are applicable to the impaction allografting case. One third of the tested sections, all below level 4, had a pure allograft-bone interface (%c=0). The average shear strength of this interface was 96% lower compared with the data reported for primary THR and only 50% of the strength reported for a re-revision (Dohmae et al., 1988). Brewster et al. measured shear strength of impacted allograft bone as a function of applied normal compressive stresses in the range from 17.6kPa to 141.6kPa (Brewster ef al., 1999). Assuming that any residual normal compressive stresses within the tested sections were below 141.6kPa, the shear strengths of the impacted allograft bone reported by Brewster et al. were 95% to 60% lower than the averaged apparent interface shear strength x a p p for the pure allograft-bone interface. This suggests that in shear, allograft failure occurs before the interface debonds (Figure 4.9). Compressive strengths of trabecular bone/ cement composites are substantially higher (between 29MPa and 50MPa) than the 98 MECHANICAL CHARACTERISTICS OF THE INTERFACE measured interface strengths (Jofe et al., 1991). With tensile strengths for the composite expected to be of similar magnitude, the composite-host bone interface, the pure allograft interfaces and the allograft layer would fail before the composite. To put the measured interface stresses in perspective to the host bone interface stresses which might occur during daily activities, the predicted Mohr-Coulomb failure envelopes for the impacted allograft, the pure allograft-host bone interface and the composite-host bone interface were constructed and illustrated in Figure 4.9. These predictions suggest that the static averaged host bone interface stresses (as determined by a finite element analysis of a primary THR) would result in failure of both, the pure allograft and the composite-host bone interface after impaction allografting (Chang et al. 1998) (Figure 4.9). While simplified in approach, if one can assume the failure predictions are valid, this raises significant concerns as the ability of the allografting technique to support the expected loads. However, the interface stresses after impaction allografting at the host bone interface are currently not known. In contrast to primary THRs, only small tensile stresses are expected to be transferred from the stem to the host bone interface by the impacted allograft bone, and therefore the host bone interface stresses might be different after impaction allografting. Clearly more research is required to clearly understand and valid the load transfer at these multiple interfaces. 9 9 MECHANICAL CHARACTERISTICS OF THE INTERFACE Stair descent • Stair ascent • c o m p o s i t e h o s t b o n e in te r face (%c> 9 0 % ) p u r e a l lograf t h o s t b o n e in te r face i m p a c t e d a l lograf t B r e w s t e r et a l . 1 9 9 9 + level walking Figure 4.9: Mohr-Coulomb failure envelopes for impacted bone graft, the pure allograft-host bone interface and the composite-host bone interface. The proximal cement-bone interface stresses were determined by finite element analysis of primary total hip replacement for level walking, stair ascent and stair descent activities (Chang ef al., 1998). The pure allograft interface failure envelope was estimated from the averaged measured apparent shear strength (0.28MPa) and assuming that the tensile strength does not exceed the maximum measured tensile stress (0.12MPa). Similarly, the composite interface (%c>90%) failure envelope was based on the averaged measured apparent shear strength (1.22MPa) and the maximum measured tensile stress (0.26MPa), which was identical to the tensile strength for the cement-bone interface of primary THRs with no cement interdigitation (Mann et al., 2001). Tick marks on axes indicate 1MPa. The allograft porosity of the sections with a pure impaction allograft-host bone interface (%c=0%) was not correlated with the apparent interface shear strength, which suggests that increased graft impaction does not enhance the interface shear strength. Similar to the interface strength, the average initial interface stiffness k of 1.54MPa mm"1 (SD 1.32) was only 94% of that reported for primary THR (Mann 100 MECHANICAL CHARACTERISTICS OF THE INTERFACE et al., 1998). This may lead to significantly larger motion at the host bone interface for the impaction allografting procedure compared with a primary THR. This excessive interface motion may result in fibrous tissue formation at the host bone interface observed in autopsies and biopsies after impaction allografting (Linder, 2000). There was a moderate correlation between apparent interface shear strength and stiffness (Figure 4.7). Based on this relationship it may be possible in animal experiments to estimate the interface shear strength in a non-destructive manner by measuring the interface stiffness. In the sections of the Type II failures, the cement was able to interdigitate with local remnants of trabecular bone or the rough, porous cortex resulting in local failure of the cement with significantly increased apparent interface strength and energy to failure. This is in agreement with primary THR where the amount of cement interdigitation was positively correlated with interface strength (Mann etal., 1999). This investigation showed that mechanical interface properties vary significantly along the femur due to different interface morphologies. Although, cement-endosteal surface contact enhanced the allograft-cement composite-bone interface strength, it was significantly weaker compared to a primary THR and may fail in tension. However, bone ingrowth would be expected to enhance the interface properties which might be the key to the success of the impaction allografting technique. 101 MECHANICAL CHARACTERISTICS OF THE INTERFACE 4.5 REFERENCES Berzins.A., Sumner.D.R., Wasielewski.R.C., and Galante.J.O. (1996) Impacted particulate allograft for femoral revision total hip arthroplasty. In vitro mechanical stability and effects of cement pressurization. J.Arthroplasty 11, 500-506. Brewster,NT., Gillespie.W.J., Howie.C.R., Madabhushi.S.P., Usmani.A.S., and Fairbairn.D.R. (1999) Mechanical considerations in impaction bone grafting. J.Bone Joint Surg.Br. 81, 118-124. Brodt.M.D., Swan,CC, and Brown,T.D. (1998) Mechanical behavior of human morselized cancellous bone in triaxial compression testing. J.Orthop.Res. 16, 43-49. Chang,P.B., Mann.K.A., and Bartel,D.L. (1998) Cemented femoral stem performance. Effects of proximal bonding, geometry, and neck length. Clin.Orthop. 355, 57-69. Dai.K.R., Liu.Y.K., Park.J.B., Clark.C.R., Nishiyama.K., and Zheng.Z.K. (1991) Bone-particle-impregnated bone cement: an in vivo weight-bearing study. J.Biomed. Mater.Res. 25, 141-156. Dohmae.Y., Bechtold.J.E., Sherman,R.E., Puno.R.M., and Gustilo.R.B. (1988) Reduction in cement-bone interface shear strength between primary and revision arthroplasty. Clin.Orthop. 236, 214-220. Franzen,H., Toksvig-Larsen.S., Lidgren.L, and Onnerfalt.R. (1995) Early migration of femoral components revised with impacted cancellous allografts and cement. A preliminary report of five patients. J.Bone Joint Surg.Br. 77, 862-864. Frei.H., Mitchell,P., Masri.B.A., Duncan.C.P., and OxIand.T.R. (2003) Allograft impaction and cement penetration with impaction allografting: An in vitro model. Orthopaedic Research Society 49th Annual Meeting, New Orleans, LA. Gere.J.M. and Timoshenko.S.P. (1990) Mechanics ot Material. PWS Publishing Company, Boston, MA. Giesen.E.B., Lamerigts.N.M., Verdonschot.N., Buma.P., Schreurs.B.W., and Huiskes.R. (1999) Mechanical characteristics of impacted morsellised bone grafts used in revision of total hip arthroplasty. J.Bone Joint Surg.Br. 81, 1052-1057. 102 MECHANICAL CHARACTERISTICS OF THE INTERFACE Haddad.F.S. and Duncan,CP. (1999) Impaction allografting of the proximal femur: fact or fad? Orthopedics. 22, 855-858. Jofe.M.H., Takeuchi.T., and Hayes.W.C. (1991) Compressive behavior of human bone-cement composites. J.Arthroplasty 6 , 213-219. Krause.W.R., Krug.W., and Miller,J. (1982) Strength of the cement-bone interface. Clin.Orthop. 163, 290-299. Kwon.S.Y., Kim.Y.S., Woo.Y.K., Kim.S.S., and Park.J.B. (1997) Hydroxyapatite impregnated bone cement: in vitro and in vivo studies. Biomed. Mater.Eng 7, 129-140. Lamerigts.N.M., Buma.P., Huiskes.R., Schreurs.W., Gardeniers.J., and Slooff.T.J. (2000) Incorporation of morsellized bone graft under controlled loading conditions. A new animal model in the goat. Biomaterials 21, 741-747. Linder.L. (2000) Cancellous impaction grafting in the human femur: histological and radiographic observations in 6 autopsy femurs and 8 biopsies. Acta Orthop.Scand 71, 543-552. Ling.R.S. (1992) The use of a collar and precoating on cemented femoral stems is unnecessary and detrimental. Clin.Orthop. 285, 73-83. MacDonald,W., Swarts.E., and Beaver.R. (1993) Penetration and shear strength of cement-bone interfaces in vivo. Clin.Orthop. 286, 283-288. Mann.K.A., Allen,M.J., and Ayers.D.C. (1998) Pre-yield and post-yield shear behavior of the cement-bone interface. J.Orthop.Res. 16, 370-378. Mann.K.A., Ayers.D.C, Werner.F.W., Nicoletta.R.J., and Fortino.M.D. (1997) Tensile strength of the cement-bone interface depends on the amount of bone interdigitated with PMMA cement. J.Biomech. 30, 339-346. Mann.K.A., Mocarski.R., Damron.L.A., Allen,M.J., and Ayers.D.C. (2001) Mixed-mode failure response of the cement-bone interface. J.Orthop.Res. 19, 1153-1161. Mann.K.A., Werner.F.W., and Ayers.D.C. (1999) Mechanical strength of the cement-bone interface is greater in shear than in tension. J.Biomech. 32, 1251-1254. 103 MECHANICAL CHARACTERISTICS OF THE INTERFACE Masterson,E.L, Masri,B.A., and Duncan,CP. (1997) The cement mantle in the Exeter impaction allografting technique. A cause for concern. J.Arthroplasty 12, 759-764. Nelissen.R.G., Valstar.E.R., Poll.R.G., Garling.E.H., and Brand,R. (2002) Factors associated with excessive migration in bone impaction hip revision surgery: a radiostereometric analysis study. J.Arthroplasty 17, 826-833. Ornstein,E., Franzen.H., Johnsson.R., Stefansdottir.A., Sundberg.M., and Tagil,M. (2003) Hip revision with impacted morselized allografts: unrestricted weight-bearing and restricted weight-bearing have similar effect on migration. A radiostereometry analysis. Arch. Orthop. Trauma Surg. 123, 261-267. Paprosky.W.G., Weeden.S.H., and Bowling,J.W., Jr. (2001) Component removal in revision total hip arthroplasty. Clin.Orthop. 393, 181-193. Tagil,M. and Aspenberg.P. (1998) Impaction of cancellous bone grafts impairs osteoconduction in titanium chambers. Clin.Orthop. 352, 231-238. Tagil,M., Johnsson.R., Stromqvist.B., and Aspenberg,P. (1999) Incomplete incorporation of morselized and impacted autologous bone graft: a histological study in 4 intracorporally grafted lumbar fractures. Acta Orthop.Scand. 70, 555-558. Ullmark.G. and Nilsson.O. (1999) Impacted corticocancellous allografts: recoil and strength. J.Arthroplasty 14, 1019-1023. van Biezen,F.C, ten Have.B.L, and Verhaar.J.A. (2000) Impaction bone-grafting of severely defective femora in revision total hip surgery: 21 hips followed for 41-85 months. Acta Orthop.Scand. 71, 135-142. Verdonschot.N. and Huiskes.R. (1996) Mechanical effects of stem cement interface characteristics in total hip replacement. Clin.Orthop. 