HIP FRACTURES: UNDERSTANDING THE MECHANISM AND SEEKING PREVENTION THROUGH PROPHYLACTIC AUGMENTATION OF THE PROXIMAL FEMUR by Peter Michael de Bakker B. Eng., University of Western Ontario, 2003 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF APPLIED SCIENCE in THE FACULTY OF GRADUATE STUDIES "Mechanical Engineering" T H E U N I V E R S I T Y OF B R I T I S H C O L U M B I A J U L Y , 2006 © Peter de Bakker, 2006 Introduction ABSTRACT Introduction: In addition to having increased mortality and decreased mobility, hip fracture patients are more likely to sustain a second hip fracture than people who have never fractured a hip. There is currently no surgical technique that can be used to instantly strengthen the contralateral femur. Objectives: The primary objective was to evaluate the mechanical feasibility of augmenting the proximal femur with a novel implant to prevent hip fracture. This involved developing implants to prevent proximal femur fracture and quantifying the effect of the implants on the propensity of femurs to fracture. The secondary objective was to describe the initiation and progression of proximal femur failure during hip fracture to better understand the injury mechanism. Methods: Four cadaveric femurs were fractured in a fall configuration while being filmed with a high-speed camera. Visual analysis of the video images was used to describe the initiation and progression of fracture. Eight implant designs were developed and assessed with a finite-element model of a proximal femur. Of these designs, the Carbon Sleeve and Gamma Nail implants underwent a preliminary experimental assessment using a single pair of femurs. The Neck-Contouring Composite was further investigated using four pairs of femurs to compare to the Femoroplasty concept, which was assessed with three pairs of femurs. Results: In the hip fracture mechanism experiments, the four femurs were found to fail in a two stage process, with fracture initiating in the superior neck followed by a second failure in the inferior neck or intertrochanteric region. The preliminary experimental assessment of the Carbon Sleeve a nd Gamma Nail implants found that neither had a large effect on femur strength, with differences in failure load between control and augmented femurs of -8% and +1% respectively. The Femoroplasty implants had failure loads that were 24% higher that controls (p = 0.071). The Neck-contouring Composite implants had failure loads that were 18% higher than controls (p = 0.095). Summary and Conclusion: Hip fractures were seen to occur in a two stage process with failure initiating in the superior neck. Although only four specimens were tested, these experiments point to the superior neck as a region of interest for the development of screening tools for hip fracture risk assessment, targeted therapy for bone strengthening and surgical prophylaxis. The augmentation experiments found that it was possible to increase in the strength of the proximal femur by up to 24%. ii Introduction T A B L E OF CONTENTS Abstract i i Table of Contents i i i List of Tables vi List of Figures vii Acknowledgements x Chapter 1: Introduction 1 1.1 Anatomy of the Hip Joint 1 1.2 Bone '. 2 1.3 Osteoporosis 5 1.4 Material Properties 6 1.4.1 Material Properties of Cancellous Bone 6 1.4.2 Material Properties of Cortical Bone 7 1.5 Hip Fracture 8 1.5.1 Hip Fracture Location 9 1.5.2 Risk Factors for Hip Fracture 10 1.5.3 Clinical Treatment of Hip Fracture 12 1.5.4 Outcome of Hip Fracture 12 1.6 Second Hip Fracture 13 1.6.1 Second Hip Fracture Location 15 1.6.2 Risk Factors for Second Hip Fracture 15 1.6.3 Outcomes of Second Fracture 17 1.7 Hip Fracture Prevention 18 1.7.1 Pharmacological Hip Fracture Prevention 18 1.7.2 Biomechanical Hip Fracture Prevention Techniques 21 1.8 Biomechanics of Hip Fracture 21 1.8.1 Fractures Due to Falls or Falls Due to Fracture? 22 1.8.2 Fall Orientation 22 1.8.3 Hip Impact Resulting from a Fall 23 1.8.4 Load Distribution in the Femur During Fall 23 1.8.5 Relative Contribution of Cortical and Cancellous Bone to the Strength of the Femoral Neck 24 1.8.6 Fracture Initiation 24 1.8.7 In Vitro Hip Fracture Testing 24 1.9 Objectives 26 1.9.1 Primary Objective 26 1.9.2 Secondary Objective 26 Chapter 2: Methods 27 2.1 Hip Fracture Mechanism: Experimental Analysis 27 2.1.1 Specimens 27 2.1.2 Experimental Apparatus 28 2.1.3 Specimen Preparation 29 2.1.4 Strain Gauges 29 2.1.5 Failure Test Protocol 30 2.1.6 Videography 30 2.2 Hip Fracture Mechanism: Finite Element Analysis 32 iii Introduction 2.2.1 Specimen 32 2.2.2 QCT Imaging 32 2.2.3 Bone Geometry 33 2.2.4 Bone Density Distribution 33 2.2.5 Mesh Generation 34 2.2.6 Material Properties 35 2.2.7 Failure Criterion 36 2.2.8 Applied Loads 36 2.2.9 Model Validation 37 2.3 Design of Femur Augmentation Implant 38 2.3.1 Design Criteria 38 2.3.2 Implant Designs 39 2.4 Experimental Assessment of Implants 41 2.4.1 Specimens 41 2.4.2 Control Group Femurs 42 2.4.3 Composite Pin 42 2.4.4 Gamma Nail 43 2.4.5 Neck-contouring Composite 44 2.4.6 Femoroplasty 46 2.4.7 Experimental Measures 47 2.4.8 Statistical Analysis 48 Chapter 3: Results 49 3.1 Fracture Mechanics: Experimental Results 49 3.1.1 Mechanical Test Results 49 3.1.2 Strain analysis 50 3.1.3 Fracture Type 52 3.1.4 Failure Location and Fracture Progression 52 3.2 Finite Element Model Results 58 3.2.1 Cadaver Model Strain Measurements 58 3.2.2 Strain Predicted by Finite Element Model 61 3.2.3 Failure of Intact Finite Element Model of Femur 63 3.3 Finite Element Analysis of Augmented Femurs 64 3.4 Cadaveric Assessment of Augmentation Designs 66 3.4.1 Carbon Sleeve (Implant b) 67 3.4.2 Gamma Nail (Implant h) 68 3.4.3 Femoroplasty 69 3.4.4 Neck-contouring Composite (Implant f) 70 Chapter 4: Discussion 73 4.1 Synthesis of Results 73 4.1.1 Fracture Mechanism Experiments 73 4.1.2 Augmentation Results 76 4.1.3 Augmentation Results: Finite Element Model 77 4.1.4 Augmentation Results: Experimental Model 78 4.1.5 Augmentation Results: Comparison of Experimental to FE Results 81 4.1.6 Augmentation Results Summary 82 4.2 Comparison to Similar Studies 83 4.2.1 Finite Element Model 83 ' iv Introduction 4.2.2 Experimental Model 86 4.2.3 Effect of Femur Augmentation 88 4.3 Strengths 90 4.3.1 Fracture Mechanism 90 4.3.2 Femur Augmentation 90 4.4 Limitations 91 4.4.1 Mechanical Test Apparatus 91 4.4.2 Finite Element Model Limitations 93 4.4.3 Fracture Mechanism Experiments 93 4.4.4 Femur Augmentation Experiments 94 4.5 Future Work 95 Chapter 5: Conclusions 98 5.1 Femur Augmentation 98 5.2 Hip Fracture Mechanism 98 References 99 APPENDIX A Load-Displacement Curves for Augmented and Control Femurs 106 v Introduction LIST OF TABLES Table 1-1 Ultimate strength for femoral cortical bone from Reilly (1975) [24] 8 Table 1-2 Modulus and Poisson's ratio for femoral cortical bone from Reilly (1975) [24] 8 Table 1-3 Second Hip Fracture 14 Table 1-4 Relative risk of second hip fracture 16 Table 2-1 Donor information for specimens used in fracture mechanism experiments 28 Table 2-2 Summary of apparent densities and modulus for the cancellous elements 36 Table 2-3 Donor information for specimens used in experimental assessment of implant designs 42 Table 3-1 Summary of results from fracture mechanics experiments 49 Table 3-2 Summary of results from carbon sleeve augmentation test 67 Table 3-3 Summary of results from Gamma Nail augmentation test 68 Table 3-4 Ultimate loads for augmented and control femurs in femoroplasty group 69 Table 3-5 Work of fracture for augmented and control femurs in femoroplasty group 70 Table 3-6 Fracture types for augmented and control femurs in femoroplasty group 70 Table 3-7 Maximum temperature increase and volume of bone cement used for each specimen that was augmented with the neck-contouring composite 70 Table 3-8 Ultimate loads for augmented and control femurs in neck-contouring composite group 71 Table 3-9 Work of fracture for augmented and control femurs in neck-contouring composite group 71 Table 3-10 Fracture types for augmented and control femurs in neck-contouring composite group 72 Table 3-11 Maximum temperature increase and volume of bone cement used for each specimen that was augmented with the neck-contouring composite 72 Table 4-1 F E M validation results from Lotz et al. [100] 83 Table 4-2 Summary of failure loads and work of fracture from in vitro hip fracture studies 87 Table 4-3 Comparison of augmentation by Heini et al. and the femoroplasty and neck-contouring composite procedures from this thesis 89 vi Introduction LIST OF FIGURES Figure 1-1 Anatomy of hip joint. Modified fromNetter, 2002 [2] 2 Figure 1-2 Cortical and cancellous bone from Khan, 2001 with permission [4] 3 Figure 1-3 Strength and modulus of cancellous bone plotted against density. From Hayes, 1991 with permission [19] 6 Figure 1-4 Hip fracture location from Zuckerman, 1996, with permission [37] 10 Figure 1-5 Fracture load vs B M D from Lotz and Hayes, 1990 [46] 11 Figure 1-6 Direction of load applied to proximal femur from acetabulum and general stress distribution through the cross section of the neck. Blue (+) indicates compressive stress and red (-) indicates tensile stress. From Turner, 2005 with permission [85] 22 Figure 1-7 Fall type loading configuration. The femur is held with the shaft at 10° from the horizontal and is internally rotated 15°. The load is applied to the greater trochanter. The head and distal shaft are free to translate in the horizontal plane From Eckstein et al., with permission [93] 25 Figure 2-1 Photograph and schematic diagram of the experimental apparatus. The yellow arrows on the photograph show the degrees of freedom at the proximal and distal ends of the femur 29 Figure 2-2 Strain gauge location on superior neck (left) and inferior neck (right). White beads were bonded to femur to investigate the possibility of using video analysis to track failure in future experiments 30 Figure 2-3 Test set-up for the fracture mechanism experiments. The load was applied to the femur in the test apparatus by the Instron. Two cameras filmed the test, one from the anterior side and one from the posterior side. Two 250W lights were used to illuminate the specimen 31 Figure 2-4 Camera still on the left shows the anterior surface of the femur and a mirror image of the inferior surface of the femur. Camera still on the right shows the posterior surface of the femur and a mirror image of the superior surface of the femur 31 Figure 2-5 Schematic of loads applied to the finite element model 37 Figure 2-6 Strain gauge locations on femur specimen 1090R 38 Figure 2-7 Finite element models of implant designs 40 Figure 2-8 Injection of bone cement (left) and insertion of carbon sleeve (right) in the composite pin procedure 43 Figure 2-9 Anteroposterior and lateral radiographs of specimen 1161 L after augmentation with the composite pin 43 Figure 2-10 Anteroposterior and lateral radiographs of 1171 R after augmentation with Gamma Nail 44 Figure 2-11 Device used to aim guide wire 45 Figure 2-12 Kyphon and custom built balloons (left) and balloon inflated inside the femur (right) 45 Figure 2-13 Anteroposterior and lateral radiographs of 1157 L 46 Figure 2-14 Anteroposterior and lateral radiographs of 1153R after undergoing femoroplasty.. 