329, 326-336. Voor.M.J., Nawab.A., Malkani.A.L, and Ullrich,CR. (2000) Mechanical properties of compacted morselized cancellous bone graft using one-dimensional consolidation testing. J.Biomech 33, 1683-1688. 104 C H A P T E R V BIOLOGICAL AND MECHANICAL CHANGES OF THE BONE-GRAFT-CEMENT INTERFACE AFTER IMPACTION ALLOGRAFTING Abstract: In impaction allografting, the host bone interface may consist of morsellized allograft alone or as a composite with bone cement. The objective of this study was to investigate the host bone temporal changes in the interface for these two materials in a rat bone chamber model. To simulate the impaired endosteal circulation after impaction allografting, bone chambers were tightened bilaterally to the endosteal surfaces of proximal tibiae of mature rats and filled with pure allograft or cement-allograft composite. The composite-host bone interface strength was significantly higher at 3 weeks and was higher than the allograft construct failure. Limited allograft, but extensive periosteal remodelling, was observed at 3 weeks which resulted in a significantly increased cortical porosity and cortical thickness. The allograft porosity decreased significantly at 6 weeks indicating extensive remodelling of the allograft bone. Little or no remodelling of the graft particles in the cement was found. At 6 weeks a new medullary canal was formed and the endosteal cortex was partially absorbed. Endosteal absorption resulting in medullary canal widening in primary THR may be responsible for clinically unstable stems after impaction allografting. 105 BIOLOGICAL AND MECHANICAL CHARACTERISTICS 5.1 INTRODUCTION Impaction allografting is an attractive technique for the treatment of failed total hip replacements (THR) (Franzen et al., 1995; Knight and Helming, 2000; Boldt et al., 2001; Fetzer et al., 2001; de Roeck and Drabu, 2001; Ullmark et al., 2002). In the technique, morsellized allograft bone is impacted into the medullary canal to fill the osteolytic defects, restore the integrity of the femur and provide a scaffold for bone ingrowth and remodelling. Postoperatively and during allograft incorporation and remodelling, the revision construct needs to support the loads to which it is subjected. If any material between the stem and the cortex or any of the interfaces between materials fails, the stem subsides. Stem subsidence exceeding 10mm has been reported to be a major complication related to impaction allografting (Elting et al., 1995; Leopold and Rosenberg, 2000; Pekkarinen etal., 2000; Leopold etal., 2000). A concern with the impaction allografting technique is the incomplete remodelling of the allograft bone. Autopsy and biopsy specimens have demonstrated that the allograft bone remodels only a few millimetres peripherally and the inner zone consists of mainly necrotic allograft bone partially embedded within fibrous tissue (Ling et al., 1993; Nelissen et al., 1995; Linder, 2000; Ullmark and Obrant, 2002). Tagil and Aspenberg showed in a bone chamber model that this fibrous tissue enhanced compressive strength of the allograft layer (Tagil and Aspenberg, 2001). However, the long-term behaviour of this fibrous tissue bone composite is not known and may not facilitate re-revision. 106 BIOLOGICAL AND MECHANICAL CHARACTERISTICS Extensive research has focussed on characterising the allograft layer in vitro (Brodt et al., 1998; Brewster et al., 1999; Giesen et al., 1999; Voor et al., 2000), allograft incorporation (Tagil and Aspenberg, 1998; Lamerigts et al., 2000; Tagil et al., 2000; van der Donk S. et al., 2002), and on the evaluation of bone graft substitutes (Verdonschot et al., 2001; Griffon et al., 2001a; Griffon et al., 2001b; Bolder et al., 2002; Pratt et al., 2002). Although important for the success of any THR, the interface between the host cortex and the graft material has not been addressed for impaction allografting. The interface morphology at the endosteal surface after impaction allografting was determined by histological analysis of cadaveric femurs described in CHAPTER II. In addition to the impacted allograft-host bone interface, an impacted allograft-cement composite-host bone interface was identified. Around the middle third of the stem virtually the entire intramedullary canal was filled with a composite of allograft bone and cement, whereas below the tip of the stem the cement did not reach the endosteal surface. Cement penetration to the endosteal cortex was shown to improve the mechanical strength of the host bone interface and this feature of the construct may be important for its success (CHAPTER IV). However, the changes in this interface with healing are not known. It has been shown in primary THRs that inorganic bone or hydroxyapatite particles embedded in bone cement could be replaced or bonded to bone to form a viable cement-bone interface (Dai et al., 1991; Oonishi et al., 2001). The shear strength of this viable interface was 107 BIOLOGICAL AND MECHANICAL CHARACTERISTICS significantly higher compared with cement only (Dai et al., 1991). However, poor bone quality and impaired endosteal circulation in a revision scenario may prevent the remodelling or bonding of the allograft particles (Gorab et al., 1993). To address the bone incorporation and remodelling of the allograft and allograft-cement composite with the host bone interface, a bone chamber model for revision arthroplasty was developed. The specific hypotheses of this study were: i. The host bone grows into the allograft-cement composite and enhances the interface strength by forming a viable interface. ii. Bone remodelling and/ or fibrous tissue ingrowth into the allograft increases the shear strength. iii. The strength of the allograft-cement composite construct is higher than the pure allograft construct. 108 BIOLOGICAL AND MECHANICAL CHARACTERISTICS 5.2 METHODS The bone chamber model was a modification of one already successfully used to study impacted allograft bone in the rat (Tagil and Aspenberg, 1998). In contrast to two subcortical bone ingrowth openings, a microenvironment was created which allowed bone ingrowth only from the endosteal surface to simulate a damaged endosteal circulation (Figure 5.1). Figure 5 .1 : Schematic diagram and photograph of the rat bone chamber model. The materials and interfaces observed after impaction allografting are re-established in a bone chamber. The endosteal circulation is impaired by a tight fit of the bone chamber to the endosteal surface. 109 BIOLOGICAL AND MECHANICAL CHARACTERISTICS A hollow threaded bone chamber (outer 02.8mm, inner 02.2mm, length 5mm, 316L stainless steel) was screwed into the proximal tibia with the end tightened to the endosteal surface to create the microenvironment. The same materials and interfaces found after impaction allografting were established in the bone chambers (Figure 5.1) (CHAPTER II). Approval was obtained from the University of British Columbia Committee on Animal Care (Protocol #A02-0104) prior to performing the study. Bone chambers were placed in both tibiae of thirty-three male mature Sprague-Dawley rats (480-550g), containing either a bone allograft or a bone allograft PMMA cement composite. Twenty-two rats were anaesthetized with Isoflurane and under aseptic conditions a 10mm incision was made over the anterior aspect of the proximal tibia. The muscles were detached circumferentially for the insertion of a drill guide (APPENDIX C). The guide ensured that the 2.7mm hole was drilled perpendicular to the lateral endosteal surface, positioned just distally to the insertion of the medial collateral ligament. The bone chambers were secured to both tibiae with a custom internal fixator. Inside the bone chamber, the endosteum and part of the endosteal cortex were removed with a flattened mill bit. The morsellized cancellous bone grafts for use in the bone chambers were harvested from the distal femur and proximal tibia from mature male Sprague-Dawley rats and were compressed in a 2mm diameter metal chamber for 2min with 80N to remove fat and bone marrow. The compressed graft was stored at -80° Celsius for several days. At the day of 110 BIOLOGICAL AND MECHANICAL CHARACTERISTICS surgery, the graft was thawed and approximately 6mm3 was loosely packed into a cylindrical Teflon mould (2mm diameter, 6mm long). A pre-manufactured cylindrical cement plug (2mm diameter, 3mm long, Simplex P) with a notch for mechanical testing was fixed to the bone graft with a small amount of low viscosity cement allowing the cement to interdigitate approximately 0.5mm with the graft. After curing of the cement, the bone graft with the plug was removed from the Teflon mould and inserted into the bone chamber. For the allograft-cement composite approximately 6mm3 of bone graft was mixed with 12mm3 methylmethacrylate powder, the monomer liquid was added and the cement graft paste was inserted into the Teflon mould. Similar to the bone graft a cement plug was attached to the sample and after curing removed from the mould. The 2mm diameter of the Teflon mould ensured clearance of 0.1mm between the sample and the bone chamber. The samples were approximately 5mm in height including the cement plug and were randomly inserted in the bone chamber in the right or left tibiae. After the graft and a spring were inserted, the bone chambers were closed with PMMA lids. The spring, which was inserted to keep the samples in place, exerted a constant compressive force of 1N, which was approximately 20% of the rat's body weight. The wound was closed with subcutaneous stitches and 5ml of saline was administered subcutaneously to account for blood loss during surgery. For three days 0.05ml buprenorphine was administered twice a day. 111 BIOLOGICAL AND MECHANICAL CHARACTERISTICS Eleven rats were operated post-mortem using the same surgical protocol. The other rats were euthanized with C 0 2 after 3 and 6 weeks. The tibiae of three of the eleven rats in each group (0 week, 3 weeks or 6 weeks) were processed for non-decalcified histology. After fixing the specimens in 10% buffered formalin for two days, the bone cement was dissolved with acetone to facilitate infiltration of the resin (Technovit 7200, Kulzer Ltd, Wehrheim, Germany). The specimens were dehydrated in 70%, 95% and 100% ethanol, and embedded in Technovit 7200. Sections were cut and ground to 50um and stained according to Goldner (APPENDIX A). From the mid section of the chamber, graft porosity, percentage fibrous tissue, cortical porosity and cortical thickness were determined with IMAGE-PRO 4.5 (Mediacybernetics, Silver Spring, MD, USA). The other specimens were mechanically tested in pure torsion. After the lids and the springs were removed from the bone chambers, the tibiae were soaked in saline for 1 hour to prevent blood coagulation. The top of the bone chambers were screwed into the testing rig, which assured an accurate alignment of the specimen. A rotational displacement of 0.5°/s with compressive dead weight of 1N was applied until failure. The torque was measured with a custom load cell (non-linearity 2%; range 11Nmm; APPENDIX B). After testing, the samples were processed for non-decalcified histology and analysed histomorphometrically as described above. From the recorded torque-rotation curves, the failure load was defined as the point where a zero slope was reached the first time. In five of the 42 tested 112 BIOLOGICAL AND MECHANICAL CHARACTERISTICS samples (i.e. total 48; 6 specimens lost) the failure load was defined as when the slope decreased the first time within the range of the failure rotations determined by the torque-displacement curves from the other specimens. The stiffness (k) was determined by a linear regression between 20% and 80% of the failure load and the failure energy (G) by numerical integration of the torque-rotation curve up to the peak load (Figure 5.2). The apparent shear stress was calculated using conventional stress analysis (Gere and Timoshenko, 1990). 0.35 0.25 H 0.15 0.05 Figure 5.2: Typical torque-rotation response (see APPENDIX E for all torque-rotation curves). The torsion tests were characterized by the initial stiffness calculated between 20% and 80% of the peak load, the peak load, and the failure energy. 113 BIOLOGICAL AND MECHANICAL CHARACTERISTICS The site of failure was determined from the histological sections. Failure occurred either at the host bone interface, through the graft/composite or at the allograft-cement interface. Apparent interface shear strength ( x a p p ) , interface rotation to failure (ID), stiffness (k), failure energy (G) and the histomorphometric parameters (i.e. graft porosity, percentage fibrous tissue, cortical porosity and cortical thickness) were compared with a two-way analysis of variance (ANOVA) with the time as factorial and the material as paired factors. Student-Newman Keuls analysis was used for post hoc comparisons and the significance level was 95%. In a pilot study, eight male mature Sprague-Dawley rats (480-550g) were used to compare the modified model with published data and to investigate the effect of the created microenvironment on bone remodelling. The surgical procedure was similar to that described except that one bone chamber was tightened to the endosteal cortex which stopped intramedullary bleeding and the other was left at least 1mm off the endosteal cortex for increased access to bone marrow. Both chambers were filled with impacted bone graft only and closed with a stainless steel lid similar as described by Tagil and Aspenberg (Tagil and Aspenberg, 1998). No internal fixator was used for the fixation of the chambers. After 3, 6 and 12 weeks, the rats were euthanized and the tibiae with the bone chambers were processed for non-decalcified histological analysis. 114 BIOLOGICAL AND MECHANICAL CHARACTERISTICS 5.3 RESULTS In the pilot study, the bone chambers were lost in two rats postoperatively, leaving 2 rats each at 3, 6 and 12 weeks for histological analysis. After 3 weeks, different zones of healing and repair were observed: 1) a zone with remodelled viable bone less dense than the allograft bone surrounded by bone marrow, 2) an active remodelling front of woven bone, 3) a zone with necrotic allograft bone with or without fibrous tissue ingrowth and 4) on top of the chamber an inflammatory response was still present (Figure 5.3A and B). In the chamber with the access to the bone marrow, the tissue was identical but the bone remodelling front and the tissue ingrowth was more advanced compared with the other chamber. At 6 weeks and 12 weeks the histology was similar to the 3 week. The bone ingrowth front with woven bone and fibrous tissue ingrowth were further advanced in the chamber with the unlimited supply compared with the three week specimens and bone chambers with the microenvironment (Figure 5.3). The thickness of the woven bone front increased with time. Between 6 and 12 weeks and no active bone remodelling was visible. Remnants of necrotic graft with little fibrous tissue ingrowth were present in the chambers with the microenvironment (Figure 5.3D and E). 115 BIOLOGICAL AND MECHANICAL CHARACTERISTICS Limited access Fibrous T i ssue Necrotic allograft B o n e marrow V iab le Trabecular bone Figure 5.3: Histology of the pilot study group at 3, 6 and 12 weeks. Four different zones of healing and repair were observed: Zone 1: Remodelled viable bone less dense than the allograft bone surrounded by bone marrow. Zone 2: Active remodelling front of woven bone. Zone 3: Necrotic allograft bone with or without fibrous tissue ingrowth. Zone 4: Inflammatory response. In the main group which addressed the hypotheses of this study, two rats were lost due to intraoperative tibial fracture and one due to post-operative fracture. Therefore ten rats in each time group were available for histological and histomorphometric analysis including seven which were tested 116 BIOLOGICAL AND MECHANICAL CHARACTERISTICS mechanically. All other rats recovered well from surgery. There were no signs of limping three days postoperatively. At 3 weeks the histology was substantially different compared with the previously described group in that the zones observed in the pilot animals were not present and substantial changes were consistently found in the cortical bone. In 40% (i.e. 8 of 20) of the bone chambers little or no fibrous tissue ingrowth was found, independent of the content of the chamber. In six bone chambers a dense membrane, and in another four an acute inflammatory reaction presumably due to surgery, was found close to the interface between the bone chamber and the cortex. In eight bone chambers the entire allograft or spaces between composite and bone chamber were embedded in fibrous tissue with only sparse or no graft and endosteal remodelling (Figure 5.4). No direct bonding of the graft particles in the cement to the endosteal surface was observed. In some specimens the bone particles in the cement were attached to the endosteal surface by a dense membrane or the fibrous tissue. The periosteal surface below the bone chamber was actively remodelled which resulted in a significantly increased cortical porosity and cortical thickness at 3 and 6 weeks compared with the 0 week group (Figure 5.4; APPENDIX G, Tables 5 to 8). There were no differences between the allograft and the composite for any the measured parameters. 117 BIOLOGICAL AND MECHANICAL CHARACTERISTICS Allograft Cement- Allograft Composite ! 73 ? s: i •* $ 20 I 10 1 0.8 r i ? E 0.7 | 0 5 i 0.4 | I*, 0.2 1 0.1 •#2r Woven Necrotic bone v allograft Viable trabecular bone Fibrous Tissue" Bone marrow n Cancellisation of U A l l ° 9 r a f t P° r o s i tyendosteal cortex Cortical porosity Percentage fibrous tissue within graft | Cortical thickness I Standard deviation Figure 5.4: Histology and histo-morphometry for the pure allograft and cement-allograft composite at 0, 3 and 6 weeks. Graphs showing the mean allograft porosity, cortical porosity, percent fibrous tissue and cortical thickness. Allograft porosity in the pure allograft group was significantly lower at 6 weeks (p<0.001). The cortical porosity and cortical thickness increased significantly at 3 and 6 weeks (p<0.002). The percentage fibrous tissue was significantly higher in the pure allograft at 3 and 6 weeks compared with the composite (p<0.005). 1 1 8 BIOLOGICAL AND MECHANICAL CHARACTERISTICS At 6 weeks the periosteal remodelling and endosteal absorption resulted in the formation of new medullary canal and cancellisation of the endosteal cortex (Figure 5.4). Graft remodelling, medullary canal formation and a remodelling front with woven bone were observed in 7 out of the 10 bone chambers with the pure allograft. The pure allograft porosity decreased significantly from 75% at 0 weeks and 72% at 3 weeks to 59% at 6 weeks (APPENDIX G, Table 8). In general the specimens showed an initial linear torque-rotation response up to the failure torque which then stayed constant or further increased up to the maximal applied rotation (Figure 5.2). The apparent strength for the allograft and the composite increased significantly from 0 to 3 week (p< 0.03; APPENDIX G, Table 9). The allograft and composite strengths decreased at 6 weeks and were not significantly different from the 0 and 3 weeks groups (Figure 5.5). The graft-cement composite strength overall was significantly higher compared with the pure allograft (p=0.03) (Figure 5.5). The stiffness of the allograft and the composite did not change significantly with time. The allograft constructs were significantly stiffer compared with the composite (p=0.03; APPENDIX G, Table 10) (Figure 5.6). The rotation at failure was significantly different among the three time groups and was highest at 3 weeks (p=0.02; APPENDIX G, Table 12). There were no significant differences in toughness of the constructs at any time (p=0.053; APPENDIX G, Table 11). 1 1 9 BIOLOGICAL AND MECHANICAL CHARACTERISTICS 1.0 0.9 0.8 0.7 £ 0.6 £ | 0.5 I 0 4 55 0.3 0.2 0.1 0.0 pure allograft • cement-allograft composite 0 week 3 week 6 week Figure 5.6: Allograft and composite stiffness of the allograft and composite at 0, 3 and 6 weeks (mean, SD). The stiffness did not change with time. The pure allograft stiffness was significantly higher than the composite (p=0.03). 121 BIOLOGICAL AND MECHANICAL CHARACTERISTICS opure allograft • cement-allograft composite 0 week 3 week 6 week Figure 5.6: Allograft and composite stiffness of the allograft and composite at 0, 3 and 6 weeks (mean, SD). The stiffness did not change with time. The pure allograft stiffness was significantly higher than the composite (p=0.03). 121 BIOLOGICAL AND MECHANICAL CHARACTERISTICS 5.4 DISCUSSION The objectives of this study were to investigate the temporal changes of the allograft and cement-allograft composite-host bone interface after impaction allografting. A rat bone chamber model, which simulated damaged and impaired endosteal circulation, was developed to investigate host bone interface strength of the allograft and allograft-cement composite. As any other animal model, this study has some limitations. Compared with other species and humans, the rat's bone composition differs and the trabecular thickness and separation are smaller (Aerssens et al., 1998; van der Donk S. ef al., 2001). In addition the metabolic rate is inversely related to the size of the animal which suggests that the rate of repair would be higher compared with larger animals or humans (Van der Donk 2001). However, comparison of bone and tissue ingrowth in the same bone chamber model in rats and goats showed little differences which increases the validity of the rat model (van der Donk S. ef al., 2001). In contrast to humans, rats have little or no secondary haversian remodelling in their cortical bone and therefore the remodelling might be different (Frost 1992). Despite these limitations of the rat as a model, it is a commonly used animal for bone research (Aerssens ef al., 1998; Yuehuei H ef al., 1999). In addition, the effect of graft impaction and bone growth factors on impaction allograft incorporation have been evaluated with a similar rat bone chamber model (Tagil and Aspenberg, 1998; Tagil ef al., 2000). The allograft particle sizes and shapes used in the rat bone chamber were substantial smaller 122 BIOLOGICAL AND MECHANICAL CHARACTERISTICS than those used in the impaction allografting procedure in humans (CHAPTER II Figure 2.3). In a larger model with larger bone particles, the particle surface area to volume fraction decreases if the allograft porosity (average 72% in human (CHAPTER II), 75% in current study) is constant which may affect ostoclastic and osteoblastic activity (Fazzalari et al., 1989). However, the different stages of