47 Figure 2-15 Typical load-displacement for control femur showing failure load. The failure was defined to occur at the ultimate load and the work of fracture was defined as the area under the load displacement curve up to the failure load 48 vii Introduction Figure 3-1 Load-displacement curves for the specimens 1004 L , 1061 L, 1169 R and 1171 L loaded to failure at 100 mm/s 50 Figure 3-2 Strain measured in the inferior and superior femoral neck of 1004 L plotted against applied load 51 Figure 3-3 Strain measured in the inferior and superior femoral neck of 1061 L plotted against applied load 51 Figure 3-4 Fracture types in specimens used in fracture mechanics tests 52 Figure 3-5 Images corresponding to the first (left) and second (right) peaks of the load-displacement curve of 1004 L. The rotation shows the movement of the head of the femur seen at the first peak. The failure indicates a visible yielding at the second peak 53 Figure 3-6 Images corresponding to the final peak (left) and after the final peak in the load-displacement curve of specimen 1004 L 54 Figure 3-7 Images corresponding to the start of the loading sequence (left) and the loaded femur after passing two small peaks (right). The numbers 1 and 2 refer to the locations of failure at the first and second peaks respectively 54 Figure 3-8 Images corresponding to failure seen at the ultimate load (left) and at the final drop in load (right) 55 Figure 3-9 Images showing femur prior to fracture (left) and immediately following the initiation of fracture (right) in specimen 1169 R 56 Figure 3-10 Image showing final failure of specimen 1169 R 56 Figure 3-11 Images showing femur prior to fracture (left) and immediately following the initiation of fracture (right) in specimen 1171 L 57 Figure 3-12 Image showing final failure of specimen 1171 L 58 Figure 3-13 Strain measurements at three locations in a plane through the proximal neck when the femur was loaded to 435 N 59 Figure 3-14 Strain measurements at three locations in a plane through the distal neck when the femur is loaded to 435 N 60 Figure 3-15 Strain measurements at three locations in a plane through the distal neck when the femur is loaded to 435 N 61 Figure 3-16 Comparison of strains measured experimentally and strains predicted by finite element model 62 Figure 3-17 Plot of predicted strain vs. measured strain at 9 locations 62 Figure 3-18 Posterior view of von Mises strain distribution in proximal femur when loaded to 3618 N in a fall configuration 63 Figure 3-19 Anterior and slightly inferior view of von Mises strain distribution in proximal femur when loaded to 3618 N in a fall configuration 64 Figure 3-20 Predicted failure load for intact model and eight implant designs 66 Figure 3-21 Ultimate load of all augmented and control femurs plotted against total proximal femur aBMD. Linear trend lines were drawn through the data points representing the control femurs (R 2 = 0.74) and the augmented femurs (R 2 = 0.79) 67 Figure 3-22 Photographs and radiograph of femur augmented with carbon sleeve after it had been fractured 68 Figure 3-23 Post-fracture images of femur augmented with Gamma Nail 69 Figure 4-1 Locations of initial compressive failure from fracture mechanics experiments. Locations are shown on a single femur 74 Figure 4-2 Post-fracture images of 1004 L showing proximity of fracture to superior gauge (left) and inferior gauge (right) 75 viii Introduction Figure 4-3 Failure location in proximal femur from Keyak et al., with permission [103] 85 Figure 4-4 Predicted failure location (grey elements) in proximal femur from Oden et al. [104] 85 Figure 4-5 Femoral anteversion shown for person lying on their side [116] 92 Figure 4-6 Load-displacement curve for the specimen 1153 R. This specimen was augmented with the femoroplasty procedure. The ultimate load, which was 6470 N , and the load at which there was an initial non-linearity in the curve, which was 4828 N , are shown 95 Figure A - l Load-displacement curve for femur augmented with carbon sleeve implant and matched-pair control from donor 1161 107 Figure A-2 Load-displacement curve for femur augmented with Gamma Nail implant and matched-pair control from donor 1171 108 Figure A-3 Load-displacement curve for femur augmented with femoroplasty implant and matched-pair control from donor 1153 109 Figure A-4 Load-displacement curve for femur augmented with femoroplasty implant and matched-pair control from donor 1160 109 Figure A-5 Load-displacement curve for femur augmented with femoroplasty implant and matched-pair control from donor 1169 110 Figure A-6 Load-displacement curve for femur augmented with neck-contouring composite implant and matched-pair control from donor 1090 I l l Figure A-7 Load-displacement curve for femur augmented with neck-contouring composite implant and matched-pair control from donor 1149 I l l Figure A-8 Load-displacement curve for femur augmented with neck-contouring composite implant and matched-pair control from donor 1154 112 Figure A-9 Load-displacement curve for femur augmented with neck-contouring composite implant and matched-pair control from donor 1157 112 i x Introduction ACKNOWLEDGEMENTS Over the course of my thesis, I have collaborated with many professors, students and staff in Division of Orthopaedic Engineering and the Bone Health Group. They have shared with me their knowledge, intellect, insight and time. In particular, I would like to give special thanks to my supervisors, Drs. Tom Oxland, Pierre Guy and Goran Fernlund. They have guided and advised me well throughout these last few years. Their passion for research has been a great inspiration to me. I would also like to thank Dr. Antony Hodgson for taking the time to serve on my committee. Several professors have been generous with their time, ideas and support. Thank you to Dr. Peter Cripton for giving me great advice on this project and beyond and for getting me started on the fracture mechanism part of the thesis. Thank you to Dr. Heather McKay for constant encouragement and to Dr. Karim Khan for great career guidance. To both Drs. McKay and Khan, thank you for including me in the Bone Health Group. Thank you to Dr. Steve Robinovitch for taking the time to help shape this project. Thank you also to Dr. Rizhi Wang, who was so generous with his knowledge about bones and bone failure. Many people in the lab collaborated with me on this project. Thank you to Cecelia Tang and Dr. Danmei Liu for helping me get started. Thank you to Dr. Teresa Liu-Ambrose and Sarah Manske. They have helped me with every aspect of this project and they were a joy to work with. Thank you to Vincent Ebacher for his help with strain gauges and to Amy Saari for her assistance with high speed video. Thank you to Simon Sjovold for being so open with his ideas and for always finding a way to fix what I thought was beyond repair. As always, my family has been my greatest support throughout this process. Thank you to my parents for teaching me the importance of finding my passion and encouraging me to pursue it. A special thank you to my Mom for her editing expertise. Finally, thank you to Jennifer Dagsvik for everything. Introduction Chapter 1: Introduction The purpose of this study was to design and evaluate surgical augmentation procedures for the prevention of hip fracture. The clinical question addressed the scenario immediately after an individual sustains a hip fracture. It could be stated as "Can the contralateral femur be augmented so that it will not break the next time the patient falls?" This question was asked because patients who fracture a hip are at an increased risk of suffering another hip fracture. Given the increased mortality and decreased mobility that result from hip fracture, patients could benefit greatly from having their contralateral femur reinforced. There are currently a number of approaches to decreasing hip fracture risk, including pharmacological treatments aimed at increasing bone mass and external mechanical devices used to dissipate the energy from a fall. There is currently no clinical technique to surgically augment the femur. That is the gap that this study aimed to fill. 1.1 Anatomy of the Hip Joint The hip joint is a ball-and-socket-type synovial joint located at the proximal end of the lower limb, connecting the lower limb and the pelvic girdle. The articulating surfaces of the hip joint are located on the acetabulum and the femoral head (Figure 1-1). These surfaces are covered with a thin layer of articular cartilage. The acetabulum is the socket of the ball-and-socket joint. The acetabulum is hemispherical with a notch located at the inferior part of the acetabular rim. This rim is extended by a fibrocartilaginous ring called the acetabular labrum, which deepens the socket and also covers the acetabular notch. The femoral head is the ball of the ball and socket joint. It is approximately two thirds of a sphere. It sits on the neck of the femur, which is oriented at about 115 to 140° from the long axis of the femur in the coronal (frontal) plane and at an angle of 10 to 30° of anteversion from the same plane [1]. The neck is also offset to the shaft, being anteriorly 1 Introduction translated. Where the neck meets the body of the femur, there are two prominent projections, the lesser and greater trochanters. Acetabular labrum Ligaments and joint capsule Synovial membrane Retinacular arteries Metaphysis Diaphysis Greater trochanter Cancellous bone Cortical bone Articular cartilage Epiphysis Acetabular branch of obturator artery Acetabular labrum Epiphyseal plate Lesser trochanter Figure 1-1 Anatomy of hip joint. Modified from Netter, 2002 |2) The hip joint is enclosed by a strong and loose fibrous capsule that is cylindrical in shape [3]. The fibrous capsule attaches medially to the acetabular rim and laterally to the base of the neck of the femur. The fibrous capsule is reinforced by three ligaments of the hip joint: the iliofemoral ligament, the ischiofemoral ligament and the pubofemoral ligament. The synovial membrane lines the fibrous capsule. This membrane completely encloses the joint cavity and contains the synovial fluid. Blood is supplied to the femur through the reticular arteries, which run under the synovial membrane into the neck and head of the femur, and by the artery of the ligamentum teres. 1.2 Bone Bone has several characteristics that are relevant to this study: its classification as cortical or cancellous, its microstructure and its ability to remodel. First, bone can be classified as cortical or cancellous. Both of these types are found in the proximal femur. Long bones such as the femur have three distinct regions: the diaphysis, the metaphysis and the epiphysis (Figure 1-1). In the proximal femur, the metaphysis and epiphysis consist primarily of cancellous bone with a 2 Introduction thin layer of cortical bone on the external surface of the femur. In contrast, the diaphysis includes a thick outer layer of cortical bone and very little cancellous bone. Figure 1-2 Cortical and cancellous bone from Khan, 2001 with permission [4] Bone is classified as cortical or cancellous depending on the organizationits structure (Figure 1-2). The most evident difference between these two types of bone is that cortical bone is much more dense than cancellous bone. The porosity of cortical bone is approximately 10%, while the porosity of cancellous bone is typically between 50% and 90% [5]. Cortical bone is made up primarily of osteons with interstitial bone between the osteons. Osteons (or Haversian systems), are cylindrical in shape and run parallel to the long axis of the bone. In the centre of the cylinder is a channel, the Haversian canal, which contains blood vessels, lymphatics, nerves and loose connective tissue. These Haversian canals are surrounded by approximately 20 to 30 cylindrical layers of lamellar bone. On the outside of each osteon is a layer of cement. The Haversian canals are interconnected by small perpendicular canals that are called Volkmann's canals. Cancellous bone, on the other hand, is characterized as a cellular solid consisting of a network of interconnected plate- and rod-like trabeculae [6]. The bone is arranged in a lattice structure that is aligned with the forces usually applied on the bone. The higher porosity of cancellous bone gives it more surface area for cellular activity. As a result, the bone turnover rate of cancellous bone is 8 times greater than that of cortical bone. The pores contain bone marrow which produces blood cells and bone cells. Cortical (compact) bone 3 Introduction In addition to classifying bone by type, it is important to look at its microstructural characteristics. At the microstructural level, bone is made of mineralized collagen. It has three major components: an organic component, an inorganic component and water. The organic matrix of bone is primarily made up of collagen. Type I collagen accounts for approximately 90% of the protein in bone and the other 10%> consists of noncollagenous matrix proteins, other collagen types, lipids and other macromolecules. The inorganic component of bone is primarily made up of a calcium phosphate mineral, which is analogous to crystalline calcium hydroxyapatite (Ca ] 0 (P0 4 )6 {OH)2). Approximately 2% of the organic matrix consists of cells that are responsible for the formation and resorption of bone. The major types of bone cells are osteoclasts, osteocytes, osteoblasts and bone lining cells. Osteoclasts are large bone resorbing cells. They attach to the bone, solubilize the apatite crystals and then digest the organic matrix. Bone lining cells line all surfaces of bone and regulate the flux of ions into and out of the bone. The layer of bone lining cells on the outside surface of bone is called the periosteum, and the layer on the inside surface is called the endosteum. Osteoblasts are bone forming cells that are derived from bone lining cells [7]. They produce unmineralized bone matrix and are also involved in the mineralization of bone. At the end of their bone forming life, osteoblasts can be transformed to bone lining cells or osteocytes, or they can die. When osteoblasts are buried between lamellae of bone, they become osteocytes. Osteocytes are the most abundant cell type in bone. They communicate with each other and with cells that line the outside of the bone through small canals called canaliculi. It is believed that osteocytes transmit information about strain magnitudes and strain distribution to other cells through processes that are housed in these canals. A third important characteristic of bone is that it remodels throughout life and adapts to the mechanical loading to which it is subjected. This behaviour was summarized by Julius Wolff in Wolffs Law, which can be stated as: "Bone will optimize structure, so as to withstand functional loading, and to ensure the metabolic efficiency of locomotion" [8]. Bone remodelling involves the coupling of bone absorbing osteoclasts and bone forming osteoblasts. In cortical bone, this coupling takes the form of cutting cones which couple osteoclasts and osteoblasts. The cutting cone bores through cortical bone forming a new osteon. Cancellous bone remodels on the 4 Introduction surface of the bone. Osteoclasts absorb bone at specific sites and the hole in then filled with osteoid. Bone remodelling occurs on bone surfaces such as periosteal, endosteal, Haversian canal and cancellous surfaces. 1.3 Osteoporosis Osteoporosis is a major risk factor for hip fracture [9-12]. The World Health Organization (WHO) defines osteoporosis as "a disease characterized by low bone mass and microarchitectural deterioration of bone tissue leading to enhanced bone fragility and a consequent increase in fracture risk" [13]. It is not well understood whether the effects of osteoporosis are the result of inadequate bone formation or too much bone resorption [14]. Most researchers in the field emphasize only the amount and distribution of bone. It is thought that in addition to decreased bone density, osteoporosis could also be associated with a reduction in the mineral content in the bone, a reduction in cancellous connectivity, the accumulation of cement lines, increased cortical porosity and the accumulation of fatigue damage [14]. A l l of these factors contribute to the increased fragility of osteoporotic bone. Osteoporosis is usually diagnosed by measuring bone mineral density (BMD). The WHO definition of osteoporosis is quantified as being a B M D that is less than 2.5 standard deviations below the mean value for a young and healthy person. Tenenhouse et al. used this definition of osteoporosis to determine the incidence of osteoporosis in Canada [15]. Among women age 50 or older, 12.1% were diagnosed with osteoporosis based on lumbar spine B M D , 7.9% were diagnosed with osteoporosis based on femoral neck B M D , and the combined incidence was 15.8%. Among men age 50 or older, 2.9% were diagnosed with osteoporosis based on lumbar spine B M D , 4.8% were diagnosed with osteoporosis based on femoral neck B M D , and the combined incidence was 6.6%. Decreased bone density has a dramatic effect on the mechanical properties of bone, and this is discussed in the following section. The relationship between bone density and the incidence of first and second hip fractures is discussed in the sections on risk factors for first and second fracture. 5 Introduction 1.4 Material Properties Bone is a very diverse tissue in terms of its material properties. In order to give some insight into the behaviour of bones, this section will discuss material properties of both cortical and cancellous bone. 1.4.1 Material Properties of Cancellous Bone The apparent density of cancellous bone is the density of a volume of cancellous when the volume includes both trabeculae and the empty space between them. The apparent density has been found to be related to the material properties of cancellous bone (Figure 1-3). This relationship is very important because it makes it possible to estimate the material properties of bone using conventional medical imaging techniques. The relationship between apparent density and ultimate strength has been reported as having both a linear [16] and a power law relationship [17]. Gibson reviewed a number of studies and plotted compressive strength against apparent density from the measurements made in the studies [18]. They found that the compressive strength of cancellous bone varies from less than 1 MPa to nearly 100 MPa. It was found that over a wide range of densities, compressive strength is related to bone density squared. This relationship was found by Carter and Hayes to be er = 68s 0 0 6 p2, where a is in MPa, s is in s~] and p is in gxcm~3 [17]. Density, p (g/cc) Density, p (g/cc) Figure 1-3 Strength and modulus of cancellous bone plotted against density. From Hayes, 1991 with permission [19] 6 Introduction Carter and Hayes found that the Young's modulus was proportional to the density cubed [17]. They quantified the relationship by the equation E = 3790s0 0 6 p3, where E is in MPa. Like other relationships between material properties and density, this relationship is not universally accepted and more recent research by Keaveny and Hayes suggests that the modulus is proportional to density squared [6]. Aging has been associated with decreases in cancellous B M D [20]. Riggs et al. used quantitative computed tomography (QCT) to measure cancellous B M D at various skeletal sites. They found an age-related difference in femoral neck cancellous B M D of -56% in women when comparing women over 80 years of age to women in the 20-29 years age group. This difference for the same age groups in men was -45%. The Poisson's ratio of cancellous bone is not a well understood property, due partly to the fact that it is very difficult to measure. In a review by Keaveny and Hayes, it was found that Poisson's ratios measured experimentally in cancellous bone varied from just below zero to just below one [6]. This large range indicates that Poisson's ratio of cancellous bone could be as variable as its strength and Young's modulus. It depends on location of the sample, direction of loading and other factors. In finite element models of bone, it is often assumed to be 0.3, which is an average value measured experimentally. 1.4.2 Material Properties of Cortical Bone The material properties of cortical bone are similarly complex. Cortical bone is anisotropic and mildly viscoelastic, and therefore its material properties depend on direction and rate of loading. Table 1-1 and Table 1-2 give some average values for ultimate strength, Young's modulus and Poisson's ratio for femoral cortical bone when loaded in both the transverse and longitudinal directions. Clearly, cortical bone is stronger when it is loaded in the longitudinal direction than in the transverse direction [21]. It is stronger and less anisotropic when loaded in compression than when loaded in tension. The strain rate dependence of the material properties of bone has been investigated by several authors [22, 23]. In the study by Wright and Hayes, human femoral bone was tested over a wide 7 Introduction range of strain rates and it was found that the ultimate strength of cortical bone is proportional to the strain rate raised to the 0.07 power [23]. Further, the modulus of elasticity was found to be proportional to the strain rate raised to the 0.05 power. Most often both strength and modulus are assumed to be proportional to strain rate raised to the 0.06 power. T a b l e 1-1 U l t i m a t e s t r e n g t h f o r f e m o r a l c o r t i c a l b o n e f r o m R e i l l y ( 1 9 7 5 ) [24] Loading Mode Ultimate Strength (MPa) Longitudinal Tension 133 Compression 193 Shear 68 Transverse Tension 51 Compression 133 T a b l e 1-2 M o d u l u s a n d P o i s s o n ' s r a t i o f o r f e m o r a l c o r t i c a l b o n e f r o m R e i l l y ( 1975) (24] Loading Mode Modulus (GPa) Poisson's Ratio Longitudinal 17.0 0.46 Transverse 11.5 0.58 Shear 3.3 The strength and modulus of elasticity of cortical bone decrease with age. The longitudinal modulus of elasticity and the tensile yield strength decrease by about 2% per decade, while the ultimate stress was also found to decrease at about 2% per decade and the ultimate strain decreased at about 5% per decade [25]. These decreases corresponded to a decrease in energy absorbed of approximately 7% per decade. Overall, these results imply that the cortical bone of the elderly is much more likely to fracture than the cortical bone of the young, because it has a lower ultimate strength and it can absorb less energy before breaking. 1.5 Hip Fracture The number of Canadians aged 65 or older who fractured a hip in 1993 was 23,375 [26]. Of these patients, 76.2% were women and 23.8% were men. The risk of hip fracture among Canadians increases exponentially with age. Using this trend along with the increase in the population of Canadians of age 65 or older projected by Statistics Canada, Papadimitropolous et 8 Introduction al. predicted an exponential increase in the incidence of hip fractures in this age group. They estimated that in the year 2041, approximately 88,124 Canadians over 65 would fracture a hip. Overall increases in hip fracture rates have also been predicted for the United States [27-29] and worldwide [28, 30]. Most studies base their predictions on current incidence of hip fracture for different age groups. However, some authors suggest that there is also an age-independent increase in hip fracture [31, 32]. A review paper has estimated the age-adjusted rate of increase to be between 0.8% and 3.0% per year for men and between -1.6% and 2.7% per year for women [31]. The annual direct cost of hip fractures in Canada is estimated at $650 million annually and is projected to increase to $2.4 billion by 2041 [26]. The cost per fracture in Canada has been estimated to be $26,527 during the first year of care [33]. The annual cost directly associated with hip fracture in the United States has been estimated to be $8.7 billion annually. The estimated cost per fracture in the United Sates has been estimated to be between $32,428 and $81,300 [34, 35]. 1.5.1 Hip Fracture Location Hip fractures can be classified according to their general location. Because fractures can span several regions of the proximal femur, the classifications developed by researchers are, to some extent, subjective but were empirically developed following recurrent patterns and therefore represent clinical conditions. Several authors have classified fractures as being in the femoral neck, intertrochanteric region or subtrochanteric region [36, 37], see Figure 1-4. Zuckerman indicated that approximately 5 to 10% of hip fractures occur in the subtrochanteric region, with the remaining 90 to 95% divided evenly between femoral neck and intertrochanteric fractures. Michelson et al. reported that 14% of hip fractures were subtrochanteric, 49% were intertrochanteric and 37% were femoral neck. Other authors classify hip fractures as intra-capsular or extracapsular depending on their relationship with the joint capsule of the hip. This study will use the classification shown in Figure 1 -4, which represents the classification that is most commonly used in a clinical setting, as it describes the most common fracture patterns and relates them to identifiable anatomical structures. 9 Introduction Subtrochanteric region Intertrochanteric region \ Greater trochanter Figure 1-4 Hip fracture location from Zuckerman, 1996, with permission [37] 1.5.2 Risk Factors for Hip Fracture A number of risk factors for hip fracture have been identified. The two most significant are falls and osteoporosis. Most of the other risk factors are related in some way to falls or osteoporosis. Over 90% of hip fractures are associated with falls [38, 39]. A fall generally refers to a fall from standing height or lower, which would not be expected to result in a fracture in young, healthy people. Between 35 and 40% of people over the age of 65 living at home fall at least once a year and between one third and one half of those fall twice or more a year [40]. Of these falls, only about 1% result in a hip fracture [41]. Although most falls do not result in a fracture, they can lead to fear of falling, loss of confidence and functional deterioration [41]. Cummings and Nevitt found that four important factors determine whether a fall will result in a hip fracture: fall orientation, protective responses, local shock absorbers and bone strength at the hip [42]. Studies have also shown that osteoporosis is a significant risk factor for hip fracture. Osteoporosis is commonly measured by dual x-ray absorptiometry (DXA), a clinical imaging tool that is the gold standard for osteoporosis diagnosis [43]. D X A uses two beams of different energy to produce a two-dimensional representation of the attenuation distribution of the bone. The output of D X A scans is aBMD (g/cm2), which is an estimation of the areal density of the 10 Introduction bone. Hip aBMD has been shown to correlate with both hip fracture risk in clinical studies [9-12] as well as femur strength in biomechanical tests [44-46]. A prospective-cohort study involving 9,704 women found that the relative risk for hip fracture was 2.4 per standard deviation decrease in femoral neck B M D [11]. The researchers also found that about 51% of hip fractures were attributable to osteoporosis when it is defined as a T-score of less than 1.5, and only about 28% of hip fractures were attributable to osteoporosis when it is defined as a T-score of less than 2. Pulkkinen et al. found that combining B M D with measurements of neck-shaft angle and cortical thickness improved the prediction of hip fracture [47]. This would highlight the fact that hip fracture risk is related to both bone density and falls; however, it could also point out the inability of D X A to characterize bone strength. 