allograft incorporation and remodelling appear to be similar in rats and in humans although they occur at different times (Tagil and Aspenberg, 1998; Linder, 2000). The interpretation of the results from the pilot study is limited by the small number at each follow up period. However, the allograft incorporation and remodelling in the chamber with the unlimited bone marrow access was virtually identical as reported in other bone chamber models reported (Tagil, 2000; van der Donk S. et al., 2001) which validates the observations. In the chamber with the damaged endosteal circulation the allograft remodelling was restricted to 0.5 to 1mm which was also observed in human autopsies and biopsies (Figure 5.3) (Linder, 2000). Therefore, we felt that this additional study validated the new model where the bone chamber was tightened to the endosteal surface. In the full study, a constant compressive load of approximately 20% of the rat's bodyweight was applied to hold the allograft and composite samples in place. The resultant compressive stress of 0.3MPa was only 15% of the compressive host bone interface stress predicted by finite element analysis for primary THRs (Verdonschot and Huiskes, 1996; Chang et al., 1998). Although, a constant load does not simulate cyclic postoperative weight bearing in patients, 123 BIOLOGICAL AND MECHANICAL CHARACTERISTICS the applied stress appeared to delay graft incorporation and remodelling as compared with the pilot study. This is in contrast to other animal studies, which showed little or no effect of load stimulus on bone incorporation and remodelling (Lamerigts et al., 2000; Tagil, 2000; van der Donk S. et al., 2002). It has been shown that increased graft impaction delays allograft incorporation (Tagil and Aspenberg, 1998). However, there was no significant decrease in allograft porosity due to the compressive load applied from 0 week to 3 weeks which may have delayed incorporation. In contrast to the pilot study at three weeks, there was only little or no bone remodelling observed independent of the material. Extensive periosteal remodelling and endosteal absorption was observed which significantly increased cortical porosity and cortical thickness (Figure 5.4). Interestingly, no periosteal cortical remodelling was observed in any of the specimens in the additional group. At six weeks the remodelling of the cortex resulted in the formation of a new medullary canal and the cancellation of the endosteal cortex. Increased cortical porosity is known to be caused by a damaged endosteal circulation which results in a necrotic endosteal cortex. This phenomena has been observed as a result of intramedullary nailing and in goats treated with THRs and impaction allografting (Schreurs ef al., 1994; Hupel ef al., 2001). The endosteal cancellation which results in medullary canal widening was observed in primary cemented THR and was correlated with radiological loosening (Kobayashi ef al., 1996). Histological examination of retrieved 124 BIOLOGICAL AND MECHANICAL CHARACTERISTICS cemented THRs confirmed that radiolucent lines represent areas of cancellation and thinning of the endosteal cortex (Kwong et al., 1992). Although interpretations of radiographs are difficult, radiolucent lines were observed after impaction allografting and were correlated with stem subsidence (Leopold et al., 2000; Kligman et al., 2002). These results suggest that the endosteal circulation may be damaged and impaired after a revision with impaction allografting resulting in medullary widening and potential stem subsidence. The significantly decreased pure allograft porosity at 6 weeks was a result of extensive remodelling of the graft and the formation of a bone ingrowth front with woven bone. It appears that necrotic endosteal cortex needs to be absorbed first before graft remodelling was possible. In contrast to inorganic bone and hydroxyapatite particle embedded in cement, there was no sign of graft particle remodelling or the attachment of the bone particle to the porous endosteal cortex at that time (Dai et al., 1991; Oonishi et al., 1997). Since interfaces fail in shear or tension and shear stresses at the host bone-cement interface exceed tensile stresses, the constructs were tested in pure shear (Verdonschot and Huiskes, 1996; Chang et al., 1998). The shear strength of the composite was significantly higher at 3 weeks compared with the pure allograft but decreased at 6 weeks (Figure 5.5). The increased strength appeared to be due to soft tissue attachment rather than direct bonding of the bone particles with the endosteal surface which is supported by its low stiffness interface compared with the allograft. The composite failed in general at the host 125 BIOLOGICAL AND MECHANICAL CHARACTERISTICS bone interface which was expected, since the cement/allograft composite strength is likely to be substantial higher than the interface strength (Jofe et al., 1991). The failure strength of the composite at 0 weeks was 85% lower than the interface strength measured after impaction allografting (CHAPTER IV). This difference is presumably due to the difference in composite and endosteal surface preparation. During impaction allografting the remnants of cement are removed with a high speed burr, this creating a somewhat rough surface where the pressurized cement can interdigitate. In contrast, in the current study the endosteal surface was flattened with a mill bit creating a smooth surface and for technical reasons the cement was cured outside the chamber and therefore the cement did not interdigitate with the endosteal surface. Although, the composite interface strength was significantly higher compared with the pure allograft, one would expect these differences to be larger in humans. Similar to the composite, the allograft construct strength increased at 3 weeks and decreased at 6 weeks. In contrast to the composite, the allograft construct failed at the cement-al log raft interface. The strength of this interface has not been determined in a clinically relevant setup. The increase in strength was most likely due to fibrous tissue ingrowth since allograft remodelling did not reach the cement at any time. 126 BIOLOGICAL AND MECHANICAL CHARACTERISTICS 5 .5 REFERENCES Aerssens.J., Boonen.S., Lowet.G., and Dequeker.J. (1998) Interspecies differences in bone composition, density, and quality: potential implications for in vivo bone research. Endocrinology 139, 663-670. Bolder.S.B., Verdonschot.N., Schreurs.B.W., and Buma.P. (2002) Acetabular defect reconstruction with impacted morsellized bone grafts or TCP/HA particles. A study on the mechanical stability of cemented cups in an artificial acetabulum model. Biomaterials 23, 659-666. Boldt.J.G., Dilawari.P., Agarwal,S., and Drabu.K.J. (2001) Revision total hip arthroplasty using impaction bone grafting with cemented nonpolished stems and charnley cups. J.Arthroplasty 16, 943-952. Brewster.N.T., Gillespie,W.J., Howie.C.R., Madabhushi.S.P., Usmani.A.S., and Fairbairn.D.R. (1999) Mechanical considerations in impaction bone grafting. J.Bone Joint Surg.Br. 81, 118-124. Brodt.M.D., Swan.C.C, and Brown,T.D. (1998) Mechanical behavior of human morselized cancellous bone in triaxial compression testing. J.Orthop.Res. 16, 43-49. Chang,P.B., Mann.K.A., and Bartel,D.L. (1998) Cemented femoral stem performance. Effects of proximal bonding, geometry, and neck length. Clin.Orthop. 355, 57-69. Dai.K.R., Liu.Y.K., Park.J.B., Clark.C.R., Nishiyama.K., and Zheng.Z.K. (1991) Bone-particle-impregnated bone cement: an in vivo weight-bearing study. J.Biomed.Mater.Res. 25, 141-156. de Roeck.N.J. and Drabu.K.J. (2001) Impaction bone grafting using freeze-dried allograft in revision hip arthroplasty. J.Arthroplasty 16, 201-206. Elting.J.J., Mikhail.W.E., Zicat.B.A., Hubbell.J.C, Lane.L.E., and House.B. (1995) Preliminary report of impaction grafting for exchange femoral arthroplasty. Clin.Orthop. 319, 159-167. Fazzalari.N.L, Crisp.D.J., and Vernon-Roberts,B. (1989) Mathematical modelling of trabecular bone structure: the evaluation of analytical and quantified surface to volume relationships in the femoral head and iliac crest. J.Biomech. 22, 901-910. 127 BIOLOGICAL AND MECHANICAL CHARACTERISTICS Fetzer.G.B., Callaghan.J.J., Templeton.J.E., Goetz.D.D., Sullivan,P.M., and Johnston,R.C. (2001) Impaction allografting with cement for extensive femoral bone loss in revision hip surgery: a 4- to 8-year follow-up study. J.Arthroplasty 16, 195-202. Franzen.H., Toksvig-Larsen.S., Lidgren.L, and Onnerfalt.R. (1995) Early migration of femoral components revised with impacted cancellous allografts and cement. A preliminary report of five patients. J.Bone Joint Surg.Br. 77, 862-864. Gere.J.M. and Timoshenko.S.P. (1990) Mechanics of material. PWS Publishing Company, Boston, MA. Giesen.E.B., Lamerigts.N.M., Verdonschot.N., Buma.P., Schreurs.B.W., and Huiskes.R. (1999) Mechanical characteristics of impacted morsellised bone grafts used in revision of total hip arthroplasty. J.Bone Joint Surg.Br. 81, 1052-1057. Gorab.R.S., Covino.B.M., and Borden,LS. (1993) The rationale for cementless revision total hip replacement with contemporary technology. Orthop Clin.North Am. 24, 627-633. Griffon,D.J., Dunlop.D.G., Howie.C.R., Gilchrist,!"., Salter.D.M., and Healy.D.M. (2001a) Early dissolution of a morsellised impacted silicate-free bioactive glass in metaphyseal defects. J.Biomed.Mater.Res. 58, 638-644. Griffon,D.J., Dunlop.D.G., Howie.C.R., Pratt.J.N., Gilchrist.T.J., and Smith,N. (2001b) An ovine model to evaluate the biologic properties of impacted morselized bone graft substitutes. J.Biomed.Mater.Res. 56, 444-451. Hupel.T.M., Weinberg,J.A., Aksenov.SA, and Schemitsch.E.H. (2001) Effect of unreamed, limited reamed, and standard reamed intramedullary nailing on cortical bone porosity and new bone formation. J.Orthop.Trauma 15, 18-27. Jofe.M.H., Takeuchi.T., and Hayes.W.C. (1991) Compressive behavior of human bone-cement composites. J.Arthroplasty 6, 213-219. Kligman,M., Con.V., and Roffman.M. (2002) Cortical and cancellous morselized allograft in revision total hip replacement. Clin.Orthop. 139-148. 128 BIOLOGICAL AND MECHANICAL CHARACTERISTICS Knight, J.L. and Helming,C. (2000) Collarless polished tapered impaction grafting of the femur during revision total hip arthroplasty: pitfalls of the surgical technique and follow-up in 31 cases. J.Arthroplasty 15, 159-165. Kobayashi,S., Eftekhar.N.S., and Terayama.K. (1996) Long term bone remodeling around the Charnley femoral prostheses. Clin.Orthop. 162-173. Kwong.L.M., Jasty.M., Mulroy.R.D., Maloney.W.J., Bragdon.C, and Harris.W.H. (1992) The histology of the radiolucent line. J.Bone Joint Surg.Br. 74, 67-73. Lamerigts.N.M., Buma.P., Huiskes.R., Schreurs.W., Gardeniers.J., and Slooff.T.J. (2000) Incorporation of morsellized bone graft under controlled loading conditions. A new animal model in the goat. Biomaterials 21, 741-747. Leopold,S.S., Jacobs,J.J., and Rosenberg,A.G. (2000) Cancellous allograft in revision total hip arthroplasty. A clinical review. Clin.Orthop. 86-97. Leopold,S.S. and Rosenberg,A.G. (2000) Current status of impaction allografting for revision of a femoral component. Instr.Course.Lect 49, 111-118. Linder.L. (2000) Cancellous impaction grafting in the human femur: histological and radiographic observations in 6 autopsy femurs and 8 biopsies. Acta Orthop.Scand 71, 543-552. Ling.R.S., Timperley.A.J., and Linder.L. (1993) Histology of cancellous impaction grafting in the femur. A case report. J.Bone Joint Surg.Br. 75, 693-696. Nelissen,R.G., Bauer.T.W., Weidenhielm.L.R., LeGolvan.D.P., and Mikhail.W.E. (1995) Revision hip arthroplasty with the use of cement and impaction grafting. Histological analysis of four cases. J.Bone Joint Surg.Am. 77, 412-422. Oonishi.H., Iwaki.Y., Kin,N., Kushitani.S., Murata.N., Wakitani.S., and Imoto.K. (1997) Hydroxyapatite in revision of total hip replacements with massive acetabular defects: 4-to 10-year clinical results. J.Bone Joint Surg.Br. 79, 87-92. Oonishi.H., Kadoya.Y., Iwaki.H., and Kin.N. (2001) Total hip arthroplasty with a modified cementing technique using hydroxyapatite granules. J.Arthroplasty 16, 784-789. 129 BIOLOGICAL AND MECHANICAL CHARACTERISTICS Pekkarinen,J., Alho.A., Lepisto.J., Ylikoski.M., Ylinen.P., and Paavilainen.T. (2000) Impaction bone grafting in revision hip surgery. A high incidence of complications. J.Bone Joint Surg.Br. 82, 103-107. PrattJ.N., Griffon,D.J., Dunlop.D.G., Smith,N., and Howie.C.R. (2002) Impaction grafting with morsellised allograft and tricalcium phosphate-hydroxyapatite: incorporation within ovine metaphyseal bone defects. Biomaterials 23, 3309-3317. Schreurs.B.W., Buma.P., Huiskes.R., Slagter.J.L, and Slooff.T.J. (1994) Morsellized allografts for fixation of the hip prosthesis femoral component. A mechanical and histological study in the goat. Acta Orthop.Scand. 65, 267-275. Tagil,M. (2000) The morselized and impacted bone graft. Animal experiments on proteins, impaction and load. Acta Orthop.Scand.Suppl. 290:1-40, 1-40. Tagil,M. and Aspenberg.P. (1998) Impaction of cancellous bone grafts impairs osteoconduction in titanium chambers. Clin.Orthop. 352, 231-238. Tagil,M. and Aspenberg.P. (2001) Fibrous tissue armoring increases the mechanical strength of an impacted bone graft. Acta Orthop.Scand. 72, 78-82. Tagil,M., Jeppsson,C, and Aspenberg.P. (2000) Bone graft incorporation. Effects of osteogenic protein-1 and impaction. Clin.Orthop. 371, 240-245. Ullmark.G., Hallin.G., and Nilsson.O. (2002) Impacted corticocancellous allografts and cement for revision of the femur component in total hip arthroplasty. J.Arthroplasty 17, 140-149. Ullmark.G. and Obrant.K.J. (2002) Histology of impacted bone-graft incorporation. J.Arthroplasty 17, 150-157. van der Donk S., Buma.P., Aspenberg.P., and Schreurs.B.W. (2001) Similarity of bone ingrowth in rats and goats: a bone chamber study. Comp Med. 51, 336-340. van der Donk S., Buma.P., Verdonschot.N., and Schreurs.B.W. (2002) Effect of load on the early incorporation of impacted morsellized allografts. Biomaterials 23, 297-303. Verdonschot.N. and Huiskes.R. (1996) Mechanical effects of stem cement interface characteristics in total hip replacement. Clin.Orthop. 329, 326-336. 130 BIOLOGICAL AND MECHANICAL CHARACTERISTICS Verdonschot.N., van Hal.C.T., Schreurs.B.W., Buma.P., Huiskes.R., and Slooff.T.J. (2001) Time-dependent mechanical properties of HA/TCP particles in relation to morsellized bone grafts for use in impaction grafting. J.Biomed.Mater.Res. 58, 599-604. Voor.M.J., Nawab.A., Malkani.A.L, and Ullrich,CR. (2000) Mechanical properties of compacted morselized cancellous bone graft using one-dimensional consolidation testing. J.Biomech 33, 1683-1688. Yuehuei H, Friedman A, and Friedman R (1999) Animal selection in orthopaedic research. In Animal Models in Orthopaedic Research (Edited by Yuehuei H and Friedman R) Pp. 39-50. CRC Press, Boca Raton. 131 CHAPTER VI GENERAL DISCUSSION AND CONCLUSIONS 132 GENERAL DISCUSSION AND CONCLUSION 6.1 DISCUSSION Total hip replacements provide pain relief and an improved quality of life for millions of patients worldwide. THRs may not last for the lifetime of a patient, in particular if they are younger and more active. The revision of the failed components is challenging particularly when associated with extensive osteolytic bone loss. There are several surgical treatment options for failed THR, including impaction allografting. Osteolytic defects filled with impacted morsellized allograft bone provide sufficient strength postoperatively to rigidly accommodate the revision components and provide a scaffold for the restoration of the host bone stock. Promising short and intermediate term clinical results have been reported for this technique. However, excessive stem subsidence and incomplete remodelling are problems associated with the impacted allograft technique, but are not well understood. Much research has focussed on characterising and optimizing allograft layer properties and understanding and optimizing the remodelling process. However, despite the host bone interface being critically important for the success of THR in general, its characteristics have not been addressed for the impaction allografting technique. This thesis focussed on the host bone interface after impaction allografting, addressing its morphology and the relevant surgical factors, its postoperative strength and the temporal changes in its strength and morphology. It provides important baseline information which will help engineers, 133 GENERAL DISCUSSION AND CONCLUSION scientists and clinicians in their research to ensure clinically stable implants that sufficiently restore host bone stock after revision of failed THR. 6.1.1 Morphology The morphology of the stem-femur-graft complex was found to vary substantially along the femur and these patterns have not been described previously. The porosity of the impacted allograft was highest proximally and lowest distally. These differences were mainly due to the different shape of impactors used during the procedure. The distal impactors were more efficient in allograft impaction. The lowest porosity achieved with the proximal impactor was 5 7 % with an average impaction force of 1530N. With the distal impactors, only 888N was necessary to achieve the same graft impaction (Figure 2.5). Therefore, a constant dense impacted graft layer may be achieved by using impactor shapes similar to the distal impactors to fill the entire femur. This may also reduce the risk of intraoperative fractures since lower impaction forces would be required and radial forces would be substantially reduced due to the shape of the impactor. However, more allograft would be necessary and the neo-medullary canal would need to be drilled and reamed (Figure 6.1). Parts of this approach are already used by surgeons as observed during the impaction allografting procedure on cadaveric specimens (CHAPTER II). If the distal impactors are used beyond the distal impaction and the tight impaction of the allograft prevents 134 GENERAL DISCUSSION AND CONCLUSION positioning of the neo-medullary canal, the impacted graft is partially removed with a circular reamer. A B C D E F G Figure 6.1: Proposed modification of the impaction allografting technique. A: Distal plug with guide wire. B to D: Allograft impaction with flattened impactors similar in shape like the distal impactors. E: Drilling of the neo-medullary canal with a tapered reamer. F: Reaming of the canal. G: Femur-allograft construct ready for cement and stem insertion. The cement mantle exceeded 2mm in some locations and was absent in others. In primary THRs, thin cement or an incomplete cement mantle has been correlated with irregular stress distribution and cement mantle fracture (Valdivia ef al., 2001). The significance of an incomplete cement mantle for the impaction allografting technique is not understood. However, it has been suggested that a 135 GENERAL DISCUSSION AND CONCLUSION cement mantle less than 2mm is considered insufficient or incomplete (Masterson et al., 1997). The cement penetration into the allograft produces a substantial cement-allograft composite (average 5.9mm Figure 2.6). Therefore, a thicker, consistent cement mantle may not be necessary. However, bone particles in the cement give rise to stress concentrations which may cause cement mantle fracture and be related to stem subsidence (Masterson et al., 1997). The cement penetration into the allograft bone during pressurisation and stem insertion was higher than expected. In the middle third of the stem almost the entire cross-section, including the simulated defects, were filled with the allograft-cement composite. The presence of the cement at the endosteum may prevent cortical remodelling. Proximal and distal to the middle third, the cement reached the endosteal surface at some but not all circumferential locations, forming a cement/allograft composite-host bone interface in addition to the allograft-host bone interface. Despite the mechanical advantages, which are discussed later, cement penetration to the endosteal surface does not allow restoration of the host bone stock. 6.1.2 Factors that determine morphology The modest correlation between cement penetration and graft porosity (r=0.55) suggested that other factors such as cement pressure and cement viscosity determine cement penetration into the impacted allograft bone. It was 136 GENERAL DISCUSSION AND CONCLUSION shown by the finite element analysis that the allograft porosity profile along the stem determines the intrusion profile and the cement volume injected determines the depth of the cement penetration. The porosity profile can be controlled intraoperatively by the impaction force and the cement volume by cement pressure, viscosity and duration of pressurisation. Cement pressure is dependent on the cement viscosity and the pressure magnitude required to displace residual fluids in the trabecular network and the quality of the proximal seal. Cement penetration up to the endosteal surface can be avoided by limiting the volume injected into the femur during pressurisation. For that purpose the cement loss at the proximal seal needs to be monitored which somewhat complicates an already time intensive and difficult surgical procedure. Stem insertion after retrograde filling of the neo-medullary canal without cement pressurisation may also limit cement penetration. The high pressures during stem insertion would force the cement into the allograft bone (McCaskie et al., 1997; Churchill et al., 2001). Alternatively, after pressurisation, some of the cement in the medullary canal could be removed which would limit cement penetration. 6.1.3 Effect of morphology on mechanical characteristics From a mechanical perspective stem subsidence occurs through failure of any material or interface between the stem and the host bone. In order to determine the likely site of failure, the stresses in the materials and at the 137 GENERAL DISCUSSION AND CONCLUSION interfaces need to be known. This data is not available for the impaction allografting construct. Although the stress field may be different after impaction allografting due to a substantially different morphology and associated material and interface properties, the complexity is illustrated by a finite element analysis of a primary THR (Figure 6.2; Weinans et al., 1990). Considering the variation in stresses, one would expect that a material or interface fails locally whereas in other locations no failure occurs. Local material failure or interface debonding due to excessive stress or fatigue do not necessary lead to stem subsidence, although they may trigger a cascade of failures which ultimately results in stem subsidence. The tapered geometry of the stem and the medullary canal may help to find a new equilibrium. H I Cortex • Cancellous bone Cement mantle Radial stress distribution along the femur Axial stress distribution along the femur and endosteal surface Figure 6.2: Radial and axial stress distribution of a primary THR with tensile separation (i.e. no tensile loads are transferred) at the cement cancellous bone interface determined by a two dimensional finite element model. Note: Fully bonded stem and different scales for radial and axial stresses. (Adapted from Weinans et al., 1990). 