5000 0-1 , 1 1 H 0 500 1000 1500 2000 2500 K 2 H P 0 4 (mg/cm 3) -Total Area (cm2) Figure 1-5 Fracture load vs BMD from Lotz and Hayes, 1990 [46] To refute the last statement, biomechanical failure tests consistently show a correlation between failure load and aBMD [44-46]. Lotz and Hayes measured B M D and fracture load in vitro [46]. Figure 1 -5 shows fracture load plotted versus average intertrochanteric equivalent bone mineral density determined by quantitative computed tomography. It can be seen from this plot that intertrochanteric B M D is a good predictor of femoral strength. Although femoral strength alone does not determine whether a hip will fracture, this in vitro study supports the results mentioned above of the ability of B M D to predict fracture risk with an important contribution of fall risk to fracture risk. 11 Introduction 1.5.3 Clinical Treatment of Hip Fracture Most hip fractures are treated with surgery, however some hip fractures can be treated non-surgically [48]. It is possible to treat non-displaced, intracapsular fracture in patients less than 70 years of age with bed rest followed by limited mobilization, although the risk of fracture displacement tends to favour some form of minimally invasive operative repair in such situations. A review of the hip fracture care in Canada by Statistics Canada further supports that only a minority of patients with hip fractures (less than 6%) are treated non-operatively [49]. A systematic review by Chilov et al. provided evidence-based guidelines for the treatment of fractured hips [50]. They suggest that trochanteric fractures should be treated surgically with a compression hip screw and plate. They found that this treatment has less chance of failure leading to reoperation making it more cost-effective in the long run. Undisplaced femoral neck fractures should have internal fixation with a method that is familiar to the surgeon such as cancellous bone screws or compression screw and plate. They did not find any treatment for displaced femoral neck fractures that is clearly superior and suggest that patient factors should guide the surgeon's decision. The two options for displaced femoral neck fractures are internal fixation or hemiarthroplasty. Hemiarthroplasty is much less likely to fail than internal fixation. The patient's age, presence of arthritis, availability and cost of the different types of treatment and surgeon experience and preference should aid in the selection of a surgical treatment. Lyons had similar recommendations and noted that hemiarthroplasty is favored in the United States and the United Kingdom while internal fixation with screws is favored in Scandinavian countries [48]. The previously mentioned report from Statistics Canada shows that 64% of those with femoral neck fracture were treated with some form of arthroplasty, while 33% underwent internal fixation [49]. 1.5.4 Outcome of Hip Fracture Hip fractures have serious consequences on mortality and mobility. The one-year mortality rate for hip fracture has been found to be between 12 and 36% [51]. A review of 32,590 hip fracture patients aged 65 or older admitted to hospital with a fracture of the femoral neck between 1968 and 1998 found that 32.7% of these patients died within a year of fracturing their hip [52]. First-year fatality rates were higher for men than for women. Fatality rates decreased over the first 12 Introduction year after fracture and levelled out by the 12 month, with no significant difference in the fatality rate of men or women compared to the general population. First-year fatality rates increased sharply with age. Fatality rates decreased between the late 1960s and early 1980s, but have not decreased significantly over the past 20 years. The change in the treatment of hip fractures over the years to a more operative approach favouring early mobilization of patients over the years suggests an overall benefit to operative care. Those who survive a hip fracture face reductions in their mobility. At one year after a hip fracture, 40% of patients cannot walk and 60% have trouble with at least one essential activity of daily living (such as driving or shopping), and 80% are restricted in other activities [53]. Magaziner et al. assessed hip fracture patients using a number of measures of independence [54]. They found that patients became more dependent on all measures, ranging from 20% being unable to put on their pants unassisted to 90% being unable to walk up 5 stairs unassisted. Rosell and Parker found that the number of people who could walk without aids fell from 59% prior to fracture to 26% after fracture [55]. 1.6 Second Hip Fracture Though several studies have examined the epidemiology of second hip fractures, the incidence and relative risk of second hip fractures is still not well known. It is generally agreed that the risk of fracturing a hip is increased by the occurrence of a previous hip fracture. A n early study on the occurrence of a second hip fracture found that the risk of suffering a hip fracture increased 20 times after an initial hip fracture [56]. More recently, Schroder et al. found that the risk increased 9 times in men and 6 times in women [57]. However, Melton et al. found that the risk increased only 1.6 times [58]. Selected studies that have looked at the occurrence of second hip fracture are summarized in Table 1-3. The studies by Shabat et al, Chiu et al, Boston, Dretakis et al, Dinah, Pearse et al. and Di Monaco et al. looked at all hip-fracture patients admitted to their respective hospitals over a period of time. When patients were admitted, they were checked for a previous contralateral fracture. Using these data, the authors found that between 2.3 and 10.6% of hip-fracture patients had previously fractured their other hip. 13 Introduction Table 1-3 Second H i p Fracture Occurrence , Mean . _ Length of . Mean Age at ^ ^ contralateral . , , 1st fracture , c ^ interval (years) fracture Study Number of subjects Number of contralateral fractures Admittance studies Boston, 1982 [59] Chiuef a/., 1992 [60] Di Monaco et al., 2002 [61] Dinah, 2002 [62] Dretakis et al., 1981 [63] Dretakis et al., 1998 [64] Pearse et al., 2003 [65] Shabat et al., 2003 [66] Retrospective studies Melton et al., 1982 [58] Schroeder et al., 1993 [57] Wolinsky and Fitzgerald, 1994 [67] Prospective studies Chapurlat et al., 2003 [68] Dolk, 1989 [69] Stewart et al, 1999 [70] 500 1514 372 186 1333 1685 886 1145 3898 368 632 282 394 54 35 39 22 99 106 49 84 81 235 27 50 49 27 10.6% 2.3% 10.5% 11.8% 7.4% 6.3% 12% 9.5% 7.1% 6.0% 7.3% or 3.0% /year 7.9% or 2.3%) /year 17.4% 6.9% 75% within 3 years 23.9 mths 3.4 years 31 mths (median) 1.7 years 2.3 years 79 (at second) 79 82 83 79 77 n . 74 - men 3.3 years _ n J 79 - women 79.7 75 34 16 3.8 10 5-10 The studies by Melton et al., Schroeder et al. and Wolinsky et al. were restrospective studies. These studies tracked hip fracture patients using hospital admittance records and Medicare billing information. They found that second hip fractures occurred in 6.0% to 7.3%) of hip-fracture patients over the periods of time studied. The studies by Chapurlat et al., Stewart et al. and Dolk et al. prospectively followed hip fracture patients. They found that over the period of their studies, between 6.9%) and 17.4% of hip fracture patients fractured their contralateral femur. 14 Introduction Overall, the incidence of second hip fracture is estimated to be between 2.3% and 17.4%. Although the incidence of second hip fracture is not accurately known, the studies tend to agree that incidence of hip fracture is greater in patients who have suffered a hip fracture than in controls. 1.6.1 Second Hip Fracture Location Two points are worth noting regarding the location of second hip fractures. First, approximately 92% of second hip fractures occur in the contralateral hip [57, 68]. After a hip fracture, the risk of fracturing the same hip again is less than what it would have been i f it had not already been fractured [58, 71]. This could be due to strengthening from internal fixation or callus formation around the healing hip. Second, many authors have reported that contralateral hip fractures tend to occur in the same part of the proximal femur as the first fracture [57-59, 64-66, 68, 71]. For example, Schroder et al. found that 68% of contralateral fractures were preceded by a fracture of the same type [57]. Therefore, most second hip fractures occur in the contralateral femur and in the same region as the first hip fracture. 1.6.2 Risk Factors for Second Hip Fracture A number of studies have analyzed risk factors for second hip fracture. This information is potentially useful in assessing candidates for a prophylactic intervention. The findings from some of these studies are summarized in Table 1-4. Chapurlat et al. followed a group of 632 women who had suffered a hip fracture to determine i f it was possible to predict who would fracture a second hip [68]. They found that 53 women fractured another hip after an average time interval of 2.3 years. This incidence of second hip fracture was 4 times the incidence of a first hip fracture in the same cohort of women. They assessed 39 potential risk factors. They found that women who walked for exercise, women who had normal depth perception and women who were taking estrogen at baseline were less likely to fracture a second hip. The relative risk was increased for women who had lost weight since the age of 25 and for those who had a low calcaneal bone mineral density. 15 Introduction Table 1-4 Relative risk of second hip fracture Variable Relative 95% Confidence Risk Interval Chapurlat et al, 2003 [68] Walking for exercise 0.5 0.3-0.9 Normal depth perception 0.5 0.3-0.9 Weight loss since age 25 2.7 1.6-4.6 Calcaneal BMD * 1.5 1.1-2.0 Currently taking estrogen 0.5 0.3-0.9 Stewart et al., 1999 [70] Weight (kg) 2.01 1.26-3.21 BUAMcC* 1.64 1.06-1.55 Neck BMD* 2.02 1.27-3.22 Total body BMD* 1.69 1.07-2.68 Mobility* 1.45 1.07-1.98 New fall since attendance 1.63 1.08-2.48 Wolinsky et al., 1994 [67] Perceived health 2.21 1.03-4.72 Dizziness 2.58 1.20-5.51 * Relative risk is given per standard deviation from the mean Stewart et al. followed 394 women with a hip fracture [70]. Of these women, 27 suffered a second hip fracture over a 5 -to- 10-year period. The average time interval between the hip fractures was 2.9 years. They found several predictors of second hip fracture to be significant. These predictors were weight, broadband ultrasound attenuation as measured by the McCue C U B A Clinical (BUAMcC), neck B M D and total body B M D . Wolinsky et al. examined 368 people who had hip fractures, and 27 of these people later fractured another hip after an average of 613 days [67]. They found that of the 29 potential risk factors that they assessed, the only ones that had significant predictive ability were perceived health status and reported problems with dizziness. Propensity to fall could have predictive value in assessing who is at risk for a second fracture. Dretakis et al. asked patients how many times they fell per year when they were admitted for either a first or second hip fracture [64]. They found that those admitted for a second fracture were 4 times as likely to have fallen twice or more per year. Stewart et al. looked at the number of times the patient had fallen in the twelve months prior to attending the clinic as well as the number of times they fell after attending [70]. The number of times the patient had fallen prior to attending was not statistically significant (RR=1.38, 95% CI [0.94-2.03]). Those patients who fell after attending were more likely to suffer a second hip fracture. This could be explained by 16 Introduction the fact that the vast majority of patients that re-fractured would have done so in a fall. Wolinsky et al. asked how many falls the patient had in the year prior to the start of the study [67]. They found no relation between this variable and a subsequent hip fracture. This could be because the surveys were filled out at the start of the study and not at the time of the first fracture. Chiu et al. found that concomitant neurological diseases (stroke or Parkinson's) were found more frequently in patients with two hip fractures [60]. They also found that sequential hip fractures occurred more commonly in institutionalized patients. They found that a biochemical change of osteomalacia at the time of the first hip fracture was also associated with the occurrence of a second hip fracture. These studies have identified B M D , weight and weight loss and perceived health, among other factors, as tools that could be used to identify people at risk for second hip fracture. 