138 GENERAL DISCUSSION AND CONCLUSION Keeping the complexity of the problem in mind, some simplistic calculation can be made to help with the interpretation of construct morphology and the mechanical characteristics of the host bone interface in terms of stem subsidence. To investigate the effect of cement penetration on stem subsidence the cement mantle/ cement allograft composite around the middle third of the stem was simplified as a thick-walled tube. The allograft was confined between the cement mantle/cement-al log raft composite and a rigid cortical shell (Figure 6.3). The inner and outer surface of the cement/composite tube were subjected to uniaxial compressive stresses determine by finite element analysis of the stem cement (Norman et al., 2001) and cement-host bone interface (Chang et al., 1998). The radial expansion of the cement/composite tube and the resulting radial compressive stresses on the allograft and the allograft-host bone interface stresses were calculated using conventional theory of elasticity (Timoshenko and Goodier, 1987). It was assumed that no tensile stresses could be transferred across the interfaces due to a polished stem and the low tensile strength of the impacted allograft-host bone interface (Brewster et al., 1999; CHAPTER IV). The shear stress in the allograft and at the interface were based on finite element prediction for primary THRs (Chang et al., 1998). The Young's modulus and strength of the cortex and confined compressive modulus for the allograft were obtained from the literature (Mow and Hayes, 1997; Giesen et al., 1999) (Figure 6.3). 139 GENERAL DISCUSSION AND CONCLUSION Figure 6.3: Simplified model of the stem-cement-allograft-bone complex around the middle third of the stem to investigate the effect of different cement/ composite layer thicknesses on stem subsidence and allograft and host bone interface failure. It was assumed that the cement/ composite is a thick-walled tube exposed to internal and external stresses (Timoshenko and Goodier, 1987). The internal 20MPa and external stress 3.3MPa were obtained from compressive stress predictions by finite element analysis for the stem-cement and the cement-bone interface of primary THR. (Chang ef al., 1998; Norman et al., 2001). The cement/ composite thickness were modelled from 0.5 to 6.75mm with corresponding allograft layer thicknesses of 6.5 to 0.25mm (CHAPTER II). Expansion of the inner cement mantle/composite diameter results in stem subsidence and depends on the cement mantle/composite tube thickness (Figure 6.4). The associated expansion of the outer cement diameter results in compression of the allograft layer (Figure 6.5). With increasing cement thickness 140 GENERAL DISCUSSION AND CONCLUSION the radial expansion decreases resulting in decreased stem subsidence and allograft compaction (Figure 6.4 and 6.5). The allograft radial compressive stress is highest with thin cement mantles and thin allograft layers (<1mm) (Figure 6.5). Cement thicknesses greater than 1mm prevent substantial radial compressive stress being transferred to the allograft. Normal compressive stresses on the allograft layer have been shown to increase its shear strength and are therefore important to prevent shear failure of the allograft layer and possible stem subsidence (Brewster et al., 1999). Although in thicker cement mantles/ composites, the radial expansion of the cement decreases substantially, high stresses are developed in the thin allograft layer (Figure 6.5). Allograft layer thickness [mm] 7 6 5 4 3 2 1 0 g . ! 0 1 2 3 4 5 6 7 Cement mantle thickness [mm] Figure 6.4: Predicted stem subsidence as a function of cement mantle/ composite thickness. With increasing cement mantle thickness, expansion of the inner diameter decreases resulting in less stem subsidence. With the entire section filled, the stem does not subside more than 1.3mm. 141 GENERAL DISCUSSION AND CONCLUSION Allograft layer thickness [mm] 7 6 5 4 3 2 1 0 Cement mantle Thickness [mm] Figure 6.5: Uniaxial compressive stress in the allograft and the host bone interface as a function of cement mantle/composite thickness. The patterned areas are required compressive stresses in the allograft or host bone interface to sustain a shear stress of 2.2MPa. The failure criteria are based on Brewster et al. and on the interface strength determined in CHAPTER IV (Brewster ef al., 1999). The stress is increasing with decreasing allograft layer thickness. Cement mantle of 2mm which is considered as intact would result in shear failure of the host bone interface and the allograft layer (Masterson ef al., 1997). No failure would occur with graft layer equal and smaller than 0.3mm. In the model, for a 2mm cement mantle considered as intact, 2.6mm stem subsidence would result with tensile failure of the mantle (clement 5* 36.7MPa; Lewis, 1997), which may further increase stem subsidence (Figure 6.4) (Masterson et al., 1997). Shear failure of the host bone interface and the allograft 142 GENERAL DISCUSSION AND CONCLUSION and presumably stem subsidence occurs due to the low radial compressive stresses in the allograft layer with a 2mm cement (Figure 6.5). In an aluminium tube model, substantial stem subsidence was observed with no cement mantle which validates this simplistic model (Robinson, 2003). The rigid distal plug used in this experimental model did not stop stem subsidence. If the stem does not subside within the cement mantle due to cement stem bonding or a rough stem surface, the stem-cement construct may subside through consolidation of the allograft layer indicated by its low elastic modulus (Figure 6.3). Based on this simplistic model, cement penetration to the endosteal cortex that leaves a thin or non-existent allograft layer prevents stem subsidence into the cement mantle and enhances the shear strength of the allograft and the host bone interface immediately postoperatively (Figure 6.4 and 6.5). If the cement reaches the endosteal cortex, additional resistance to subsidence is increased by locking of the composite into the tapered femur (average taper angle 3.4° SD 2.1°) (Figure 6.6). Hoop stresses in the cortex caused by the composite were below the transverse tensile strength of cortical bone (51 MPa; Mow and Hayes, 1997), for cortical thicknesses greater than 1mm. This suggests that immediately postoperatively, substantial subsidence may not occur with cement infiltration up to the endosteal cortex. This is supported by structural cadaveric experiments which measured stem subsidence of less than 2mm (Berzins et al., 1996; Malkani etal., 1996; Hbstner et al., 2001; Kligman etal., 2003). 143 GENERAL DISCUSSION AND CONCLUSION Figure 6.6: Locking of the cement-allograft in the tapered medullary may prevent stem-composite subsidence. Structural cadaveric experiments are critical to verify these hypotheses, due to the simplified geometry and stress field. However, it appears that cement penetration to the endosteal surface in the middle third of the stem is important for the immediate postoperative clinical stability of the stem-cement-allograft-bone construct after impaction allografting. 6.1.4 Temporal changes of morphology and strength Cadaveric experiments and clinical RSA studies have shown that the risk of stem subsidence increases from being initially clinically stable to subsiding stems within the first postoperative year (Berzins et al., 1996; Malkani et al., 144 GENERAL DISCUSSION AND CONCLUSION 1996; Karrholm era/., 1999; Hostner et al., 2001; Nelissen etal., 2002; Kligman et al., 2003). Fatigue failure of the materials and interfaces and/or their biological changes may be responsible for these temporal changes of the stem-cement allograft-bone construct. The biological changes were investigated with a rat bone chamber model which simulated the damaged endosteal circulation in a microenvironment after impaction allografting. Based on previous work with inorganic bone and hydroxyapatite particles embedded in cement, we hypothesized that the cement-allograft composite interface strength would further increase by bonding of the graft particles to the endosteal surface. The shear strengths of the composite-host bone interface and the allograft construct did increase significantly at 3 weeks postoperatively. However, histological examination showed that the increased interface strength was due to bonding of the embedded graft particles by fibrous tissue. In contrast to the composite, the allograft construct failed at the allograft-cement interface rather than at the host bone interface. There is no study available which characterizes allograft-cement interface, but it appears to be the weakest link before and after allograft incorporation. At all times, the composite interface strength was higher than the allograft construct. After six weeks the interface strengths of the composite decreased, most likely due to the changes of endosteal surface which are discussed next. Extensive periosteal remodelling and endosteal absorption was observed at 3 weeks, leading to cortical porosis and cancellation of the cortex by 6 145 GENERAL DISCUSSION AND CONCLUSION weeks. Cortical porosis is a well known problem in intramedullary nailing and is caused by damage of the endosteal circulation due to reaming (Hupel et al., 2001). Cortical porosis and cancellation has also been observed in goats treated with impaction allografting and resulted in medullary canal widening after cemented and non-cemented primary THR which has been associated with loosening and stem subsidence (Hofmann et al., 1989; Jasty et al., 1990; Schreurs et al., 1994; Robinson et al., 1994). Decreased cortical blood perfusion in the femur after primary THRs and in animal models suggests that, similar to intramedulary nailing, cortical porosis and subsequent medullary canal widening is caused by damaged endosteal circulation (Gruen et al., 1979; Rhinelander et al., 1979; Sund and Rosenquist, 1983; de Waal etal., 1988; Hupel etal., 2000) rather than stress shielding (Finkelstein et al., 1995). This is supported by the findings of the rat bone chamber experiment (CHAPTER V). Devascularisation of the inner one-third of the proximal femoral diaphysis was found in both cemented and non-cemented THRs in goats (de Waal et al., 1988). However, if cement was used, revascularisation was substantially delayed (Rhinelander era/., 1979; Sund and Rosenquist, 1983; de Waal et al., 1988). In the rat bone chamber, no difference in cortical porosis between the pure allograft and the composite was found, which was presumably due to the sealing effect of the bone chamber and needs further investigation. Surprisingly, no periosteal remodelling was found in any of the unloaded animals in the pilot study, which