1.6.3 Outcomes of Second Fracture It has been suggested by several authors that the outcome of a second hip fracture is worse than that of a first fracture [59, 64, 65]. The three outcomes that were found to be worse in second hip fractures were the type of fracture, patient's mortality rate and the patient's loss of mobility. With regard to the type of fracture, Dretakis et al. found that second hip fractures tended to be more unstable and displaced than the first [64]. They suggested that this could be due to a progressive reduction of bone mass as well as the impairment of the patient's mobility. Boston et al. looked at the mortality rate of hip fracture patients. They found that patients who fractured their first hip had a mortality rate of 13% at 3 months while those who fractured a second hip had a mortality rate of 30% at 3 months [59]. This difference suggests that an increased mortality rate is associated with second hip fractures when compared to first hip fractures. Pearse et al. looked at the mobility of patients with sequential contralateral hip fractures [65]. They found that after a first fracture, only one of 24 patients who could walk without supports before the fracture could do so after, though they could still walk without the assistance of an able-bodied person. After their first fracture, 91% of the patients who could walk without assistance of an able-bodied person prior to their fracture could do so after. After a second fracture, only 53%) of patients who could walk independently (including those who used.supports 17 Introduction such as a cane) prior to the second fracture could still walk without help. They also found that second hip fractures had a negative effect on social independence. Of the patients in their study who lived at home prior to their first fracture, 64% were able to return home. After these patients suffered another fracture, only 54% returned home. There was no statistically significant difference between the length of hospital stay after the first and second fractures. Overall, they found that patients who fractured a hip on average became less mobile and less socially independent. After a second fracture, patients became even less mobile and socially independent. These changes in level of function following a second hip fracture along with the noted increase in mortality motivate the need to develop second hip fracture prevention strategies. 1.7 Hip Fracture Prevention There are a number of pharmacological and biomechanical interventions that are used clinically or have been proposed to be used clinically to reduce the risk of hip fracture. Pharmacological interventions aim to improve the strength of the bone by increasing the bone density. Biomechanical interventions aim to dissipate the energy from a fall or increase the strength of the femur. A l l of these types of interventions are discussed in the following sections 1.7.1 Pharmacological Hip Fracture Prevention A number of drugs have been studied as treatments for osteoporosis including calcium and vitamin D, bisphosphonates, hormone replacement therapy, calcitonin, parathyroid hormone and selective estrogen-reception modulators. These pharmacological interventions can reduce the risk of hip fracture by slowing the rate of bone turnover. The ability of these drugs to increase B M D and prevent hip fracture as well as side effects associated with them are discussed below. Some of these drugs have been shown to reduce the risk of hip fracture, while others appear to only reduce the risk of vertebral fractures. A l l of these drugs have been shown to increase B M D . Because hip fractures are less common than vertebral fractures, larger studies are required to show a significant effect. Calcium and vitamin D3 supplements are known to be effective in the prevention of osteoporosis. The Canadian Medical Association guidelines for the treatment of osteoporosis state that "adequate calcium and vitamin D through diet or supplements are essential for the 18 Introduction prevention of osteoporosis and, taken together, are essential adjuncts to preventative therapy" [72]. They caution that calcium and vitamin D should not be used as the sole treatment for osteoporosis. A large placebo-control study found that women who were treated with calcium and vitamin D had 43% fewer hip fractures than controls [73]. They also had a 2.7% increase in B M D compared with a 4.6% decrease in B M D for controls. Another study by this group found that calcium and vitamin D were a cost effective method of hip fracture prevention [74]. A n additional benefit of vitamin D therapy is that it increases muscle strength and thus could potentially reduce the risk of falling in the elderly [75]. Bisphosphonates are a class of compounds that act by selectively inhibiting osteoclast function. This slows the rate of bone resorption and thus increases bone density. The Canadian Medical Association recommends bisphosphonates as a first line preventative therapy in postmenopausal women and men with low bone mass or osteoporosis. Hodsman et al. reviewed a number of randomized control trials that looked at the effect of these drugs on the incidence of vertebral, non-vertebral and hip fracture [76]. They found that patients who were treated for 3-4 years had B M D increases o f 1.6 to 3.8%> in the femoral neck. These increases in B M D translated to significant reductions in vertebral fracture incidence, however only two of the studies reviewed found a statistically significant reduction in hip fracture incidence (RR = 0.5). The use of estrogen replacement therapy for prevention of osteoporotic fractures is controversial, and the results of most studies that look at hip fracture are inconclusive [77]. The Women's Health Initiative is a large randomized control trial that studied the effect of estrogen and progestin on 16,608 women aged 50 to 79 [78]. They found that daily use of estrogen plus progestin reduced the risk of hip fracture (RR=0.66, CI (0.45-0.98)) but increased the risk of breast cancer (RR=1.26), venous thromboembolic disease (RR=2.11), stroke (RR=1.41) and coronary heart disease (RR=1.29). The benefits of hormone replacement therapy may be short term. A study by Yates et al. used data from women who participated in the National Osteoporosis Risk Assessment [79]. They found that women who had ceased using hormone replacement therapy for 5 years had a risk of hip fracture similar to the general population. Women who had ceased using hormone replacement therapy within the previous 5 years had an increased risk of hip fracture (RR=1.65, CI (1.05-2.59)). 19 Introduction Calcitonin is a peptide hormone that occurs naturally in the body. It cannot be taken orally and is usually taken in a nasal spray. Recombinant salmon calcitonin is the standard form of the drug because it is more potent in humans than the human form of calcitonin. A review by Brown and Josse found that nasal calcitonin has been shown to be effective in the prevention of vertebral fractures in severely osteoporotic women but has not been proven to reduce non-vertebral fractures [72]. Another study found a reduced risk of hip fracture (RR = 0.1) in patients taking a low dose of calcitonin [80]. They did not find a reduced risk in patients taking higher dosages. Parathyroid hormone can stimulate both bone formation and resorption resulting in an increase or decrease in B M D , depending on how it is taken. Continuous infusion leads to greater bone resorption than daily injections. Neer et al. studied the effect of parathyroid hormone on the incidence of fractures in 1,627 postmenopausal women with prior vertebral fractures [81]. The study was terminated early because of the development of osteosarcomas in a long-term toxicology study in rats. Selective estrogen-receptor modulators block conformational changes of the estrogen receptor. Raloxifene hydrochloride is the only selective estrogen-receptor modulator that is approved for the prevention and treatment of osteoporosis [72]. In a study by Ettinger et al. the effect of raloxifene on fracture risk was examined in postmenopausal women [82]. They found a 30% reduction in the incidence of vertebral fractures but no significant difference in the number of hip fractures in particular. They did find that raloxifene increased femoral B M D . Overall, pharmacological interventions have been shown to increase B M D and reduce the risk of fractures in general. However, few drugs have been shown to reduce hip fractures specifically, and all carry risks of side effects. In the case where a risk reduction was observed, the results occurred after 3-4 years of treatment with the least effect on elderly frail individuals. In the setting of prevention of a second hip fracture where most occur within 2 years, a strategy which increases bone strength earlier would be favourable. 20 Introduction 1.7.2 Biomechanical Hip Fracture Prevention Techniques Biomechanical interventions for hip fracture prevention include hip protectors, which are used clinically, and the "femoroplasty" procedure, which has been investigated in an in vitro biomechanical study. Hip protectors have been widely investigated as a method of reducing hip fracture incidence, but their efficacy is a subject of debate. Hip protectors are made of some form of external padding that helps to absorb the impact energy of a fall. A systematic review by Parker et al. looked at 13 randomized control studies [83]. They separated the studies into those in which the participants were cluster-randomized by care unit, nursing home and nursing home ward and those in which the participants were individually randomized. They pooled the studies with individual randomization and found that the incidence of hip fracture was not reduced by wearing hip protectors. The studies with cluster randomization found that, for those living in institutional care with a high background incidence of hip fracture, hip protectors appear to reduce the incidence of hip fracture. They found that compliance was a problem. This could be because the hip protectors are uncomfortable, unattractive or, in institutional populations, difficult to manage in the setting of concurrent incontinence management. A new surgical procedure called a "femoroplasty" has been proposed to prevent hip fracture. In this procedure, polymethylmethacrylate (PMMA) bone cement is injected into the proximal femur. In a cadaver study, this procedure was found to increase the failure load of the proximal femur by 82% when it was loaded to failure in a fall loading condition [84]. In this study, they caution that this procedure generates a significant amount of heat with a resultant increase in temperature to an unacceptable level. This procedure has not been evaluated clinically. 