suggests that a "non-physiological" load triggers 146 GENERAL DISCUSSION AND CONCLUSION these processes. The constant load applied was only 15% of the rat's body weight which is only 10% of the compressive host bone interface stress reported for primary THRs (Chang et al., 1998). Since a constant load does not simulate the cyclic postoperative weight bearing in patients, an in vivo actuator based on shape memory alloys is in development. The actuator can be implanted into the rat and is activated through mutual inductance. A functional prototype was built, but it was beyond the scope of this thesis to finalize and test this actuator. Stem subsidence reported in clinical studies of primary and revised THRs with impaction allografting could not be reproduced by in vitro experimentation or finite element analysis (Weinans et al., 1990; Berzins et al., 1996; Malkani et al., 1996; Chang et al., 1998; Hostner et al., 2001; Kligman et al., 2003). This highlights the significance of these biological observations. It appears that for primary THRs, medullary canal widening is a known phenomenon but has not been recently associated with femoral component subsidence or loosening. After impaction allografting, stems subside in the first postoperative year and then stabilize, presumably after revascularisation. However, necrotic parts were still present in the cortex after impaction allografting more than 2 years postoperatively (Linder, 2000). Therefore a damaged endosteal circulation resulting in medullary widening is a likely cause of clinical unstable stems (Figure 6.7). In addition, it remains to be determined how much of the viable trabecular bone, observed in autopsies and biopsies, is remodelled impacted allograft or cancellised endosteal cortex. 147 GENERAL DISCUSSION AND CONCLUSION F i g u r e 6.7: Schematic drawing illustrating medullary canal widening as observed after primary THR. This phenomenon presumably is caused by a damaged endosteal circulation and may cause stem-cement-allograft construct subsidence after impaction allografting. From a mechanical perspective, the morphology, the interface characteristics and the increased interface strength would favour cement penetration to the endosteal surface. However, the cement is not removed biologically and this does not allow reconstruction of the host bone stock. In addition, cement penetration may delay endosteal revascularisation and results in more endosteal absorption compared with a pure allograft interface where revascularisation is facilitated by the porous structure. 148 GENERAL DISCUSSION AND CONCLUSION 6.1.5 Clinical significance With its potential to restore host bone stock, the impaction allografting procedure is a promising technique for the revision of failed THR associated with severe bone loss. The technique is time intensive, challenging and special training is required, particularly in severe cases (Ling, 1997). Therefore, any attempt to improve the clinical outcome of the impaction allografting technique should also address the surgical technique. With the modification of technique described above (Figure 6.1), higher allograft impaction can be achieved proximally which may reduce cement penetration and decrease the risk of femoral fracture. The circular reamer and stem shaped reamer to create the neo-medullary canal may be available from toolsets used for primary THRs. Cement penetration up to the endosteal cortex can be avoided by limiting the cement volume injected during pressurisation. The cement volume injected is also decreased by decreasing the cement pressure during pressurisation, limiting time of pressurisation, increased cement viscosity by late pressurisation and stem insertion which can be varied intraoperatively. It appears that cement penetration up to the endosteal cortex is important for the postoperative clinical stability of the stem-cement-allograft construct. Therefore, cadaveric experiments have to show that immediately postoperative, clinical stability can be achieved with limited cement penetration before any of the above mentioned modifications to the technique are made. 1 4 9 GENERAL DISCUSSION AND CONCLUSION The average cement mantle observed herein was 1.7mm which would be considered as insufficient (Masterson et al., 1997). This suggests the use of larger proximal impactors. The cement penetration into the allograft was on average 5.9mm, thereby forming a cement allograft composite with lower compressive strength than pure cement (Jofe et al., 1991). Therefore a thicker pure cement mantle appears to be beneficial for the strength of the construct. The host bone interface shear strength was similar to cemented revision THR in areas were the cement reached the endosteal surface. This suggests that postoperative weight bearing regime after impaction allografting could be similar to that after cemented revisions. The host bone interface strength could be increased by roughening of the endosteal surface. Medullary widening is a potential problem after impaction allografting and may lead to clinically unstable stems. Although it might be not possible to avoid damage to the endovascular circulation during the revision procedure, revascularisation should be facilitated were possible. Distal revascularisation can be enhanced by a preamble distal plug to create access to the medullary canal distally. 150 GENERAL DISCUSSION AND CONCLUSION 6 .2 CONCLUSIONS 1. The morphology of the stem-cement-allograft bone construct after impaction allografting varied along the femur and was mainly determined by the allograft impaction profile along the femur and the amount of cement injected during cement pressurisation. 2. In the middle third of the stem-femur complex, virtually the entire cross-section was penetrated with bone cement, thereby forming an allograft-cement composite. The cement reached the endosteal bone interface also in other locations along the stem, although it was not present at all circumferential positions. 3. The interface strengths varied along the femur and were correlated with the amount of cement contact with the host bone. The allograft-cement composite interface strength was higher compared with a pure allograft-host bone interface. The measured interface shear strengths were similar to those reported for cemented revision proximally and somewhat lower distally. 4. The shear strength increased after 3 weeks in a rat bone chamber model for the pure allograft and composite construct, mainly due to graft incorporation and soft tissue attachment of the graft particles to the endosteal surface. 5. In contrast to host bone interface failure of the composite, the allograft construct failed at the cement-allograft interface and was weaker than that 151 GENERAL DISCUSSION AND CONCLUSION of the composite. At 6 weeks the interface strength of the composite decreased presumably due to endosteal cancellisation. 6. The damaged endosteal circulation resulted in cortical cancellisation and endosteal absorption. Subsequent medullary canal widening correlated with radiological loosening in primary THRs and may also cause stem subsidence in revision THR with impaction allografting. 6.3 CONTRIBUTIONS The following novel contributions to the field were made: 1. The morphology of the stem-cement-allograft-bone construct determined in this thesis is important baseline information for further research on the understanding and improvement of the impaction allografting technique. In particular, the observation of the degree of cement penetration is novel and potentially important. 2. The finite element model showed that cement penetration into impacted allograft bone can be accurately predicted with the specific finite element formulation. This formulation will also help to predict cement flow in other areas of orthopaedic research. The cement penetration depth can be controlled by the volume whereas the direction of flow is governed by the porosity. 3. The mechanical characterisation of the host bone interface for the impaction allografting technique has not been previously determined. This 152 GENERAL DISCUSSION AND CONCLUSION information is important for mathematical models and will help in the development and preoperative validation of alternative techniques or new materials for revision hip reconstruction. The bone chamber model with a microenvironment that resembles the revision hip scenario is a novel development. Medullary canal widening due to a damaged endosteal circulation was identified as a problem associated with impaction allografting and other THRs. This bone chamber model will help to develop and validate strategies and new materials for the revision of failed THR. FUTURE WORK In areas of excessive cement penetration, restoration of the host bone is not possible. However, before the impaction allografting technique is modified, the effect of a reduced cement penetration on the structural behaviour of the stem-cement-allograft construct needs to be determined in cadaveric experiments. Stem-cement, cement-allograft and allograft-bone interface stresses and the stresses in the cement mantle, cement-allograft composite and the allograft are not known after impaction allografting. These stresses are important in order to understand the failure mechanisms and can be determined by finite element analysis of the stem-cement-allograft construct. However, important parameters for the finite element model 153 GENERAL DISCUSSION AND CONCLUSION including the mechanical properties of the cement-allograft composite and the cement-allograft interface need to be determined experimentally. The long term effect (after allograft remodelling stops) of the microenvironment on strength of the pure allograft and the cement-allograft composite is not known and needs to be investigated. The prototype of the in vivo actuator needs further development and testing and can be used to address the effect of dynamic loads and different load magnitudes on graft remodelling and cortical porosis and cancellisation. To what extent the cement/allograft composite-host bone delays endosteal revascularisation is not clear. The composite needs to be compared with a pure allograft interface in a modified bone chamber model with increased access to the medullary canal. Strategies to enhance allograft remodelling and endosteal revascularisation in a hypo-vascularised environment need to be developed and tested. 154 GENERAL DISCUSSION AND CONCLUSION 6.5 REFERENCES Berzins.A., Sumner.D.R., Wasielewski.R.O, and Galante.J.O. (1996) Impacted particulate allograft for femoral revision total hip arthroplasty. In vitro mechanical stability and effects of cement pressurization. J.Arthroplasty 11, 500-506. Brewster.N.T., Gillespie, W.J., Howie.C.R., Madabhushi.S.P., Usmani.A.S., and Fairbairn.D.R. (1999) Mechanical considerations in impaction bone grafting. J.Bone Joint Surg.Br. 81, 118-124. Chang,P.B., Mann.K.A., and Bartel.D.L. (1998) Cemented femoral stem performance. Effects of proximal bonding, geometry, and neck length. Clin.Orthop. 355, 57-69. Churchill,D.L., Incavo.S.J., Uroskie.J.A., and Beynnon.B.D. (2001) Femoral stem insertion generates high bone cement pressurization. Clin.Orthop. 393, 335-344. de Waal.M.J., Slooff.T.J., Huiskes.R., de Laat.E.A., and Barentsz.J.O. (1988) Vascular changes following hip arthroplasty. The femur in goats studied with and without cementation. Acta Orthop Scand. 