1.8 Biomechanics of Hip Fracture The study of hip fracture requires an understanding of the biomechanics of hip fracture, including the mechanism of fracture, the characteristics of falls, load distributions in the femur, the contribution of cortical and trabecular bone to the strength of the femoral neck, fracture initiation and previous biomechanical research. The loading on the femur during gait is much different than that seen in a fall (Figure 1-6). The diagram in Figure 1-6 is a simplification of the stress in the femoral neck as it only shows the force applied to the femur from the pelvis. This general stress distribution demonstrates that during gait the inferior neck experiences large 21 Introduction compressive stresses, while the superior neck experiences relatively small tensile stresses. In a fall, this general stress distribution is reversed, with the superior neck experiencing large compressive stresses and the inferior neck experiencing smaller tensile stresses. Figure 1-6 Direction of load applied to proximal femur from acetabulum and general stress distribution through the cross section of the neck. Blue (+) indicates compressive stress and red (-) indicates tensile stress. From Turner, 2005 with permission [85] 1.8.1 Fractures Due to Falls or Falls Due to Fracture? There is some controversy over what proportion of hip fractures occur due to the fall mechanism described in the previous section and what proportion occur by other mechanisms. It is often quoted that over 90% of hip fractures are associated with falls [28], but it is unknown what proportion of these fractures associated with a fall were caused by the fall and what proportion occurred immediately prior to the fall [86]. Intrinsic factors that have been proposed to cause hip fractures without trauma include severe osteoporosis, muscle contraction, fatigue or stress fractures and localized disease such as metastatic cancer. A review by Y o u m et al. found that between 11%> and 25%> of hip fractures were caused by the "leg giving away" or "spontaneous fracture" [87]. Although it is often stated that 90%> of fractures are caused by a fall, it is likely that a proportion of these fractures are caused by other mechanisms. 1.8.2 Fall Orientation The direction of a fall has been associated with both the likelihood of sustaining a fracture and the type of fracture that occurs. Falls to the side are more likely to cause hip fracture than other types of falls [9, 88-90]. Nevitt and Cummings found that elderly community-dwelling women who fractured a hip were more likely to have fallen on their side or straight down (odds ratio X _ Falling 22 Introduction (OR) =3.3) and they were more likely to have landed on or near their hip (OR=32.5) than those whose fall did not result in a fracture [89]. Different types of hip fracture have been found to be associated with different types of falls. Hopkinson-Woolley and Parker found that falls that involved a twisting motion were associated with extracapsular fractures [91]. Overall, falls to the side and falls where the hip contacts the ground first are more likely to cause a fracture, and falls that involve a twist motion are more likely to result in extracapsular fractures. 1.8.3 Hip Impact Resulting from a Fall Fall biomechanics experiments have been used to measure velocity at impact and peak ground force and to estimate the loading rate on the femur. When the hip hits the ground in a fall, the energy of the fall is absorbed by the body. Researchers have studied this impact using human subjects fall and crash dummies in pelvis release experiments. Using video analysis of subjects falling on mats, the velocity of impact at the greater trochanter during a sideways fall has been estimated to be between 1.99 and 4.79 m/s [41, 92, 93]. The velocities were found to be higher in subjects that contracted their muscles during the fall. In order to estimate the forces in a fall, the impulse at impact has been investigated using pelvis release experiments, where live subjects are dropped from small heights onto a force plate [41, 94, 95]. The peak forces have been measured to be between 1145 and 6100 N . The peak force was found to be higher in people who landed with their trunk upright and in people who flexed their muscles when they fell. These impulse curves from these tests along with estimates of the stiffness of the soft tissue surrounding the femur have been used to estimate the displacement rate at the greater trochanter of the proximal femur. Dr. Steve Robinovitch estimated this loading rate to be approximately 330 mm/s [96]. 1.8.4 Load Distribution in the Femur During Fall Finite element modelling has been used to examine the relative load carried by the cortical and cancellous bone in the proximal femur. A finite element analysis of the proximal femur during a fall found that the relative load carried by cortical and cancellous bone during elastic loading varied with location in the proximal femur [97]. Lotz et al. found that cortical bone supports 30% of the load in the subcapital region, 50% of the load at the mid-neck, 96% of the load at the 23 Introduction base of the neck, and 70% of the load in the intertrochanteric region during a fall. This analysis found that the superior neck was loaded primarily in compression and the inferior neck was loaded primarily in tension. This indicates that in a fall, cortical bone carries most of the load in the lateral neck and intertrochanteric region, and cancellous bone carries most of the load in the medial neck. 1.8.5 Relative Contribution of Cortical and Cancellous Bone to the Strength of the Femoral Neck Cortical and cancellous bone both contribute to the strength of the proximal femur. Destructive mechanical tests have been used to measure the effect of removing the cancellous bone on the strength of the femoral neck [98]. The researchers found that removing the cancellous bone from the femoral neck resulted in 40% decrease in failure load compared with matched pair controls when the neck was loaded in isolation from the rest of the proximal femur. 1.8.6 Fracture Initiation Although it has been shown that cancellous bone plays a significant role in carrying load during a fall, a number of studies suggest that it is cortical bone that is of primary importance in the structural failure of the femur. In an editorial by Ferretti et al., it is asserted that hip fractures initiate in the intertrochanteric and femoral neck cortices; however, no conclusive experimental evidence is provided to back up these assertions [99]. A group at Cambridge has also proposed that it is cortical bone that initiates hip fractures in the femoral neck. Their fracture model hypothesizes that the impact of a fall causes a compressive buckling in the superolateral cortex of the femoral neck followed by failure due to tensile or torsional stresses in the remainder of the neck [100]. This theory is based on structural differences between people who fracture their hip and those who do not, as well as structural changes in the cortex of the femur with aging. Their studies comparing femoral necks from people who have fractured their femurs to those of people who have not have found differences between the two groups in cortical thickness and porosity. 1.8.7 In Vitro Hip Fracture Testing Hip fractures have been reproduced experimentally by loading cadaver femurs to failure in apparatuses that simulate falls as well as apparatus that simulate stance loading. Experiments that 24 Introduction load femurs in a stance configuration have fixed the shaft of the femur in a vertical position and applied a downward force to the head of the femur. A study by Beck et al. that loaded femurs to failure in a stance loading configuration found that the femurs broke in a vertical plane through the femoral neck from the base of the femoral head to the lesser trochanter [101]. They noted that the fracture plane differed from what was seen clinically. A s discussed in Section 1.9.1, the injury mechanism for spontaneous fractures is likely to be more complicated than a straight vertical force. F i g u r e 1-7 F a l l t y p e l o a d i n g c o n f i g u r a t i o n . T h e f e m u r is h e l d w i t h t h e s h a f t a t 1 0 ° f r o m t h e h o r i z o n t a l a n d is i n t e r n a l l y r o t a t e d 1 5 ° . T h e l o a d is a p p l i e d to the g r e a t e r t r o c h a n t e r . T h e h e a d a n d d i s t a l s h a f t a r e f r e e to t r a n s l a t e i n t h e h o r i z o n t a l p l a n e F r o m E c k s t e i n et at., w i t h p e r m i s s i o n [93] A number of studies have reproduced hip fractures similar to the fractures seen clinically by simulating fall loading. Although these studies generally use similar test apparatuses, they vary slightly in terms of the constraints on the femur, the location where the load is applied and the loading rate. In the most commonly used configuration, the femur is held with the shaft 10° from the horizontal and is internally rotated 15°. Lochmuller et al. and Eckstein et al. used this configuration, with the load applied to the femur at the greater trochanter [102, 103] (Figure 1-7). In this configuration, the crosshead acts like the floor hitting the greater trochanter. Other researchers have used the same femur orientation and applied the load to the head and restricted the vertical translation of the greater trochanter [44, 104, 105]. Pini l la et al. examined the effect of internal rotation on the failure load of cadaver femurs and found that the failure load decreased with increasing angle of internal rotation [106]. The internal rotation of 15° is not, however, based on published fall biomechanics data and has been used because it is a reasonable approximation of a sideways and slightly backwards fall. 25 Introduction Loading rate has been shown to have an effect on the fracture load of femurs. Courtney et al. showed that increasing the loading rate from 2 mm/s to 100 mm/s increased the failure load of femurs by 20% [104]. The increase in loading rate did not, however, have an effect on the energy to failure. Most other researchers looking at hip fracture have used quasi-static loading rates of 0.5 mm/s to 6.6 mm/s. 1.9 Objectives As stated earlier, there is a need to prevent second hip fracture. There is currently no surgical technique to instantly strengthen proximal femur for hip fracture prevention. To address this need, the following primary and secondary objectives were identified. 1.9.1 Primary Objective The primary objective of this study was to evaluate the mechanical feasibility of augmenting the proximal femur with a novel implant in order to prevent hip fracture. Specifically, I aimed to develop implants to prevent proximal femur fracture and quantify the effect of implants on the propensity of femurs to fracture through finite element modelling and mechanical testing with a cadaver model. 1.9.2 Secondary Objective The secondary objective of this project was to describe the initiation and progression of proximal femur failure using an experimental hip fracture model, and a finite element analysis of hip fracture. 26 Methods Chapter 2: Methods The Methods section of this thesis is divided into the following four sections, the first two relating to the secondary objective of the study, and the third and fourth sections relating to the primary objective: 1. experimental methods relating to the determination of the failure mechanism of hip fracture; 2. analytic methods relating to the determination of the failure mechanism of hip fracture; 3. methods used in the design of implants; and 4. methods relating the assessment of the implants. To meet the objectives of this experiment, two models of hip fracture were developed: an experimental model and an analytic model. These models are fully described in the Hip Fracture Mechanism sections. The other two sections describe how the models were used to meet the objectives of the section. They refer back to the first two sections for full descriptions of the models. 2.1 Hip Fracture Mechanism: Experimental Analysis 2.1.1 Specimens Four fresh-frozen, human proximal femur specimens were obtained from the Faculty of Medicine, University of British Columbia and LifeLegacy Foundation (Tuscon, A Z , USA). The femurs were stored in a freezer at -21° C. Anteroposterior radiographs were obtained and examined by an Orthopaedic Surgeon (Dr. Pierre Guy) for evidence of previous fracture, local area of lucency or pathology (e.g. metastases) which could alter mechanical testing results. D X A scans were performed with a Hologic QDR 4500W bone densitometer (Hologic Inc.,Waltham, M A ) using the standard protocol for the proximal femur. Donor information along with total proximal femur areal bone mineral density (aBMD) is presented in Table 2-1. 27 Methods Table 2-1 Donor information for specimens used in fracture mechanism experiments Donor Side Sex Age Weight (kg) Height (cm) a B M D (g/cm 2) 1004 L M - - - 0.867 1061 L F 72 63 157 0.844 1169 R M 72 90.7 170.18 0.764 1171 L F 91 40.8 157.48 0.722 2.1.2 Experimental Apparatus A n apparatus was built to reproduce the loads on the femur during a fall (Figure 2-1). This device was modelled after the device used by Lochmuller et al. [103]. The apparatus allowed the application of a load to the greater trochanter by a materials testing system while supporting the head of the femur and the distal shaft. The support at the head allowed full translation and rotation of the head, restricting only its vertical movement. The potting that held the distal end of the femur allowed translation in the horizontal plane and rotation in the coronal plane. The bottom of the apparatus consisted of two 11.5-mm-thick hardened steel plates, ground flat to within 50 pm. The bottom plate was 43 cm x 38 cm and sat on the base of the materials testing system (Instron 8874, Instron Corporation, Canton, M A ) . The second plate was 51 cm x 31 cm and was separated from the bottom plate by a layer of 0.95 cm diameter ball bearing set in steel plates. Two angled pieces of aluminum were bolted to the top plate. These angled plates supported a 19.05 mm diameter steel pin on circular Rulon bearing surfaces. The height of the pin was adjustable to accommodate different sizes of femur. The pin supported an aluminum tube that was 7.3 cm in diameter and 20.2 cm in length. This aluminum tube held the distal end of the femur. The head of the femur was supported by a steel plate that sat on a layer of 0.95 cm diameter ball bearings. 