59, 643-649. Finkelstein.J.A., Anderson,G.I., Waddell.J.P., Richards.R.R., and Humeniuk.B. (1995) A Madreporic-surfaced femoral component in a canine total hip arthroplasty model: bone remodelling response at 6 and 24 months. Can.J.Surg. 38, 501-506. Giesen.E.B., Lamerigts.N.M., Verdonschot.N., Buma.P., Schreurs.B.W., and Huiskes.R. (1999) Mechanical characteristics of impacted morsellised bone grafts used in revision of total hip arthroplasty. J.Bone Joint Surg.Br. 81, 1052-1057. Gruen,T.A., McNeice.G.M., and Amstutz.H.C. (1979) "Modes of failure" of cemented stem-type femoral components: a radiographic analysis of loosening. Clin.Orthop. 141, 17-27. Hofmann.AA, Wyatt.R.W., France.E.P., Bigler.G.T., Daniels.A.U., and Hess.W.E. (1989) Endosteal bone loss after total hip arthroplasty. Clin.Orthop. 245, 138-144. Hostner.J., Hultmark.P., Karrholm.J., Malchau.H., and Tveit.M. (2001) Impaction technique and graft treatment in revisions of the femoral component: laboratory studies and clinical validation. J.Arthroplasty 16, 76-82. 155 GENERAL DISCUSSION AND CONCLUSION Hupel.T.M., Schemitsch.E.H., Aksenov.SA., and Waddell.J.P. (2000) Blood flow changes to the proximal femur during total hip arthroplasty. Can.J.Surg. 43, 359-364. Hupel.T.M., Weinberg,J.A., Aksenov.SA, and Schemitsch.E.H. (2001) Effect of unreamed, limited reamed, and standard reamed intramedullary nailing on cortical bone porosity and new bone formation. J.Orthop.Trauma 15, 18-27. Jasty.M., Maloney.W.J., Bragdon.C.R., Haire.T., and Harris,W.H. (1990) Histomorphological studies of the long-term skeletal responses to well fixed cemented femoral components. J.Bone Joint Surg.Am. 72, 1220-1229. Jofe.M.H., Takeuchi.T., and Hayes,W.C. (1991) Compressive behavior of human bone-cement composites. J.Arthroplasty 6, 213-219. Karrholm.J., Hultmark.P., Carlsson.L, and Malchau.H. (1999) Subsidence of a non-polished stem in revisions of the hip using impaction allograft. Evaluation with radiostereometry and dual-energy X-ray absorptiometry. J.Bone Joint Surg.Br. 81, 135-142. Kligman,M., Rotem.A., and Roffman.M. (2003) Cancellous and cortical morselized allograft in revision total hip replacement: A biomechanical study of implant stability. J.Biomech. 36, 797-802. Lewis,G. (1997) Properties of acrylic bone cement: state of the art review. J.Biomed.Mater.Res 38, 155-182. Linder.L. (2000) Cancellous impaction grafting in the human femur: histological and radiographic observations in 6 autopsy femurs and 8 biopsies. Acta Orthop.Scand 71, 543-552. Ling.R.S. (1997) Femoral component revision using impacted morsellised cancellous graft. J.Bone Joint Surg.Br. 79, 874-875. Malkani.AL, Voor.M.J., Fee.K.A., and Bates,CS. (1996) Femoral component revision using impacted morsellised cancellous graft. A biomechanical study of implant stability. J.Bone Joint Surg.Br. 78, 973-978. Masterson,E.L., Masri,B.A., and Duncan,CP. (1997) The cement mantle in the Exeter impaction allografting technique. A cause for concern. J.Arthroplasty 12, 759-764. 156 GENERAL DISCUSSION AND CONCLUSION McCaskie,A.W., Barnes,M.R., Lin.E., Harper,W.M., and Gregg,P.J. (1997) Cement pressurisation during hip replacement. J.Bone Joint Surg.Br. 79, 379-384. Mow.V.C. and Hayes.W.C. (1997) Basic Orthopaedic Biomechanics. Lippincott-Raven Publishers, New York. Nelissen.R.G., Valstar.E.R., Poll.R.G., Garling.E.H., and Brand,R. (2002) Factors associated with excessive migration in bone impaction hip revision surgery: a radiostereometric analysis study. J.Arthroplasty 17, 826-833. Norman,T.L., Thyagarajan.G., Saligrama.V.C, Gruen,T.A., and Blaha.J.D. (2001) Stem surface roughness alters creep induced subsidence and 'taper-lock' in a cemented femoral hip prosthesis. J.Biomech. 34, 1325-1333. Rhinelander.F.W., Nelson,C.L., Stewart.R.D., and Stewart.C.L. (1979) Experimental reaming of the proximal femur and acrylic cement implantation: vascular and histologic effects. Clin.Orthop. 141, 74-89. Robinson,D., Hendel.D., and Halperin.N. (1994) Changes in femur dimensions in asymptomatic non-cemented hip arthroplasties. 20 cases followed for 5-8 years. Acta Orthop Scand. 65, 415-417. Robinson,M.C. (2003) An in vitro biomechanical study of impaction allografting for revision total hip arthroplasty. Master of Applied Science Thesis, University of British Columbia. Schreurs.B.W., Buma.P., Huiskes.R., Slagter.J.L, and Slooff.T.J. (1994) Morsellized allografts for fixation of the hip prosthesis femoral component. A mechanical and histological study in the goat. Acta Orthop.Scand. 65, 267-275. Sund.G. and Rosenquist, J. (1983) Morphological changes in bone following intramedullary implantation of methyl methacrylate. Effects of medullary occlusion: a morphometrical study. Acta Orthop. Scand. 54, 148-156. Timoshenko.S.P. and Goodier.J.N. (1987) Two-Dimensional Problems in Polar Coordinates. In Theory of Elasticity Pp. 65-149. McGraw-Hill, New York. 157 GENERAL DISCUSSION AND CONCLUSION Valdivia.G.G., Dunbar.M.J., Parker.D.A., Woolfrey.M.R., MacDonald.S.J., McCalden.R.W, Rorabeck.C.H., and Bourne,R.B. (2001) The John Charnley Award: Three-dimensional analysis of the cement mantle in total hip arthroplasty. Clin.Orthop. 393 38-51. Weinans,H., Huiskes.R., and Grootenboer.H.J. (1990) Trends of mechanical consequences and modeling of a fibrous membrane around femoral hip prostheses. J.Biomech. 23, 991-1000. 158 APPENDIX A HISTOLOGICAL PROCESSING PROTOCOLS AND STAINS 159 HISTOLOGICAL PROCESSING PROTOCOLS AND STAINS Tissue Processing Protocol for Cadaveric Bone with PMMA Method without dissolving PMMA Stage Time [days] After use 1 10% Buffered Formalin 2 reuse 2 Dehydrate in 70% Ethanol/ 30% distilled water 2 dispose 3 Dehydrate in 95% Ethanol/ 5% distilled water 2 dispose 4 Dehydrate in 100% Ethanol 2 dispose 5 Dehydrate in 100% Ethanol 2 dispose 6 Remove Ethanol in Oven @ 60 degrees 0.25 7 Embed in 100% Technovit 72001 ) in incubator 0.5 reuse 8 Polymerize under UV-Light • 0.5 Tissue Processing Protocol for the Bone Chamber Method with dissolving PMMA Stage Time [days] After use 1 10% Buffered Formalin 2 reuse 2 100% Acetone to dissolve PMMA 2 dispose 3 Dehydrate in 70% Ethanol/ 30% distilled water 2 4 Dehydrate in 95% Ethanol/ 5% distilled water 2 dispose 5 Dehydrate in 100% Ethanol 2 dispose 6 Dehydrate in 100% Ethanol 2 dispose 7 Infiltrate 50% Technovit 72001'/ 50% Ethanol 2 reuse 8 Infiltrate 75% Technovit 72001'/ 25% Ethanol 2 reuse 9 Infiltrate 100% Technovit 72001> 2 reuse 10 Infiltrate 100% Technovit 72001 ) 2 reuse 11 Embed in 100% Technovit 72001> in incubator 0.25 12 Polymerize under UV-Light 0.5 Technovit 7200, Kulzer Ltd, Wehrheim, Germany 160 HISTOLOGICAL PROCESSING PROTOCOLS AND STAINS GOLDNER TRICHROME STAIN FOR BONE Solutions: Weigert's Hematoxylin (mix Solution A and B 1:1) Solution A 95% Alcohol 1% Hematoxylin Solution B 20ml Ferric Chloride (29%) 475ml Distilled water 5ml Concentrated HCI Light Green 1.0g Light Green 100ml Acetic Acid 0.2% Method for slides ground to approx. 50um Stage Time [minutes] After use 1 Rinse in water 2 2 Stain in Weigert's Hematoxylin 20 dispose 3 Wash well in water 5 4 Stain in Ponceau Mixture 20 reuse 5 Rince in Acetic Acid 0.2% 2 dispose 6 Differentiate in Phosphomolybdic Acid 10 reuse 7 Rince in Acetic Acid 0.2% 5 dispose 8 Stain in Light Green 15 reuse 9 Rince in Acetic Acid 0.2% 0.5 dispose 10 Rinse in water 3 Results: Osteoid (collagen) Orange Calcified Bone Green Muscle Red Nuclei Black Ponceau Mixture 0.5g Acid Fuchsin 0.5g Ponceau de Xylidine 1.0g Orange G 500ml Acetic Acid 0.2% Phosphomolybdic Acid 2.0g Phosphomolybdic Acid 100ml Distilled water 161 HISTOLOGICAL PROCESSING PROTOCOLS AND STAINS ALIZARIN RED S FOR BONE Solution: 0.1ml Ammonium Hydroxide (NH4OH) C P . 58% 100ml Distilled Water 1 % Aqueous Alizarin Red S (pH 6.36 to 6.4) Stir dye into distilled water till only a few grains remain undissolved. Add 10ml of 0.1 % NH4OH slowly with constant stirring. Check pH. Solution is stable for one month. Method for slides ground to approx. 100um Stage Time [minutes] After use 1 Rinse in water 2 Alizarin Red S 5 reuse 3 Rinse in water Results: Calcified Bone Intense Red to Orange 162 APPENDIX B TORQUE CELL 163 TORQUE CELL All units in mm Scale 1:1 11.5 r - 5 I | r - U . 5 1 , 5 Material Strain Gages Bridge Type Max. Torque Non linearity full Aluminium 350Q Full Bridge 11Nmm 2% Wire configuration Red Black Green White Excitation + Excitation - Signal.+ Signal -164 TORQUE CELL Amplifier Settings SCXI 1334 National Instruments DAQ Gain 2 Bridge Gain 1000 Input limits +/- 0.0025 Bridge Ex Voltage 5 Scan rate [Hz] 20 Refresh rate [Hz] 5 Sensitivity [Nmm/V] 8479.2 Shunt A Resistor [Ohm] 1000 Shunt Torque [Nmm] 21.19 Filter [Hz] 10 165 APPENDIX C DRILL GUIDE FOR BONE CHAMBER 166 DRILL GUIDE FOR BONE CHAMBER in in OJ LO lOJ — OJ - E s -E E T in OJ i n oo •e u 167 DRILL GUIDE FOR BONE CHAMBER OJ Hen in OJ oo •e-0 0 E -S2 E 0 => ro = o < CO L O L O CO OJ OJ - 3 " OJ i CO 168 DRILL GUIDE FOR BONE CHAMBER 169 DRILL GUIDE FOR BONE CHAMBER 170 APPENDIX D PUSH-OUT LOAD-DISPLACEMENT CURVES 171 PUSH-OUT LOAD-DISPLACEMENT CURVES Section 15 Section 17 1 2 3 Displacement [mm] 1 2 Displacement [mm] Section 19 Section 21 0.5 1 1.5 Displacement [mm] 0.2 0.4 0.6 Displacement [mm] Section 23 Section 25 0.5 1 1.5 Displacement [mm] 1 2 3 Displacement [mm] Section 27 2 3 Displacement [mm] Specimen 1024 172 PUSH-OUT LOAD-DISPLACEMENT CURVES Section 14 0.5 1 1.5 Displacement [mm] Section 16 0.2 0.4 0.6 Displacement [mm] Section 18 0.5 1 15 Displacement [mm] 90 80 70 60 50 40 30 20 10 0 Section 20 / 2 3 Displacement [mm] Section 22 Section 24 Displacement [mm] Displacement [mm] Section 26 Specimen 1034 Displacement [mm] 173 PUSH-OUT LOAD-DISPLACEMENT CURVES 1200 Section 12 Section 14 0.5 1 1.5 Ds placement [mm] 1 2 3 Displacement [mm] Section 16 Section 18 1 2 Displacement [mm] 2 3 Displacement [mm] Section 20 Section 22 1 2 3 Displacement [mm] 1 2 3 Displacement [mm] Section 24 2 3 Displacement [mm] Specimen 1066 174 PUSH-OUT LOAD-DISPLACEMENT CURVES 250 Section 14 Section 16 200 3 CL 1 2 Displacement [mm] 1 2 Displacement [mm] Section 18 Section 20 0.5 1 Displacement [mm] 1 2 3 Displacement [mm] Section 22 Section 24 2 3 4 Displacement [mm] 1 2 Displacement [mm] Section 26 Displacement [mm] Specimen 1068 175 PUSH-OUT LOAD-DISPLACEMENT CURVES Section 12 Section 14 0.5 1 1.5 Displacement [mm] 1 2 Displacement [mm] Section 16 0.5 1 Displacement [mm] 500 450 350 Section 18 t 250 200 1 2 Displacement [mm] Section 20 Section 22 0 1 2 3 4 Displacement [mm] 176 PUSH-OUT LOAD-DISPLACEMENT CURVES Section 14 1 2 Displacement [mm] 250 Section 16 1 2 Displacement [mm] Section 18 Section 20 1 2 Displacement [mm] 1 1.5 Displacement [mm] Section 22 Section 24 1 2 Displacement [mm] 1 2 Displacement [mm] Section 26 0_ 2 3 Displacement [mm] Specimen 1094 177 APPENDIX E BONE CHAMBER TORQUE-ROTATION CURVES 178 TORQUE-ROTATION CURVES Specimen 4 Graft 0 w eek 1 1.5 Rotation [Deg] Specimen 5 Graft 0 week 1 1.5 Rotation [Deg] Specimen 4 Composite 0 w eek 1 1.5 Rotation [Deg] Specimen 5 Composite 0 w eek 0.5 1 1.5 Rotation [Deg] Rotation [Deg] Rotation [Deg] 1 7 9 TORQUE-ROTATION CURVES Specimen 9 Graft 0 w eek Rotation [Deg] Specimen 9 Composite 0 w eek 0 0.5 1 1.5 2 2.5 Rotation [Deg] Specimen 10 Graft Oweek 0.35 -, Rotation [Deg] Specimen 11 Graft 0 w eek 0.6 0.5 E 0.4 E z 0.3