28 Methods Figure 2-1 Photograph and schematic diagram of the experimental apparatus. The yellow arrows on the photograph show the degrees of freedom at the proximal and distal ends of the femur. 2.1.3 Specimen Preparation Prior to testing, each femur was thawed to room temperature overnight. The femur was cut at the midpoint between the top of the greater trochanter and the bottom of the lateral condyle and the soft tissue was dissected away. Care was taken to remove as much of the periosteum in the intertrochateric and femoral neck region as possible without damaging the underlying cortical bone. To pot the femur, it was positioned in the aluminum tube so that the distance from the top of the greater trochanter to the support pin was 2/3 of the length of the femur. The shaft axis of the femur was oriented parallel to the tube. The neck axis of the femur was internally rotated 15° [103]. The pot was then filled with P M M A covering the femur between the mid-length cut and the proximal l / 3 r d . The potted femur was then placed in the apparatus with the head supported by a tennis ball shell and a 0.7-mm-thick piece of foam to distribute the load on the femoral head and prevent local crushing. The height of the distal pin was adjusted so that the shaft of the femur was at an angle of 10° with respect to the horizontal [103]. A P M M A pad was moulded to the greater trochanter and the loading plate of the materials testing machine to prevent local crushing by distributing the applied load. 2.1.4 Strain Gauges Femoral neck strain was measured on the surface of two of the femurs in this test (Specimen numbers 1004L and 1061L) using two uniaxial strain gauges (KFG-3-350-C1-11L1M2R, Omega Engineering Inc., Stamford, CT) . The surface was prepared and the gauges were applied following the application protocol of Carter et al. [107]. One gauge was applied to the superior 29 Methods surface of the femoral neck and one gauge was applied to the inferior neck (Figure 2-2). The gauges were connected to a signal conditioner (SCXI-1000, National Instruments, Austin, T X ) and sampled along with the axial load from the load cell at a frequency of 500 Hz, which was the limit of the software. Figure 2-2 Strain gauge location on superior neck (left) and inferior neck (right). White beads were bonded to femur to investigate the possibility of using video analysis to track failure in future experiments. 2.1.5 Failure Test Protocol A l l femurs were loaded to failure with a set displacement rate of -100 mm/s applied at the greater trochanter using a materials testing system (Instron 8874, Instron Corporation, Canton, M A ) . The tests were controlled using a Fastrak Console (Instron Corporation, Canton, M A ) . Load and displacement data were acquired using a 12-bit data acquisition card, sampled at 1000 Hz. A x i a l force was measured using a biaxial load cell (Model 211-113, SensorData Technologies Inc., Sterling Heights, M I ; serial number 97533). Displacement was measured using an Instron Linear Variable Differential Transformer (± 50 mm; serial number 0291). 2.1.6 Videography Two high-speed video cameras were used to film the tests to track the failure of the femur (Phantom v9.0, Vis ion Research, Wayne, N J , U S A ) . Images were captured at a resolution of 384 x 384 pixels, a sample rate of 9,111 frames per second and an exposure time of 99 ps. The cameras were controlled using Phantom Camera Control V 8.4.630 (Vision Research, Wayne, NJ). The femurs were illuminated using two 250 W photographic analysis lights (North Star, Wayne, NJ) . 30 Methods Two femurs (1169R and 1171L) were initially tested with both cameras filming the anterior surface of the femur. After these initial tests, the protocol was refined and the last two femurs (1004L and 1061L) were filmed from both sides (Figure 2-3), using mirrors to see the top and the bottom of the femoral neck (Figure 2-4). These last two femurs had the strain gauges attached and had small white beads bonded to them. The beads were added to track the deformation of the femur. The conditioned load signal from the load cell was also acquired by the Phantom Camera Control software at a rate of 9,111 Hz so that the applied load at each frame was recorded. Figure 2-3 Test set-up for the fracture mechanism experiments. The load was applied to the femur in the test apparatus by the Instron. Two cameras filmed the test, one from the anterior side and one from the posterior side. Two 250W lights were used to illuminate the specimen. Figure 2-4 Camera still on the left shows the anterior surface of the femur and a mirror image of the inferior surface of the femur. Camera still on the right shows the posterior surface of the femur and a mirror image of the superior surface of the femur. 31 Methods 2.2 Hip Fracture Mechanism: Finite Element Analysis To aid in the design and analysis of the femoral augmentation device, a finite element model (FEM) of a proximal femur was developed. To develop this model, a Quantitative Computed Tomography (QCT) scan was taken of a femur and the geometry and density were analyzed and exported to the FEM software, where a mesh was generated. Finally, the model was validated using mechanical experiments. 2.2.1 Specimen The specimen used for this model was a previously frozen human proximal femur that was obtained from the Faculty of Medicine, University of British Columbia. The donor for this specimen was a 73-year-old woman. An anterioposterior radiograph was obtained using standard femur protocol. A radiologist (Dr. Bruce Forster) examined the film and found no indication of previous fracture or metastatic bone disease. A DXA was performed using a Hologic QDR 4500W bone densitometer (Hologic Inc.,Waltham, MA) to measure the bone mineral density of the specimen. The scan was performed using the standard protocol for the proximal femur. The T-Score of the proximal femur was found to be -2.2 in the total proximal femur region. 2.2.2 QCT Imaging To prepare the specimen, the femur was cut at mid-shaft and the muscles, tendons and ligaments were dissected away. The proximal end of the femur was submerged in ultrasound gel to provide a medium surrounding the bone. The bone was held rigid, with the shaft axis perpendicular to the scan plane and the femoral neck axis parallel to the horizontal. The bone was placed on top of Model 3 CT Calibration Phantom (Mindways Software Inc., San Francisco, CA, USA) so that each image of the bone would also contain the phantom. The bone was scanned using a GE LightSpeed Ultra 16 slice helical scanner (General Electric Healthcare Technologies, Waukesha, WI, USA). The scan was performed by a trained technologist at settings of 120.0 kV, 100 mA, table height of 129.5 cm, slice thickness of 1.25 mm, matrix size of 512 x 512 pixels and pixel size of 0.49 mm x 0.49 mm. A total of 168 consecutive images were obtained. 32 Methods 2.2.3 Bone Geometry The CT images were segmented using Analyze 5.0 (AnalyzeDirect Inc, Lenexa, KS , USA) to determine the geometry of the femur. The segmenting procedure involved tracing two boundaries on each slice, the outer surface of the bone and the boundary that separates the cortical or subchondral from the cancellous bone or intermedullary canal. This segmentation was done using the thresholding function of the program, with manual override to ensure continuity of the bone. The cortical bone was separated from the cancellous bone, so that during meshing the cortical elements would contain only cortical bone. When the geometry had been defined for the entire proximal femur, it was exported into the finite element analysis software (ANSYS 8.0, ANSYS Inc., Canonsburg, PA, USA). 2.2.4 Bone Density Distribution Analyze 5.0 was also used to determine the density distribution of the bone. A density map of the proximal femur was exported using the program. The density map contained the location of each voxel and the corresponding attenuation expressed in Hounsfield Units (HU). The phantom was used to relate the H U values to a K2HP04 equivalent density for each voxel. The equivalent K2HP04 density has been shown to relate to the material properties of bone [108]. This equivalent density was later used to determine the distribution of material properties and to apply these material properties to the cancellous elements in the completed mesh. Calibration Phantom The phantom consisted of five rods of reference materials that contained known amounts of low and high atomic number materials. The rods were previously calibrated against liquid K2HPOA /water solutions and replicate solutions of precisely known water and K2HPOA densities. The equivalent water density of the rods varied from 923.2 mg/ cm 3 to 1119.5 mg/cm3. The K2HPOA equivalent density varied from -53.4 mg/ cm 3 to 375.8 mg/ cm 3. Calibration The scan was calibrated following the calibration procedure outlined in the phantom manual. Briefly, this involved tracing a region of interest around each of the columns in the phantom in the Analyze 5.0 program. The average attenuation in H U was then found for each column. A 33 Methods linear regression was then performed using these values and the known equivalent water and K2HPOi equivalent densities of each column. This analysis yielded two imaging-technique specific parameters that were used to relate the attenuation of the bone to a K2HP04 equivalent density with the following linear function: N _ ^ROI ~ P e r o~CT Where, MROI ~ CT number within a region of interest of the bone, in HU pKHP0) ~ K2HP04 equivalent density of the bone within the region of interest w o o 5 6000 4000 2000 0 -2000 -4000 -6000 -| -8000 Inferior 000 1500 2000 2500 3000 3500 4Q00 Superior Load (N) Figure 3-3 Strain measured in the inferior and superior femoral neck of 1061 L plotted against applied load 51 Results 3.1.3 Fracture Type The fractured specimens are shown below in Figure 3-4. The fracture in specimen 1004 L went through the distal femoral neck and is classified as basicervical. The fracture in this femur was jagged and passed close to the strain gauges mounted to the femur. The fracture ran beneath the distal end of the superior strain gauge and passed approximately 1 cm distal to the inferior gauge. The fracture in specimen 1061 L passed superiorly through the distal femoral neck and inferiorly through the lesser trochanter. This type of fracture is classified as intertrochanteric. The fracture was located approximately 1 cm distal to the superior gauge and was not close to the inferior gauge. The fracture in specimen 1169 R ran obliquely though the femoral neck. This fracture was classified as a femoral neck fracture. The fracture in 1171 L also ran obliquely through the femoral neck and was also classified as a femoral neck fracture. There were no strain gauges on the specimens 1169 R and 1171 L. 1004 L 1061 L 1169 R 1171 L Basicervical Intertrocantheric Femoral Neck Femoral Neck Fracture Fracture Fracture Fracture Figure 3-4 Fracture types in specimens used in fracture mechanics tests 3.1.4 Failure Location and Fracture Progression In this section, the progression of the fractures through the femurs will be qualitatively discussed for each femur. Specimen 1004 L The load displacement for this femur had three distinct peaks (Figure 3-1). At the first peak of 4390 N , the head and neck was seen to rotate slightly towards the shaft of the femur (Figure 3-5). At this point in the test, the inferior surface of the femur was loaded in tension, and the superior 52 Results surface was loaded in compression, as shown in the strain gauge readings (Figure 3-2). This suggests that this first peak was most likely the result of a failure in the head of the femur or in the support structure. Inspection of the fractured specimen did not reveal any obvious crushing in the femoral head, so this failure was likely the result of failure in the support foam or a shifting in the apparatus beneath the femoral head. At the second peak of 4607 N , a visible fracture appeared in the superior cortex of the proximal neck (Figure 3-5). Figure 3-5 Images corresponding to the first (left) and second (right) peaks of the load-displacement curve of 1004 L. The rotation shows the movement of the head of the femur seen at the first peak. The failure indicates a visible yielding at the second peak. The third peak in the test corresponded to the tensile failure seen in the inferior femoral neck (Figure 3-6). This peak occurred at a load of 3582 N . There was a time delay of 9.0 ms between the initiation of the compressive crack seen at the second peak and the initiation of the tensile crack seen at the third peak. This crack in the inferior neck took 1.5 ms to propagate through the neck and reach the crack in the superior neck. 53 Results Specimen 1061 L The load-displacement curve for this specimen had two small drops in load as the load increased to the ultimate load. It also had a large drop that occurred shortly after the ultimate load had been reached (Figure 3-1). Figure 3-7 Images corresponding to the start of the loading sequence (left) and the loaded femur after passing two small peaks (right). The numbers 1 and 2 refer to the locations of failure at the first and second peaks respectively. 5 4 Results The first peak load of 1714 N corresponded with a release of fluid from the posterior surface of the femur at the junction of the femoral neck and head (Figure 3-7). The second peak load of 3040 N corresponded with a release of fluid at the junction of the neck and head of the femur on both the superior and inferior surfaces (Figure 3-7). Figure 3-8 Images corresponding to failure seen at the ultimate load (left) and at the final drop in load (right) The third peak load of 3361 N corresponded to a compressive failure seen in the distal superior femoral neck (Figure 3-8). This was followed by a tensile failure in the medial intertrochanteric region (Figure 3-8). The tensile failure occurred 28.9 ms after the compressive failure occurs. It was not possible to see how long the fracture took to progress across the intertrochanteric region. Specimen 1169 R The load-displacement curve for this specimen was linear until it had almost reached the ultimate load. There was a short horizontal region on the curve before the specimen reached the ultimate load of 3038 N (Figure 3-1). The load data was not synchronized with the video, so it was not possible to see exact loads at specific images. However, it was possible to estimate the loads for various images using the time stamp on the video. The fracture progressed through two distinct phases that were clearly evident from the video: an initial compressive failure followed by a tensile failure. 55 Results Figure 3-9 Images showing femur prior to fracture (left) and immediately following the initiation of fracture (right) in specimen 1169 R A crack first appeared on the anterior-superior surface of the cortex at the junction of the femoral neck and the head of the femur (Figure 3-9). This appeared to be a compressive failure and occurred at the time that the load displacement curve leveled off. Figure 3-10 Image showing final failure of specimen 1169 R After the femur fractured in the anterior surface, it fractured in the inferior neck. As the inferior neck fractured, the head pulled away (Figure 3-10). The time elapsed between the first fracture 56 Results and the second fracture was 4.6 ms. The time elapsed between the plateau in the load displacement curve and the ultimate load was approximately 5 ms. Specimen 1171 L The load-displacement curve for this specimen was linear until it reached the ultimate load of 3841 N (Figure 3-1). The load dropped to 3095 N and then increased to 3841 N and subsequently dropped off. The load data for this test was also not synchronized with the video. This specimen also failed initially in compression, followed by a tensile failure. Figure 3-11 Images showing femur prior to fracture (left) and immediately following the initiation of fracture (right) in specimen 1171 L A crack first appeared in the superior cortex of the proximal femoral neck (Figure 3-11). This crack appeared to be a compressive failure. The time that the crack appeared on the video corresponded with the first peak in the load displacement curve. 57 Results Figure 3-12 Image showing final failure of specimen 1171 L The view of the initiation of the crack in the inferior neck was partially obscured by padding under the head of the femur (Figure 3-12). Based on what was visible and the motion of the femur, the fracture corresponded to the final drop in load seen in the load-displacement curve. The inferior neck fracture began 19.3 ms after the fracture in the superior neck began. 3.2 Finite Element Model Results In this section, the results of the validation test will be presented. These will be followed by an analysis of the failure of the femur predicted by the model. Finally, the predicted strengthening effects of some implant designs will be shown. 3.2.1 Cadaver Model Strain Measurements In this section the strain measured in the femur 1090R during non-destructive testing, which was used to develop the finite element, will be presented. This test was performed to validate the finite element model, but these results are also interesting because they reveal further information about the strain distribution in the femur during a fall. 58 Results Load(N) Figure 3-13 Strain measurements at three locations in a plane through the proximal neck when the femur was loaded to 435 N The strains measured during one load cycle of the cadaver specimen in the simulated fall apparatus are plotted against the applied load in Figure 3-13, Figure 3-14 and Figure 3-15. The max applied load in this cycle was 435 N , which is well within the elastic region of the bone. The strains in the proximal femoral neck plane varied linearly with applied load (all R 2 > 0.99). The largest strain was in the superior neck (PI) and reached -607 ps. The posterior neck gauge (P3) was also in compression and reached a strain of -217 ps. The anterior gauge was in tension and measured a strain of 182 ps. 59 Results Anterior Posterior 200 300 400 5*0 D1 D 2 ] Section D Superior • Inferior L o a d 0.99). The highest strain was seen in the anterior neck (D2), with the gauge reading 568 ps at the peak applied load. The superior (Dl) and posterior (D3) gauges both measured compressive strain with maximum load readings of -536 pe and -547 ps respectively. 60 Results Lateral • Anterior Posterior Section S Loacl(N) Figure 3-15 Strain measurements at three locations in a plane through the distal neck when the femur is loaded to 435 N In the shaft (Figure 3-15), the measured strain also varied linearly with applied load (R 2 > 0.99 for S2 and S3, R 2 = 0.97 for SI). The anterior gauge (S2) measured the highest strain, with a reading of 434 pe at maximum load. The lateral and posterior gauges measured compressive strain and had maximum load readings of -334 pe and -107 pe respectively. 3.2.2 Strain Predicted by Finite Element Model The strains predicted by the finite element model at the nine gauge locations when loaded to 435 N are presented along with the measured strain in Figure 3-16. The mean absolute difference between measured and predicted strain was 138 ps. The strains predicted in the proximal neck were the least accurate. The predicted strain underestimated the measured strain at PI by 61% and at P3 by 60%. The predicted strain at P2 overestimated the measured strain by 90%. The predicted strains at the distal neck were more accurate. The finite element model underestimated the strain at D l by 13%. The model overestimated the strain at D2 and D3 by 16% and 37% respectively. With the exception of location S3, where the model predicted a negligible amount of strain, the predicted strain was similar to the measured strain in the shaft. The gauges at SI and S2 underestimated the measured strain by 4% and 23% respectively. The model predicted -1 pe at location S3 and the gauge measured 107 pe at this location. 61 Results Figure 3-16 Comparison of strains measured experimentally and strains predicted by finite element model Predicted strain was plotted against experimentally measured strain in a manner similar to Anderson et al. [112]. The predicted strain was found to correlate strongly with the measured strain (r2 = 0.87), indicating that the model behaviour is a reasonable approximation of the actual behaviour of the femur. -800 c Z -800 to < u. 800 Experimental Strain (pe) Figure 3-17 Plot of predicted strain vs. measured strain at 9 locations 62 Results 3.2.3 Failure of Intact Finite Element Model of Femur The intact model of the proximal femur was predicted to have a failure load of 3618 N . At this load the von Mises maximum strain of 0.011 was exceeded in the superior cortex of the distal femoral neck. The von Mises strain distribution of the posterior surface of the proximal femur when loaded to 3618 N is shown in Figure 3-18. At this load several elements in the superior cortex experienced high compressive strains that exceeded or were close to exceeding the von Mises failure criterion. A view of the anterior-inferior surface of the proximal femur is shown in Figure 3-19. In this figure it can be seen that the femur was quite far from failing in the inferior surface of the femur which was primarily loaded in tension. The predicted failure load was 11% higher than the 3285 N failure load of the contralateral femur. Figure 3-18 Posterior view of von Mises strain distribution in proximal femur when loaded to 3618 N in a fall configuration 63 Results Figure 3-19 Anterior and slightly inferior view of von Mises strain distribution in proximal femur when loaded to 3618 N in a fall configuration 3.3 Finite Element Analysis of Augmented Femurs The predicted failure loads of the eight augmented femur designs are shown below in Figure 3-20. Implant (a), which was a composite rod through the inferior neck, was predicted to have the least effect on the strength of the femur. The femur was predicted to fail at 4071 N , representing an increase of 13% over the intact femur model. The failure was predicted to occur in the superior distal neck in a manner similar to the intact femur. Implant (b), which was a composite rod through the superior distal neck, had a greater strengthening effect. This model was predicted to fail in the superior distal neck at a load of 5584 N , representing an increase of 54% over the intact femur model. 64 Results Implant (c), which combined the rods of implants (a) and (b), had an effect similar to implant (b). It predicted failure at a load of 5765 N in the superior distal neck, representing an increase of 59% over the intact femur model. Implant (d), which was a rod through the centre of the femoral neck, was not predicted to strengthen the femur by very much. The model predicted failure at a load of 4317 N , representing an increase of 19% over the intact femur model. The predicted location of failure was the superior distal neck with high strains also seen on the posterior surface of the base of the femoral neck. Implant (e), which was a larger rod through the centre of the femoral neck, had a greater strengthening effect than implant (d). The model predicted failure at 5164 N , representing an increase of 43% over the intact femur model. The fracture was predicted to begin in the superior distal neck. Implant (f), which was the neck-contouring composite, was predicted to fail at a load of 5952 N , representing an increase of 65% over the intact femur model. The predicted location of failure was the superior distal neck. Implant (g), which is the same as implant (b) with a wire in the inferior neck to hold tension, had a lower failure load than implant (b). It was predicted to fail at 4693 N , representing an increase of 30%o over the intact femur model. The fracture was predicted to initiate in the superior distal neck. Implant (h), which was the Gamma Nail, was predicted to have the highest failure load. It was predicted to fail at a load of 6659 N , representing an increase of 84% over the intact femur model. The femur was predicted to fail in the superior distal neck but high strains were also seen in the inferior neck. The elements adjacent to the implant were ignored in the failure analysis. This is because they were assumed to be rigidly attached to the implant. This assumption caused very high strains in the elements in the greater trochanter and femoral neck that contacted the 65 Results implant. If these elements are not ignored, the predicted failure load of the proximal femur i much less. 8000 7000 6000 T3 (0 5000 o _l